Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
00001.tif
(USC Thesis Other)
00001.tif
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Information Company 300 North Zed) Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/321-0600 ANALYSIS OF HIGH RESOLUTION PHYSICAIVBIO-OPTICAL MOORED TIME SERIES AND COMPARISON TO A 1-D INTERDISCIPLINARY OCEAN MODEL by Jerem y David Wiggert A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (E arth Sciences) August, 1995 Copyright 1995 Jerem y David Wiggert UM I Number: 9621646 UMI Microform 9621646 Copyright 1996, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeb Road Ann Arbor, MI 48103 UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90007 This dissertation, written by Jeremy David Wiggert under the direction of k .is Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY Dean of Graduate Studies Date DISSERTATION CO M M TTTEE O uirptm n A cknow ledgem ents I would like to express to Tommy Dickey my deep appreciation for providing the Biowatt data set, allowing me to choose my own analysis path, providing the computer facilities and acting as my principal advisor. I would also like to thank Burt Jones, Dale Kiefer, Jim Kremer, Tony Maxworthy and Libe W ashburn for their guidance while serving as members of the thesis committee. I would like to emphasize my gratitude to Jim Kremer for standing in for Dale Kiefer and for his thorough, and much appreciated, editorial comments. I would also like to express my appreciation to Tony Maxworthy for getting me started a t USC. Also, many thanks are due to past and present members of the USC/OPG for some fun times and continuous moral support. This especially includes Tim Granata and Mike Hamilton (2/3 of the Biowatt post-processing trium virate and backpacking cohorts), Isabelle Taupier-Letage (mastering the finer points of GIDO), M argaret Stram ska (co-commiserator on the state of the micro-VAX) and Anne Petrenko (for numerous reasons but especially for sharing one of my darker periods). I would also like to thank the various members of the Bedrocker intram ural teams (especially Reese, Kevin, Erik, Jim , Tom, Whitey and Josh) for some thrilling efforts. Finally, I would like to thank Rene (for always solving my USC protocol questions), Semele (for numerous word processing breaks), Sam (the lunch bunch stalw art and volleyball babe) and Charlotte (for letting me hook my vcr up to her tv). And, of course, I would like to thank my family (Beth, Vic, Joy and Aggie) for their patience and support. Acknowledgements........................................................................................... i Table of C ontents..............................................................................................iii List of Figures.................................................................................................... viii List of Tables....................................................................................................... xiii List of Abbreviations and Symbols...................................................................xiv A bstract............................................................................................................ xix C hapter 1 Introduction..............................................................................................1 Chapter 2 The Biowatt Experim ent........................................................................ 7 Physical background for the site..................................................7 Biological characteristics and interaction with the physical field.......................................................................................... 8 Seasonal cycles observed in previous biological and geochemical studies in the Sargasso Sea.............................. 10 Description of the present experiment.......................................... 12 Usage of the bio-optical instrum ents.......................................... 15 Presentation and discussion of the Biowatt moored interdisciplinary time series....................................................18 Synopsis of in situ observations: Deployment 1............22 Synopsis of in situ observations: Deployment 2 ............29 Synopsis of in situ observations: Deployment 3 ............32 Chapter 3 Primary productivity based on moored time series.................................33 Introduction to phytoplankton photosynthesis..........................33 Models of gross production............................................................ 34 Cc as it relates to net production and particle specific accumulation rates................................................................. 36 M ethods.........................................................................................38 Estimating gross growth rates, GPP and C.Chl ratios.............................................................................38 The Kiefer model..................................................... 38 Error minimization.................................................. 40 Estim ating C:Chl ratio........................................... 44 Estimation of deployment averaged net production, growth and grazing rates............................................. 45 Results........................................................................................... 48 Description of GPP estim ates and comparison to previous resu lts............................................................48 Estimated Net Production................................................. 52 Estimated productivity, respiration and grazing rate s..............................................................................58 Discussion..................................................................................... 60 C hapter 4 Spectral Analysis of the Moored Time Series: The Search for Linear Interaction Mechanisms............................................................... 66 Introduction to Physical/Biological Interactions.......................... 66 Analytical methods and presentation technique........................ 74 Deployment averaged autospectra............................................. 89 Horizontal Currents............................................................89 Tem perature..................................................................... 89 Dissolved Oxygen (DO)...................................................... 90 Bio-optical spectra (cc and C hi)....................................... 91 Chronology of Observed Physical/Biological Interactions........... 92 Deployment 1.....................................................................93 Period 1: (JD 60-78)...............................................98 Period 2: (JD 78-90)............................................... 107 Period 3: (JD90-130)............................................. 116 Deployment 2 ..................................................................... 119 Period 4: (JD 135-190).............................................119 Period 5: (JD 190-220).............................................125 Period 6: (JD 220-240).......... 126 Discussion......................................................................................128 Chapter 5 A one dimensional interdisciplinary model............................................. 137 Introduction...................................................................................137 Objectives behind the development of a 1-D interdisciplinary model............................................................137 Previous interdisciplinary modeling studies......................140 Development of the interdisdplinaiy model..................................143 The physical model..............................................................144 Components of the ecosystem model..................................146 Coupling between the physical and ecological system s.........................................................................151 Application of the interdisciplinary model................................... 154 Applied boundary conditions and model initialization 154 Synopsis of the site’ s physical and biological regim es..........................................................................156 Observations of the coupled system based on the locally derived heat budget...........................................158 Comparison of simulated physical fields with in-situ m easurem ents..............................................................159 Comparison of simulated ecosystem fields with in- situ m easurem ents...................................................... 164 Discussion..................................................................................... 179 Conclusion............................................................................................................186 Appendix 4.1 Determination of significant coherence..................................................188 Appendix 4.2: Presentation of spectral analysis results..................................190 Spatial/Temporal distribution of normalized spectral density..................................................................................... 190 Horizontal Current Speed...................................................190 Tem perature......................................................................192 Beam attenuation coefficient (cc)........................................195 Chlorophyll-Fluorescence (Chi).........................................196 Spatial/Temporal distribution of coherence................................ 198 Coherence between oceanic currents and tem perature................................................................. 198 Coherence between oceanic currents and dissolved oxygen........................................................................... 204 Coherence between oceanic currents and bio-optical variables........................................................................ 204 Coherence between tem perature and DO........................ 208 Coherence between tem perature and cc..........................211 Coherence between temperature and Chi......................... 213 Coherence between C c and Chi......................................... 213 Appendix 5.1 Equation set for the level 2.5 MeUor-Yamada mixed layer model..........217 Appendix 5.2 Equation set used for the pelagic ecosystem model.............................. 220 Appendix 5.3 Ecosystem model param eter definitions................................................223 Appendix 5.4 Ecosystem model constant definitions...................................................224 Appendix 5.5 Nitrate-Temperature relation derived from in-situ profiles................... 225 Appendix 5.6 Determination of the local heat budget.................................................... 227 Bibliography....................................................................................................... 236 vii List ftf Figures Figure 1.1 Illustration of the temporal and spatial scales relevant to the physical and biological processes sampled during the Biowatt experiment..............................2 Figure 1.2 Experiment site in the Sargasso Sea................................... 2 Figure 1.3 Configuration of MVMS system used for the Biowatt experim ent........................................................................... 3 Figure 2.1 Surface heat and momentum fluxes and optical characteristics of the water column...................................... 14 Figure 2.2 Time series of averaged current stick vectors.....................19 Figure 2.3 Superimposed in-situ temperature time series.................... 19 Figure 2.4 Individual time series of PAR................................................ 20 Figure 2.5 Individual time series of beam attenuation coefficient........20 Figure 2.6 Individual time series of chlorophyll..................................... 21 Figure 2.7 Individual time series of dissolved oxygen............................ 21 Figure 2.8-10 SST maps providing a synoptic view of the Sargasso Sea in the vicinity of the mooring..........................................23 Figure 2.11 Contour map of tem perature................................................ 27 Figure 2.12 Contour map of diffuse attenuation coefficient.................... 27 Figure 2.13 Contour map of beam attenuation coefficient.................... 30 Figure 2.14 Contour map of chlorophyll...................................................30 Figure 2.15 Contour map of stratification...............................................31 Figure 2.16 Contour maps of shear.......................................................... 31 Figure 3.1 Comparison between gross production estim ates obtained from ship based incubations and estimates made by optical production models....................................... 42 Figure 3.2 Time series of daily integrated gross prim ary productivity...........................................................................49 Figure 3.3 Contour map of 6* from 50 to 100 m eters...........................49 Figure 3.4 Contour maps of gross prim ary production..........................51 Figure 3.5 Time series of areal production based on the Kiefer model results.........................................................................53 Figure 3.6 Deployment averaged diel cycles of A ce................................ 54 Figure 3.7 Comparison between gross production and net production profiles........................................... 56 Figure 3.8 Contour map of net production...............................................5 7 Figure 3.9 Comparison of the quantum yield function used in the LDGO model and daily integrated values predicted by the Kiefer model.................................................................... 63 Figure 4.1 Deployment averaged rotary spectra for currents a t 23 and 101 m eters................................................................79 Figure 4.2 Deployment averaged spectra for tem perature at 23 and 101 m eters.....................................................................80 Figure 4.3 Deployment averaged spectra for beam attenuation coefficient at 23 and 101 meters........................................... 81 Figure 4.4 Deployment averaged spectra for chlorophyll at 23 and 101 m eters.....................................................................82 Figure 4.5 Deployment averaged spectra for dissolved oxygen a t 23 and 101 m eters................................................................83 Figure 4.6 GW/.t) for current speed........................................................ 94 Figure 4.7 GjsK/.t) for tem perature with associated Brunt- Vfiisfilfi frequency time series................................................96 Figure 4.8 GuCf.t) for beam attenuation coefficient..............................99 Figure 4.9 GnCM) for chlorophyll............................................................. 102 Figure 4.10 Coherence between tem perature individual current directions................................................................................104 Figure 4.11 Coherence between tem perature and dissolved oxygen...................................... 108 Figure 4.12 Coherence between tem perature and chlorophyll................110 Figure 4.13 Coherence between beam attenuation coefficient and chlorophyll.............................................................................. 112 Figure 4.14 Coherence between dissolved oxygen and individual current directions.................................................................. 117 Figure 4.15 Coherence between tem perature and beam attenuation coefficient.......................................................... 121 Figure 5.1 Intercompartm ental flows for the ecosystem model........... 139 Figure 5.2 Coupling mechanisms within the interdisciplinary model......................................................................................152 Figure 5.3 Comparison of MLD based on the in-situ measurements and the model results..................................160 Figure 5.4 Comparison of measured and modeled SST...........................160 Figure 5.5 Contour map of seasonal tem perature cycle predicted by the physical model............................................................162 Figure 5.6 Contour map of seasonal stratification cycle predicted by the physical model............................................................162 Figure 5.7 Contour map of seasonal cycle in current shear predicted by the physical model........................................... 163 Figure 5.8 Contour maps of chlorophyll concentration...........................165 Figure 5.9 Comparison of areal production calculated using the Kiefer model with the in-situ data and as predicted by the interdisciplinary model....................................................167 Figure 5.10 Contour map of gross primary production as predicted by the interdisciplinary model............................................... 167 Figure 5.11 Contour map of zooplankton concentration as predicted by the interdisciplinary model.............................. 168 Figure 5.12 Contour maps of modeled and measured PAR.................... 169 Figure 5.13 Comparison of 1% light depth between estim ates based on in-situ PAR measurements and the irradiance field as predicted by the interdisciplinary model......................................................................................171 Figure 5.14 Contour maps of modeled nitrate concentration, empirically derived nitrate concentration, modeled ammonium concentration and f-ratio.................................. 172 Figure 5.15 Distribution of DON predicted by the interdisciplinary model......................................................................................175 Figure 5.16 Model and in-situ distribution of bacteria............................. 176 Figure 5.17 Model and in-situ distributions of PON and time series of particle flux at 200m........................................................ 177 Figure 5.18 Comparison of Qsw time series obtained from the surface buoy and maximum theoretical value assuming no cloud cover...................................................... 183 Figure A4.1.1 Significant coherence threshold..............................................189 Figure A4.2.1 Gn( /^ ) for the current field....................................................191 Figure A4.2.2 Gn(/,Z) for the temperature field...........................................193 Figure A4.2.3 Coherence individual current vectors and tem perature.......................................................................... 199 Figure A4.2.4 Coherence individual current vectors and tem perature during the third deployment.................................................202 Figure A4.2.5 Coherence between the two current directions and the bio-optical variables for the first deployment...................... 205 Figure A4.2.6 Coherence between the two current directions and the bio-optical variables for the second deployment................. 206 Figure A4.2.7 Coherence between the two current directions and the bio-optical variables for the third deployment..................... 207 Figure A4.2.8 Coherence between tem perature and dissolved oxygen for the second deployment....................................................209 Figure A4.2.9 Vertical coherence between tem perature and beam attenuation coefficient......................................................... 212 Figure A4.2.10 Coherence between tem perature and chlorophyll...............214 Figure A5.5.1 Empirical relation between tem perature and nitrate..........226 Figure A5.6.1 Sea surface heat fluxes measured during Biowatt.............. 230 Figure A5.6.2 Time series of terms in the heat flux budget.........................232 List of Tables Table 3.1 Chlorophyll specific absorption............................................. 39 Table 3.2 Prim ary productivity growth and loss rates determined using the Kiefer model and the transmissometer data............................................................59 Table 4.1 Comparison of Biowatt productivity to estim ates from a warm-core ring.......................................................... 71 Table 4.2 D ata available for spectral analysis.....................................76 Table 4.3 Brunt-VSisfilS frequency estimated from the in-situ tem perature and current time series.................................. 78 Table 4.4 Transformation between frequency in standard units and the frequency axis used in the contour maps of Gn and y...................................................................................... 86 L ist o f A bbreviations and Sym bols achi: Chlorophyll-specific Absorption Coefficient [m2 /g Chi a] AVHRR: Advanced Very High Resolution Radiometer BOMS: Bio-optical moored system BOPS: Bio-optical profiling system be: Empirical constant from Eq. 2.7 [4.56] c*c: Carbon-specific C c [3.92 • 10*3 m2 / gC] C: Community biomass in carbon units [mgC/m3 ] cc: Beam attenuation coefficient at 660 nm [1/m] CCW: Counterclockwise Chi: Chlorophyll-Fluorescence from moored data [mg Chi a /m3 ] Cl: Confidence interval C p: Specific heat of water [J/kg^C] cph: Cycles per hour [1/h] ce: C c reduced by 0* [1/m] CTD: Conductivity tem perature depth profiler CW: Clockwise CZCS: Coastal Zone Color Scanner D: Photoperiod [fraction of 24 hours] DCM: Deep chlorophyll maximum D 02: O 2 m e a s'0 2 aat [jtM ] DO: Dissolved oxygen [jiM] DOM: Dissolved organic m atter DON: Dissolved oiganic nitrogen EE: Expected Error f-ratio* PPW3----- i ratio. ppnh4 + PPnos FFT: Fast Fourier Transform g: Gross carbon specific fixation rate [1/day] < g >: Light period averaged gross carbon fixation rate [1/day] GCM: General circulation model Gn(/, t): Temporally varying normalized spectral density function Gsif, Z): Spatially varying normalized spectral density function GPP: Gross Prim ary Productivity [mgC/m3 /day] gaat'. Light-saturated, carbon-specific cell growth rate [1/day] Gxx(f, t): Temporally varying auto-spectral density function Gxx(f, Z): Spatially varying auto-spectral density function Gxy(f, t): Temporally varying cross-spectral density function Gxy(f, Z): Spatially varying cross-spectral density function h: Heterotrophic grazing rate [1/day] HKE: Horizontal Kinetic Energy [m^s2 ] IGW: Internal Gravity Wave Iir: Infrared Radiation [W/m2 ] I o = PAR [Ein/m2 /d] JD: Julian day IQ: Chlorophyll Specific Attenuation Coefficient [m2 /mg Chi a] Kh: Thermal Difiusivity [m2 /s] Khi: Background IQ] due to Internal Waves [m2 /s] Khs: Background Kh due to Shear Instability [m2 /s] Km: Momentum Difiusivity [m2 /s] Kmi: Background Km due to Internal Waves [m2 /s] Kms: Background Km due to Shear Instability [m2 /s] xv K p a r : Diffuse attenuation coefficient [1/m] <w0>: Vertical turbulent heat flux <wu>, <wu>: Vertical turbulent momentum flux LOTUS: Long Term Upper Ocean Study MVMS: Multi-variable Moored System NH4 +: Ammonium N 03 ': N itrate NP: Net Production [mgC/m3 /day] Nph: N utrients [pM] NSV: Normalized Spectral Variance P: Instantaneous carbon-specific photosynthetic rate [1/hr] PAR: Photosynthetically Available Radiation P B : Chlorophyll-specific photosynthetic rate [g C/g Chl/day] Pc: Particle specific rate based on c « [1/day] <Pa>: Averaged particle specific accumulation rate [1/day] <Pi>: Averaged particle specific loss rate [1/day] POC: Particulate Organic Carbon [mgC/m3 ] PON: Particulate Organic Nitrogen [mMol N/m3 ] P P N H 4: Regenerated production (ammonium driven) P P nos: New production (nitrate driven) PZN: Phytoplankton-Zooplankton-Nutrient Ola'- Latent heat flux [W/m2 ] Q lw * Long wave heat flux [W/m2 ] Qse: Sensible heat flux [W/m2 ] Q « w * Short wave heat flux [W/m2 ] r Specific respiration rate [1/day] xvi SI: Sampling Interval [days] SPM: Suspended particulate m atter [mgC/m3 ] SSH: Sea surface height SST: Sea surface tem perature [f i C] T: Temperature [® C ] t: Time U: 3-D velocity vector Uh: Horizontal velocity vector [2-D] VMCM: Vector Measuring Current Meter W: Upwelling velocity XBT: Expendable Bathythermograph Z : Depth [m] Znjd: Mixed layer depth [m] Znf. Reference depth for vertical integration in heat budget [m] T : Average tem perature of vertical profile through Znf fs v - Brunt-Vfiisfila frequency [cph] / d : Diurnal frequency [cph] fi: Inertial frequency [cph] flGW'- Internal gravity wave band [i.e., f \ < = > / bv» cph] / m: Mesoscale frequency band [i.e., = > / d , cph] fsD' Semi-diurnal frequency [cph] a: Albedo A c« : Del c [1/m] $: Quantum Yield [mol C/ mol photons] Maximum Quantum Yield [0.084 mol C/mol photons] xvii YxyCf. t): Temporally varying ordinary coherence function YxyCf. Z): Spatially varying ordinary coherence function YxyCflGW. t): Temporally varying coherence within the 1GW band YxyCflGW. Z): Spatially varying coherence within the IGW band YxyCfM> t): Temporally varying coherence within the mesoscale band YxyCfM. Z): Spatially varying coherence within the mesoscale band p: Specific growth rate [1/day] 6*: Ratio of cell C to total C [Eq. 3.10a] 6: Carbon:Chl ratio [g C/ g Chi a] 6: Turbulent tem perature [%] 6 k: C:Chl Ratio from model [g C/ g Chi a] 0^: Minimum C:Chl Ratio [36 gC/g Chi a] 6u: C:Chl Ratio estimated from time series [gC/g Chi a] pw : Density of w ater [kg/m3 ] T i n : Mixing time scale [s] T p h: Minimum Steady-state Reaction Center Turnover Time £: Solar zenith angle [degrees] V: Gradient operator V h: Horizontal gradient operator A b stract A moored data set consisting of high resolution time series of physical and bio-optical variables from the western Sargasso Sea is presented. The duration of the experiment provided for observing characteristics of the region’ s annual hydrodynamic and pelagic cycles. The time series and ancillary synoptic maps reveal this region to be strongly affected by mesoscale activity associated with proximity to the Gulf Stream. High resolution estim ates of primary productivity were generated from the bio-optical time series. These reveal variations in phytoplankton biomass and prim ary production as having a strong episodic component superimposed on the well established diel cycle. Time series analysis reveals signatures in spectral variance associated with mesoscale activity a t low frequencies and internal wave motions a t higher frequencies in the bio-optical as well as the physical measurements. Coherence between variables has been used to isolate and emphasize physical/biological interactions such as distinct ecosystems contained within mesoscale features and a nutrient injection driven by internal wave pumping which manifests as temporarily elevated phytoplankton biomass. Finally, a 1- D interdisciplinary model has been developed by coupling the Mellor-Yamada mixed layer model with the Fasham ecosystem model. Using boundary condition time series obtained from the surface buoy associated with the mooring experiment, this model successfully emulates the timing of the spring bloom, shows a remarkable ability to recreate the variability observed in summertime areal production and predicts remineralization rates associated with the microbial loop and carbon flux associated with sinking particulate m atter. C hapter 1: Introduction In-situ studies of oceanic ecosystems require an interdisciplinary sampling approach since light intensity and nutrient concentration, the main requirements for photosynthesis, are fundamentally linked to the physical environment. Biological processes of interest include diel photosynthetic cydes, primary productivity and sinking fluxes of biogenic partides. In addition to studying basic biological mechanisms, modulation of these processes through interaction with the physical environment is a topic of ongoing investigation as well. Physical phenomena which are of particular interest with regard to such interactions indude tidal, inertial and internal wave oscillations, the seasonal cyde of water column stratification, episodic synoptic forcing and advective coherent mesoscale features. Clearly, advancing the understanding of these varied processes necessitates the ability to sample a broad suite of physical and environmental variables over an extensive range of time scales (Fig. 1.1). These observational goals were accomplished using an interdisciplinary approach as part of the ONR sponsored Biowatt project which took place in 1987 in the Sargasso Sea (3 4 ® N 70s W, Fig. 1.2). The data set utilized here was obtained with a set of eight instrum ent arrays, dubbed multi-variable moored systems (MVMS, Fig. 1.3). Each MVMS package consisted of a vector measuring current meter (VMCM) with attached instrum entation which provided colocated, four minute resolution measurements of tem perature, photosynthetically available radiation (PAR), stimulated chlorophyll- fluorescence (Chi), beam attenuation a t 660nm (C c ) and dissolved oxygen (DO). 10 km 100 km 1000 km 100 m 1 km 10 m 1 m 1 mm 1 cm 1 sec. Molecular Processes Surface Waves 1 mln. Individual Turbulent .Movement Patch Size Langmuir Cells Inertial/Internal Waves 1 hour Internal Tides Plankton Migration Synoptic Storms 1 day Fronts 0 1 ^ N Eddies /Seasonal MLD8 V VBiomass Cycles Zooplankton Patch 1 month I Gyre iCirculatk 1 year Figure 1.1. Illustration of temporal and spatial scales relevant to the physical and biological processes sampled during the Biowatt experiment. ' W H O ! Mooring Site i i m f Ifotni 80 75 70 65 60 55 Figure 1.2. Experiment site in the Sargasso Sea. Inset shows distribution of instrument packages within the water column. VMCM Orthogonal Current Meters Thermistor PAR Sensor Fluorometer Transmissometer I U \7 Dissolved Oxygen Sensor Figure 1.3. Configuration of Multi-Variable Moored System (MVMS) used for the Biowatt experiment. 3 These packages were part of a moored array which also featured a surface meteorological buoy which sampled heat and momentum fluxes a t the air-sea interface. The mooring was placed for three consecutive deployments, spanning 260 days, which allowed for a t least partial coverage of the full range of seasonal conditions. One continuous deployment was not feasible due to data storage limitations, battery life and data degradation due to instrum ent bio-fouling which is especially detrimental to optical devices which depend on transparent windows for data collection. This series of interdisciplinary moored experiments was the first of its kind at an oligotrophic site. Typically > time series measurements of bio-optical properties have been restricted in temporal resolution [Menzel and Ryther, 1960; Menzel and Ryther, 1961; Lohrenz et al., 1992; Winn et al., 1993], limited in experiment duration [Fasham and Pugh, 1976; Dickey et al., 1986; Vandevelde et al., 1987; Washburn et al., 1989] or done in coastal [Cullen et al., 1983; Whitledge and Wirrick, 1983; Falkowski et al., 1988] or limnological [Powell et al., 1975] environments. However, this experiment consisted of extended, high resolution time series of co-located physical and bio-optical measurements, sampled a t eight depths in the upper water column (down to 160m). The ability to obtain temporally extensive, high resolution measurements is a major asset of these moored systems since this minimizes temporal aliasing, a known error source in coarser resolution sampling schemes [Wiggert et al., 1994]. The data obtained from moored experiments are also ideal for aiding the development of interdisciplinary models since simultaneous measurements 4 surface forcing and in-situ ecosystem variables are necessary for assessing such a model's viability. These data are available over a considerable range of time scales which encompass a variety of relevant oceanic processes [Dickey, 1991]. Recently, there has been considerable interest in developing interdisciplinary (i.e., coupled physical and ecosystem) models in order to: 1) study the ocean's role in the global dim ate system (e.g., the potential impact of vertical carbon cycling on mitigating anthropogenic emissions); 2) investigate how individual compartments, which are poorly defined a t present, may contribute to the overall system and; 3) gain additional insight into physical/biological interaction mechanisms through process studies (i.e., ecosystem response to the application of episodic atmospheric forring or vertical oscillations) upon the development of a robust interdisdplinary scheme. The m aterial presented here is an accumulation of research into the areas presented in this introduction. Chapter 2 provides a review of previous oceanographic studies, both physical and bio-optical, in the Sargasso Sea and a brief overview of the essential time series results obtained from the MVMS packages. A more thorough presentation of the collected data is unnecessary as it has already appeared in the literature [Dickey et al., 1991; Dickey et al., 1993]. Chapter 3 presents the results obtained using bio-optical data to estim ate rates of gross primary productivity, net primary productivity and net community productivity, leading to the formation of a planktonic rate budget. Chapter 4 presents the results of spectral analyses carried out on the individual time series. These are used to isolate energetic events in the physical and bio-optical data within frequency bands which cover processes 5 w ith time scales of m inutes to weeks. These spectral analyses are also used to investigate coherence between signals as a means of identifying physical/biological interaction mechanisms. Chapter 5 presents the development of a 1-D interdisciplinary model and its application to the Biowatt site using the moored bio-optical time series and ancillary ecosystem data collected as part of the BATS/JGOFS time series to assess the model's predictive capability. 6 C h a n te r 2 ; T he B iow att E xperim ent Physical background for the site The dynamics of the region surrounding the experimental site have been studied extensively. The site is on the southern edge of a North Atlantic region which has the Atlantic's greatest heat loss to the atmosphere, due to evaporative heat flux driven by the wintertime presence of cold dry air flowing SE across the Gulf Stream [Bunker, 1976; Bunker and Worthington, 1976]. This heat flux is responsible, via processes akin to deep water formation in higher latitudes, for the generation of the 18° C water mass typified by tem peratures of 17.9±0.3 °C and salinities of36.50%«±0.10%o. These waters lie between the seasonal and perm anent therm odines of the Sargasso Sea and are seasonally ventilated north of 32N [Worthington, 1959]. The borders of the Sargasso Sea are commonly defined by the Gulf Stream to the north and west, the North Equatorial Current to the south and weak recirculation flow to the east. Temporal variability at this site is dominated by the passage of mesoscale eddies or Gulf Stream rings, associated warm outbreaks, and the growth and decay of the seasonal therm odine. Synoptic maps of sea surface tem perature (SST), generated using advanced very high resolution radiometer (AVHRR) images and data from ships of opportunity, reveal mesoscale advective features including cold core rings and warm outbreaks associated with the Gulf Stream and mesoscale eddies [Hitchcock et al., 1985; Comillon et al., 1986]. A previous study focusing on physical processes (the Long term 7 upper-ocean study or LOTUS) obtained two years of moored velocity and tem perature time series and a pair of three month data sets of current profiles (from 30 to 180 meters) a t the site. This study was designed to investigate the seasonal variability of energy levels in both low-frequency eddies and internal gravity waves (IGW) and the response of the IGW field to passing mesoscale features [Trask et al., 1982]. Both the moored time series and the profile data were used to form time series of energy in bands divided into sub-inertial, inertial and high frequency, the latter corresponding to internal waves. Time series of energy in the inertial band, taken as .94f -1.04 f i (where f\ is the inertial frequency corresponding to a period of 21.46 hrs.), reveal more energy in the late winter than during the summer due to increased wind stress during this time period. It was also observed that during the summer/fall period, when the upper water column is well-stratified, there is a greater difference between the energy levels of the inertial band and the .94/i - 1.64/i band since the increased stratification provides greater support for the internal wave field. [Briscoe and Weller, 1984; Eriksen, 1988]. Biological characteristics a n d interaction with the physical field The seasonal hydrodynamic cycle, in conjunction with the surface wind stress, plays a major role in establishing the magnitude of the spring phytoplankton bloom in the northern part of the Sargasso Sea, where a correlation has been observed between the intensity and frequency of winter storms and the magnitude of the following phytoplankton bloom [Menzel and Ryther, 1961]. W intertime ventilation of the 18° water m ass determines nutrient concentrations within the upper part of the water column, since deep w aters have greater concentrations than typical near-surface waters, due to remineralization processes [Siegel et al., 1990b]. W ith the onset of the spring, levels of photosynthetically available radiation (PAR) increase and net heat flux across the air-sea interface becomes positive, causing the surface waters to stratify. This results in an increase in the rates of primary productivity, the so-called spring bloom [Sverdrup, 1953], since the phytoplankton population receives more light over the course of each day, due to a combination of increased light level and restricted vertical motion, in an environment which is nutrient rich. As near-surface nutrients are utilized but light intensities continue to increase, the level of the maximum chlorophyll concentration moves deeper into the water column, forming what is termed the deep chlorophyll maximum (DCM) which persists through the summer. The high nitrate concentrations are quickly utilized however, resulting in a short-lived bloom followed by a transition of phytoplankton assemblage. The duration of the springtime phytoplankton bloom in the Sargasso Sea is on the order of 10-15 days, according to Coastal Zone Color Scanner (CZCS) images of near-surface chlorophyll concentration from 1985 [Smith et al., 1987]. The first phase of the Biowatt project, which occurred in 1985, was designed to study the spring bloom as it spread northward across the Sargasso Sea. This experiment consisted of a meridional transect along 70W (from 24N to 35N), during which a variety of in-situ profile measurements suitable for defining physical, optical, biological and chemical characteristics [Siegel et al., 1990b] were made a t a number of stations and analyzed in conjunction with AVHRR and CZCS images. By returning to stations visited previously on the southbound leg, bloom and post-bloom characteristics within the upper water 9 column (down to 250m) were obtained. During the bloom at 35N 70W, within 200km of the present site, the phytoplankton population was dominated by diatoms [Bidigare et al., 1990] which thrived in the high nutrient conditions. Upon returning to this site sixteen days later, the diatom community had been replaced by dinoflagellate-containing radiolarians near the surface, cyanobacteria with a maximum concentration near 40m and nanoplankton with a maximum concentration near 50m [Bidigare et al., 1990]. Seasonal cycles, observed itutcevious biological and geochemical studies in the Sargasso Sea Seasonal variations have been observed in a number of studies measuring a variety of biological characteristics at sites which are in the vicinity of Bermuda (around 32N 64W), making the acquisition of repeated data throughout the year logistically feasible. Time series, with resolutions ranging from 2 weeks to 2 months, of areal production, particle fluxes in the surface and deep ocean and oxygen concentration in the upper ocean have been obtained from this region of the western Sargasso Sea [Menzel and Ryther, 1960; Jenkins and Goldman, 1985; Deuser, 1986]. Generally, time series of areal production, from 1 4 C [Steemann Nielsen and Hansen, 1959] and bio-optical models, show m axim a in the late winter/early spring (February- April) and the late fall (October-November) associated with the onset and breakdown of seasonal stratification and m in im a in the mid-winter (December- January) ju st before seasonal ventilation occurs [Menzel and Ryther, 1960; Menzel and Ryther, 1961; Lohrenz et al., 1992]. Data from upper ocean sediment traps in the Sargasso Sea (150-200m) reveal tight temporal correspondence with the areal production time series ju st discussed while estim ations of £NOs' in the upper 100m have been observed to lead these maxima by about 11/2 months [Altabet, 1989]. Data from traps located slightly deeper in the water column (down to 400m) are not as well correlated with the surface and the possibility th at the maxima may be lagged a t time scales associated with particle sinking were not resolved by the experiments' temporal resolution [Lohrenz et al., 1992]. A composite annual cycle from six years of observations using deep ocean (3200m) sediment traps consists of a maximum in late April and a minimum in Oct./Nov. [Deuser, 1986]. It is interesting to note th at the springtime particle flux maximum is temporally coherent with the time series of areal production. Considering the sampling intervals associated with these studies (1-2 months) a lag between the surface and the deep time series of less than one month is not resolvable. However, this does provide a constraint with which to estim ate the amount of horizontal dispersion likely to occur between particle generation in the euphotic zone and settling to the deep ocean. Mesoscale currents are the driving mechanism for this horizontal redistribution and the relationship of dispersion to horizontal eddy kinetic energy (HKE) and sinking time has been considered in a previous work [Siegel et al., 1990a]. Typical levels of HKE in the Sargasso Sea are between 200 and 600 (cm2 /s2) depending on the proximity of mesoscale features [Granata et al., 1994]. This corresponds to a horizontal dispersion length scale of -800 km [Siegel e t al., 1990a] and serves to illustrate one of the difficulties in reconciling production estim ates from surface and deep waters [Platt, 1984]. 11 A composite seasonal cycle of AO 2 (defined as O2 mearC>2 M t) above 100m shows a maximum in August-September, just prior to stratification breakdown, and the surfacing of the zero line of the contour in March-April, corresponding to w ater column ventilation. A mirror image of the AO2 maximum (i.e., a AO2 minimum) appears about three m onths later in the waters between 10O and 200m. This lag is due to the breakdown of seasonal stratification which allows the organic m atter from summertime production to mix down into the aphotic zone where remineralization processes appear as oxygen consumption [Jenkins and Goldman, 1985]. This brief review of seasonal biological and geochemical cycles in the Sargasso Sea, from a region relatively near the Biowatt site, emphasizes the fact th a t these cycles are strongly modulated by annual and interannual variations in local hydrodynamic cycles, thus stressing the need for long-term, co-located, interdisciplinary measurements such as those provided by the MVMS packages utilized in the Biowatt II experiment. Description o f the present experiment O ur experiment consisted of four ship cruises each of which consisted of two legs. One leg was used to perform profiling studies near the site while the other w as used to deploy/recover the mooring system. Bio-optical profiling system (BOPS, [Sm ith et al., 1991]) and CTD casts (down to 200m) were made to obtain vertical profiles of conductivity, tem perature, q , Chi, PAR, diffuse attenuation coefficient ( K p a r ), upwelling spectral radiance and downwelling spectral irradiance, a t wavelengths described below for the moored version o f the sensor. These profile data illustrated, albeit coarsely, seasonal 12 evolution of each param eter [Smith et al., 1991; M arra et al., 1992; Dickey et al., 1993] and could be used as an aid in cahbrating/intercomparing the moored data, although horizontal spatial variability necessitated judicious use of this information. In addition, profiles of C c were utilized during the detrending operations applied to the C c time series in order to remove the effects of signal degradation due to the variation of the optical properties of the anti-foulant OMP-8 [Spinrad, 1987] applied to the transmissometer windows. Also, discrete w ater samples over the upper 200m were acquired during each cruise a t 8 to 12 depths. These were used in 1 4 C, nutrient and pigment analyses (using high-performance liquid chromatography). The primary productivity estim ates were used to determine an empirical function for apparent quantum yield which was then used in the generation of productivity tim e series from the moored data [Marra et al., 1992]. Finally, pigment analyses were carried out in order to determine the seasonal variation in species composition of the phytoplankton assemblage [Dickey et al., 1993]. The tim e series portion of these experiments consisted of a number of instrum ent packages attached to a taut-line mooring which was connected to a surface buoy containing a suite of meteorological instrum ents (Fig. 1.2). The meteorological package was used to quantify momentum and heat fluxes a t the air-sea interface by recording values of atmospheric pressure, wind speed and direction, short wave insolation, air tem perature, sea surface tem perature and relative humidity every 7.5 minutes [Weller et al., 1990]. Drag coefficients [Kondo, 1975] were used with these meteorological measurements in order to estimate the heat and momentum fluxes [Liu et al., 1984]. A sample of the measurements and derived surface fluxes are shown (Fig. 2.1). The in-situ 13 I*. Light Level T m M L D W ind Stress Net M eal Flu* S W Radiation Atm.Pressure (Meters) (sec) (Meters) (Pa) (W olfs/m 2 ) (W atts/m 2 ) (mb) J W ' ■ 4 1 0 2 9 1 0 2 2 J l O I S 4 1 0 0 8 J lO O l ! I 0 8 0 8 1 0 9 4 0 2 7 0 0 1I O 8O 1 7 2 0 1 3 6 0 r 3 6 7 2 1 0 8 144 60 1 0 0 140 180 220 260 300 JULIAN DAY 1987 3 2 64 96 128 160 2.1a 2.1b 1 * 9 6 0 2.1 C -fo.36 2.1 d 2 .1 e 1 0 7 t o * I 0 » 0 » 2.1 f 2-1 g Figure 2.1. Surface heat and momentum fluxes measured by the meteorological package on the surface buoy and derived variables describing the physical and optical state of the water column. Figures a-d show surface measurements of atmospheric pressure, short wave insolation, net heat flux and wind stress. Figure e shows mixed layer depth (MLD) based on the in-situ temperature measurements. Figure f shows mixing time scale which is derived from MLD/u*, where u* is surface friction velocity estimated from the wind stress. Figure g is 1% light level which is based on the in-situ estimates of photosynthetically available radiation (PAR). 14 instrum entation consisted of three bio-optical moored systems (BOMS), which provided high resolution measurements of downwelling irradiance (@ 410,441, 488,520 and 560 nm) and upwelling radiance (0 410,441,488,520 and 683 nm) [Smith et al., 1991],.along with the eight MVMSs (Fig. 1.2). Each MVMS package (Fig. 1.3) utilizes a vector measuring current m eter (VMCM), which consists of two orthogonal rotors for measuring horizontal currents [Weller, 1978; Weller and Davis, 1980]. The pressure casing of each VMCM was outfitted with a thermistor. In addition, a set of bio- optical instrum ents were mounted on a ’ backpack' which was attached to the cage surrounding each VMCM. The bio-optical instrum entation consisted of a beam transm issom eter for m easuring C c , used to estim ate particle concentration for particle sizes ranging from 1 to 80 pm [Bartz et al., 1978], a strobe fluorometer for measuring Chi [Bartz et al., 1988], a pulsed electrode dissolved oxygen probe [Langdon, 1984] and a PAR sensor for measuring broadband (400-700 nm) quantum scalar irradiance [Booth, 1976]. It m ust be remembered, when using the data gathered by the bio-optical instrum entation, th a t their variability may be the result of a variety of mechanisms such as phytoplankton physiology and ocean chemistry. The three param eters for which this is m ost likely to occur in the moored time series presented here are C c , Chi and dissolved oxygen (DO). Usage o f the bio-ontical instrum ents In the case of C c , the param eter of interest is a relative measure of particle concentration [Bishop, 1986; Spinrad, 1986]. In case I waters [Jerlov, 15 1976], away from sources of terrigenous input and resuspended bottom sediments, C c is useful for determining total particulate carbon and estimating net primary production [Siegel et al., 1989; Cullen et al., 1992a]. The theoretical beam attenuation coefficient (c) is defined as the sum of coefficients due to absorption (a) and scattering (b) and is considered to be an inherent optical property. The transm issom eter measures the attenuation of a collimated light beam (with a wavelength of660 nm) across a 25 cm column of water. The measured beam attenuation coefficient is derived from the percentage of light transm itted and is the sum of the attenuation coefficient due to clear ocean water (cw =0.36 n r1 , [Bishop, 1986]) and the attenuation coefficient due to suspended particulate m atter (SPM). Attenuation coefficients based on the presence of dissolved m aterials and detritus are generally disregarded, since they are not well quantified. For the analysis presented in this work, it will be assumed that C to t= C w + C c , where ctot and C c are the total attenuation coefficient and the attenuation coefficient due to living and non-living cells, respectively. Temporal changes in the latter will be used for estim ating net primary production and net community production. Beam attenuation, in addition to being a strong function of changes in phytoplankton concentration, has also been observed to reflect variations in scattering caused by size and shape distributions of particles [Baker and Lavelle, 1984] as well as changes in phytoplankton refractive index due to physiological a4justments to nutrient and/or light fluctuations [Kiefer, 1973; Morel and Bricaud, 1986; Stram ska and Dickey, 1992; Ackelson et al., 1993]. For chlorophyll fluorescence, the param eter of interest is an estim ate of chlorophyll concentration. These measurements are used in bio-optical models 16 which estim ate primary production [Kiefer and Mitchell, 1983; Bidigare et al., 1992; Kiefer, 1993]. The strobe fluorometer emits short pulses of light with a wavelength centered on 483 nm and measures the response of the local phytoplankton population a t wavelengths centered on 685 nm. Interpretation of chlorophyll fluorescence as chlorophyll concentration is not straightforward since the ratio of fluorescence yield to chlorophyll concentration (Fl/Chl) is not constant. These variations are due to changing environmental conditions (e.g., light, nutrients and/or temperature) which cause a4justm ents in phytoplankton physiology. These include photoadaptation, which maintains maximum photosynthetic efficiency via pigment-dependent changes, and photoinhibition, which is a self-protection mechanism triggered by a saturating light field [Falkowski, 1984; Falkowski and Kiefer, 1985; Prezelin et al., 1991; Falkowski et al., 1992]. The DO sensor used for these measurements combines a conventional membrane-covered polarographic oxygen sensor with pulse potential voltammetry. This approach removes errors traceable to fluctuations in steady-state currents seen in non-pulsed instrum ents [Langdon, 1984], In the present context, it is desirable to obtain estim ates of net production from the DO time series in order to compare with the assessments obtained with the two methods mentioned above, however, there are a number of processes governing the concentration of DO in oceanic waters, making the task of partitioning the relative contributions of each process difficult. Assuming variability due to advective processes to be minimal, concentrations of DO on daily time scales will reflect [Broecker and Peng, 1982]: 1) the diel cycle of the phytoplankton population since oxygen is produced during photosynthesis and consumed during respiration; 2) changes in w ater tem perature which determines oxygen solubility [Weiss, 1970]; 3) the chemical impetus for equilibrating the partial pressures of oxygen in the surface ocean and the overlying atmosphere; 4) bubble injection due to breaking surface waves driven by surface winds [Liss and Merlivat, 1986; Wannikhof, 1992] and; 5) remineralization processes which utilize oxygen to transform organic materials sinking out of the euphotic zone into inoiganic nutrients such as phosphate, nitrate and dissolved inoiganic carbon [Nqjjar, 1992], Presentation and d iscussion o f the B iow att moored interdisciplinary time series The in-situ time series analyses presented in this work will focus on the data obtained from the MVMS packages previously described. These were placed a t nominal depths of 14,23,43,62,81,101,120,160 meters. As mentioned previously, a series of three deployments resulted in a total time series length of260 days, with gaps of 5 and 2 days between deployments. Data from the sensor suite were recorded every four minutes on a tape storage device located in the pressure housing of the VMCM. A complete description of instrumentation, calibrations and data return is contained within an unpublished data report [Dickey e t al., 1990]. Data return is illustrated here by the stacked time series (Figs. 2.2-2.7) which show a lower return rate for the bio-optical measurements, due to bio-fouling and instrum ent malfunction. Variations in the parameters measured by the mooring have been described previously [Dickey et al., 1991; Dickey et al., 1993]. In these papers, 18 Temperature (C) f ji l l H f r o -IOO 1 0 0 o -IOO I O O 0 - 1 0 0 1 0 0 o -1 0 0 1 0 0 o -IOO IOO - . o o f e [ ______ A .u ^ r "•* JT Depth (m) 1 4 r4 3 -------- 6 2 8 1 IO I ■ 160 60 100 M O ISO 220 Julian Day 1987 280 300 Figure 2.2. Time series of averaged current stick vectors. Each vector represents the daily averaged current speed and direction. Units are in (cm/s). Julian Day 1987 Figure 2.3. Superimposed in-situ temperature time series. 19 C c (1/m) 1 1 1 1 1 1 •- 1 1 1 ‘ ‘ ‘ 1 1 ‘ ‘ « 1 1 » _ » » ‘ * 60 1 0 0 1 4 0 1 8 0 2 2 0 260 300 JULIAN D A Y 1987 Figure 2.4. Individual time series of photosynthetically available radiation (PAR). Time series at 120 and 160 meters are downward irradiance at 488 nm. 0.7 0.7 14m 0.7 0.7 0.7 0.7 0.7 0.7 r l l r f M W tt' ‘ . . .. . — |0 I V O - ------------------ 43 >41^. -■ ■ i--..-— 62 1 2 0 60 100 14 0 1 8 0 220 260 300 Julian Day 1987 Figure 2.5. Individual time series of beam attenuation coefficient (Cc). D O (|iM ) C hi (p g C hi a/1) 2.4 0 2.4 O 2.4 0 2.4 0 2.4 I [ Depth (m) 14 23 43 ■ j l n l r ‘ < i m r t i f m j *» AWMliU^Uifcpv^ J^IT "*I ,1111 ■ nil 11 - -. , 0 2.4r 2.4 0 2.4 1 0 1 120 1 6 0 ifrmn h 100 140 IS O 220 260 300 Julian Day 1987 Figure 2.6. Individual time series of chlorophyll-fluorescence (Chi). 26o |- I95F "— — — . ■ < ■ - - ■ ■ — 1 4 m 1 s r B 0 i 9 sF 260j- | ,-1_ _ _ _, ^_ _ _ _ _ _ _ _ _ _ _ _ 23 19 9 —. 260i I93(- ®4 •4 2 6 0 j- I9»f 26 I99i 2 193 1 0 1 26 1 9 9 1 261 I99| t 1 2 0 160 ■ ■ ■ ■ ■ ■ * * ■ * ■ .................. 60 10 0 140 I B O 220 260 300 Julian Day 1987 Figure 2.7. Individual time series of dissolved oxygen (DO). 21 time series from each of the three deployments have been broken up into six observational periods in order to facilitate interpretation. These periods are not of equal length and have been chosen based on roughly distinct oceanic regimes defined by current, tem perature, bio-optical or mesoscale features witnessed in the data set. D eterm in in g the presence of mesoscale influences within the time series was aided by SST images obtained from the NOAA Ocean Services U nit (Figs. 2.8-2.10). One image is presented for each of the observational periods in order to provide synoptic background for the site during each time period. Synopsis of in situ observations: Deployment 1 It is interesting to note th at the three deployments correspond to three relatively distinct physical and biological regimes. The atmospheric pressure and wind stress time series show elevated synoptic activity associated with the passage of winter storms typical of this time period. The oceanic environment of the first deployment is characterized by elevated mesoscale activity in the form of cold core rings and warm outbreaks from the Gulf Stream which have dimensions of 100 to 200 km and time scales of 10-20 days [Cornillon et al., 1986]. These features may be seen in the SST maps for the deployment (Fig. 2.8). The warm outbreaks in particular are easily recognized between Julian Day (JD) 60 and 80 in the stacked tem perature time series (Fig. 2.3) and the contour map of these tem perature data (Fig. 2.11). Following this period characterized by the relatively frequent passage of warm outbreaks is a regime showing deep mixing which is indicative of water column ventilation (Fig. 2.11 JD 75-95 at 160m). This deep ventilation period is followed by the 22 75W rc)W G S iW 3 JO 96 ; V , i Y H -#- 35N 7 c / H 73 W 70 W 65W JO 68 75 W 7 C IW 6! > W 2 ■ JO 64 , CO m :Pm :* ? ' Mooring b 73W 70W 65W JD 103 i- 40N 7 5W 70W 7 5 W 70W 65W Figure 2.8. SST maps providing a synoptic view of the Sargasso Sea in the vicinity of the mooring. The Gulf Stream path is shown. Gulf Stream meanders and its interaction with mesoscale eddies are indicated. These panels show six periods from the first deployment. 75W 7 0 W 65W JO 147 35N 65 W 70** 7 5 * 7! iW . 70W 65W 8 i g f j JO 170 y iT V # ✓ 0 5 S * '. \% T i 'I 1 t b 70** 6 9 * * 79** JO 203 3 9 N or 75* 7 1 3 * * 6!i* * 1 2 JO 236 I r l ! ® ......v :< f T O W 65W 75W 39N Figure 2.9. SST maps providing a synoptic view of the Sargasso Sea in the vicinity of the mooring. The Gulf Stream path is shown. Gulf Stream meanders and its interaction with mesoscale eddies are indicated. These panels show six periods from the second deployment. 7 S W 70W 63W JO 232 35N TOW 75W 35N JO 268 40N 33N 65W * JD 271 —- - - - - - - -40N 75 W ^ 7 1 3W 6 * sw i? JO 289 '."'3 % / 7s f o y I M M rif tf 35 N e JO 317 35N Figure 2.10. SST maps providing a synoptic view of the Sargasso Sea in the vicinity of the mooring. The Gulf Stream path is shown. Gulf Stream meanders and its interaction with mesoscale eddies are indicated. These panels show six periods from the third deployment. to cn perm anent onset of seasonal stratification in the surface waters (JD 85), though mesoscale activity may still be seen in the data. The current stick plots (Fig. 2.2) also indicate elevated energy levels during this deployment. The highlight of these tim e series are currents in excess of 1 m/s (~ JD 120) which rotate from a southwesterly to a northeasterly direction over a period of 45 days. This is due to the propagation of a cold core ring edge past the mooring (Fig. 2.8 d-f). The 14 and 23m moored tem perature time series (Fig. 2.3) appears to indicate th at this ring interacts w ith the Gulf Stream for a significant portion of this time period, forming a Class C outbreak [Cornillon et al., 1986] of unusually long duration. The optical properties of the water are driven by a combination of light level and water column stratification, since nutrients are not limiting at this time due to the recent occurrence of ventilation. Prior to JD 80, the presence of elevated levels of particulate m atter in the water column, seen in all depths of the beam c time series for JD 60-70 (Fig. 2.5), is due to the advection of m aterials within the warm outbreaks, the passage of which are evidenced by fluctuations recorded in the mixed layer depth (Z m id, defined here as the depth at which there is a 0.5° C difference from the surface, Fig. 2.1e). The occurrence of the extended ventilation event coincides w ith a sustained minimum in PAR throughout the water column (Fig. 2.4) and the dominant low pressure system recorded in the meteorological time series (Fig. 2.1a). The entire w ater column a t this tim e contains extremely low levels of particulate m atter. Evidence for this may be seen in the time series of 1% light level (Fig. 2.1g), which is beyond the deepest PAR sensor (101m), and in the contour map of K p a r (Fig. 2.12) which shows this to be a period of minimal light attenuation, 26 D epth (m) SO 40 60 60 1 1 00 140 180 S O O 240 280 1 2 0 160 260 200 2 2 0 Ju lian D ay 1987 Figure 2.11. Contour map of temperature. Units are degrees C. to 20 o - 30 40 O ' S O 60 70 0 .0 4 O 80 .0 .0 3 - 90 60 140 220 260 300 Ju lia n D ay 1987 Figure 2.12. Contour map of diffuse attenuation coefficient (Kpar), based on in-situ PAR. Units are (1/m). 27 though the w ater column is not completely devoid of particles since the .04 n r1 value ofKpAR indicated on the contour map is greater than the value range of K p a r (.029-.021 n r1 ) for d ear ocean waters between 25 and 70m [Baker and Frouin, 1987]. The C c time series from the upper 60m of the water column show m inim a during this period while the Chi and DO time series show no significant changes in magnitude from the beginning of the deployment, although all of the bio-optical time series continue to show high frequency variability (Figs. 2.5-2.7). The onset of seasonal stratification, seen in the stacked tem perature time series (Fig. 2.3) and the Z m id time series (Fig. 2.1e) following JD 85, coincides with the appearance of a phytoplankton bloom recorded in the time series of C c in the upper 60m (Fig. 2.5) and the 23m time series of Chi (Fig. 2.6). Dramatic decreases in the depth of the 1% light level (Fig. 2.1g) and the mixing time scale (tmsZmid/U*, where U* ( s \ l — ) is the surface friction velodty due \ Pw to wind stress, Fig. 2.1f) indicate increases in biomass (i.e., increased light attenuation) and increases in the daily dose of PAR (i.e., elevated production rates) respectively. It is interesting to note th at the near-surface DO time series (Fig. 2.7) do not show a corresponding response to the elevated levels of production indicated by the C c and Chi time series. In fact, the surface DO data show a significant decrease in concentration a few days following the productivity event. This may be due to surface gas exchange processes which tend to rectify the difference in the partial pressures of oxygen across the air- sea interface due to the increasing tem perature of the overlying air mass. The end of this deployment is characterized by increasing stratification (Fig. 2.3) as 28 well as an increase in the time series of C e and Chi after JD 110 (Figs. 2.5 and 2.6). Contour maps of C c and Chi (Figs. 2.13 and 2.14) show the development of a surface particle maximum (down to 60m) and the developmental stages of a sub-surface Chi maximum (between 40 and 70m) during this time. The 23m DO signal also appears to increase slightly in response to the increasing levels of primary productivity, while the 14m DO signal still appears to be responding to surface gas exchange processes (Fig. 2.7). Synopsis of in situ observations• Deployment 2 The second deployment is characterized by decreased surface wind stress (Fig. 2. Id) and decreased mesoscale activity around the mooring (Figs. 2.2 and 2.9). The mesoscale field shown in the synoptic maps shows a number of cold core rings in the area but they do not interact with the mooring, except for a couple weeks around JD 200 when a warm outbreak of Gulf Stream water, entrained by a cold core ring, is in the vicinity. This feature is clearly evident in the tem perature time series (Fig. 2.3) which, except for this mesoscale event, show increasing stratification until JD 220 (Fig. 2.15). Z m id is generally between 15 and 25m (Fig. 2.1e) throughout this time period while the strength of the horizontal velocity field (Fig. 2.2) is significantly reduced in comparison to the elevated energies witnessed in the first deployment. Due to limited vertical exchange caused by the physical environment, the near surface waters become nutrient poor resulting in low levels of Chi biomass, despite the presence of summertime light levels. The subsurface Chi maximum which began developing at the end of the first deployment continues to evolve, reaching its maximum (>0.64 |ig/l) between JD 140 and 160 at a 29 D epth (m ) D epth (m) SO .0 4 0 >0 6 0 .0 8 0 .0 1 00 .0 120 .0 1 4 0 .0 1 6 0 .0 1 0 0 . 0 1 S 0 . 0 S O O .0 3 5 0 .0 J u lia n D ay 1987 Figure 2.13. Contour map of Cc based on in-situ moored data. Units are 1/m. SO 6 0 0.38 B O 0 . 5 6 1 00 1 40 16 60 1 00 ISO 140 160 180 SOO SSO 8 4 0 8 6 0 8 6 0 5 0 0 5 8 0 J u lia n D a y 1987 Figure 2.14. Contour map of Chi based on in-situ moored data. Units are pg Chi a /l. . D epth (m ) D epth (m) so 40 60 80 O Q s 1 1 00 140 160 SOO 340 S60 S80 520 1 8 0 180 SOO J u lia n D ay 1987 Figure 2.15. Contour map of stratification based on in-situ temperature measurements. Units are deg. C/m. SO 40 60 80 1 1 00 140 160 180 SOO 2 2 0 860 8 8 0 500 5S0 J u lia n D ay 1987 Figure 2.16. Contour map of shear based on in-situ velocity measurements. Units are 1/sec. depth of 90m and consisting of an envelope which extends between 50 and 120m (Fig. 2.14). This is reflected in the contour m ap of K p a r (Fig. 2.12) which indicates values greater than 0.06 m 1 in the chlorophyll maximum region. After the passage of the dominant mesoscale feature in the deployment (JD 200), the subsurface chlorophyll maximum begins to dissipate, consisting of a reduced envelope extending between 50 and 70m. Synopsis of in situ observations: Deployment 3 The third deployment is characterized by the transition from the strongly stratified, low energy current regime of the summer to the breakdown of seasonal stratification and mixed layer deepening. This is caused by a decrease in surface heat flux while wind-induced mixing is becoming more prominent due to an increase in frequency and magnitude of low pressure atmospheric systems and the associated periods of elevated wind stress which combine to deepen the mixed layer (Fig. 2.3). There is also an increase in mesoscale activity which is visible in the tem perature and current measurements (Figs. 2.2,2.3). The SST maps (Fig. 2.10) do not effectively depict this increase in mesoscale activity, however, the sea surface height (SSH) maps [Dickey et al., 1993] show the increase in mesoscale behavior near the mooring. The subsurface chlorophyll maximum, which had already begun to dissipate by the end of the second deployment, was gone by JD 300 (Fig. 2.14). However, the end of each Chi time series between 23 and 101m shows an increase starting around JD 310 (Fig. 2.6), due to the entrainm ent of nutrients into the surface w aters as conditions become favorable for vertical exchange with the deep waters. C h an ter S: P rim ary pro d u ctiv ity b ased on m oored tim e series Introduction to nhvtnplankton photosynthesis Time series of PAR, Chi and (< are utilized to estim ate rates of gross (GPP) and net (NP) community production (mgC/m3 /d). GPP, (Eq. 3.1), represents the photosynthetically driven fixation rate of cell carbon. It is defined in term s of a gross carbon fixation rate (g) consisting of the specific growth (|x), specific respiration (r) and heterotrophic grazing (h) rates of a planktonic community (Eq. 3.1, where C is phytoplankton biomass in mgC/m3 ). GPP = gC = (p + r + h)C 3.1) The specific respiration rate of a cell suspension is described empirically as the sum of a basal respiration rate (ar = .046 /day) and a linear dependence on the specific growth rate (Eq. 3.2). Generally, 10-20% of p is lost to respiration [Kiefer and Mitchell, 1983]. r = ar + br p 3.2) NP (Eq. 3.3) represents the photosynthetically driven fixation rate of cell carbon less losses due to cell respiration and heterotrophic grazing. NP = (g - r - h) C. 3.3) 33 Models, of gross. j2 rodu£tion GPP is estimated here using the PAR and Chi time series (Figs. 2.4 and 2.6) as input to an absorption-based optical production model [Kiefer, 1993]. A number of production models have appeared in the literature and may be separated into two basic groups depending on whether the model is absorption based or P-I based. Generally, formulation differences are m anifest in the param eterizations of quantum yield ( < j > ) and chlorophyll specific absorption (achi) for the absorption based models [Cullen, 1990; Bidigare et al., 1992] or the functional form of the photosynthesis-irradiance curve used in the P-I models [Jassby and P latt, 1976]. As the field of in-situ optical measurements has progressed, advances in productivity modeling are being driven by developing abilities to obtain spectrally decomposed data for both irradiance and phytoplankton absorption [Bidigare et al., 1987] which may be incorporated into spectral primary productivity relations [Morel, 1991; Bidigare et al., 1992], When estimating GPP, an understanding of instrum ent limitations and photosynthetic processes and the latter's manifestation within the optical signals is required. This is due to basic physiological processes occurring during the normal diel cycle [Prezelin, 1992], adaptive strategies (e.g„ photoinhibition and quenching) employed under stressful environmental conditions [Richardson et al., 1983; Falkowski, 1984], and bio-fouling which may result in significant signal degradation during extended deployments [Spinrad, 1987]. Additionally, the use of stimulated fluorescence as a direct indicator of chlorophyll 34 concentration, phytoplankton biomass or photosynthetic rate is inherently uncertain [Falkowski and Kiefer, 1985; Cullen et al., 1988; Marra, 1992], The recently developed pump and probe or double flash fluorometer is capable of quantifying several characteristics of phytoplankton photosystems. Assessment of adaptive strategies (e.g., quenching) and direct determination of physiological characteristics (e.g., maximum quantum yield O f t m ) and minimum steady-state reaction center turnover time (tp h)) is possible with this fluorometric technique [Falkowski et al., 1992]. Knowledge of these attributes allows for ascertaining photosynthetic rates which more accurately reflect carbon fixation under stressful environmental conditions. In-situ m easurements using this device have only recently become available. These data have been used to estim ate photosynthetic rates which agree well with estim ates determined using passive fluorescence and standard 1 4 C uptake m easurements [Kolber and Falkowski, 1993]. The irradiance field is the environmental characteristic with the most direct effect on phytoplankton growth [Langdon, 1988]. In addition, photosynthetic rate is affected by tem perature [Eppley, 1972] and the concentration of both dominant (e.g., [Laws and Bannister, 1980; Cleveland et al., 1989] and trace [M artin et al., 1989; Falkowski et al., 1992] nutrients. However, Carbon:Chlorophyll ratio (0) varies significantly as a physiological response to variations in these ecosystem param eters. Thus changes in specific growth rate (|i) do not correspond to changes in chlorophyll specific photosynthetic rate (PB ). In fact, a t growth irradiance, it has been found that pB is typically independent of nitrogen limited growth rate [Sakshaug et al., 35 1989; Cullen et al., 1992b] indicating th at phytoplankton growth is adapted to a nominal performance range which is chiefly modulated by irradiance. Cf as it rela tes to n e t production a n d p a rticle specific accum ulation ratea Light in the oligotrophic ocean is attenuated by water, phytoplankton, detritus and dissolved materials. The transmissometers used in the moored experiment measured the summation of these contributions to attenuation with a collimated beam of light centered a t 660nm with a spectral line half width of 20 nm. This wavelength was chosen to utilize a stable light source and reduce the significance of dissolved m aterial attenuation [Bartz et al., 1978; Mobley, 1994]. Furthermore, w ater attenuation is constant and may be disregarded when investigating temporal changes in C e . Thus, variability in C c may in principle be attributed to the daily accumulation of detritus and physiological processes associated with phytoplankton photosynthesis. The presence of these diel processes within the C c data was accentuated by subtracting the daily mean from each day’ s signal. The resulting signal, termed A q. [Siegel et al., 1989], generally consists of a minimum near dawn and a maximum near sundown. This idealized pattern, based upon local ecosystem dynamics (cf. Fig. 11 in [Cullen, 1991]), is extremely sensitive to alterations introduced by advected water masses and vertical mixing. Estim ates of NP have previously been made assuming a linear relationship between diel changes in c « and SPM [Bishop, 1986] and a constant percentage of particulate organic carbon (POC) within SPM [Bishop et al., 1986]. Thus a 36 constant coefficient for carbon specific beam attenuation (c% ) is utilized to estim ate NP [Siegel et al., 1989]. The assumptions utilized in estimating NP from diel changes in C c are presently undergoing further study. Attributing all of the diel change to changes in SPM has recently been challenged. Laboratory studies indicate th at observed changes in C c may be equally partitioned into changes in particle concentration and cellular attenuation cross-section [Stramski and Reynolds, 1993]. Additionally, there are indications that c * c is not constant. Calculations have indicated th at a 2% change in c% per hour would lead to a 60% overestimation in NP over the course of a day [Ackelson et al., 1993]. However, detailed investigations revealing variations in c * e have yet to be performed and it is presently unclear whether this behavior appears outside of laboratory experiments during which irradiance levels were 2-3 times higher than values typically found in the natural regime [Ackelson et al., 1990; Ackelson et al., 1993]. These findings indicate th at estim ates of NP from A c ^ may be in error by 0(100%) due to the inability to differentiate between sources of variability within the C c signal and present knowledge (or rather lack thereof) concerning natural variability in c * c [Cullen et al., 1992a; Ackelson et al., 1993; Stramski and Reynolds, 1993]. 37 Methods Estimating gross growth rates. GPP and C:CKL ratios The Kiefer model 1 4 C estim ates of prim ary production were made during the four deployment/recovery cruises. Water samples were taken a t dawn and in-situ incubations were carried out until dusk. Four replicates were placed at each depth and spaced 5-20m ap art through the euphotic zone. Details of the techniques employed have been published previously [Marra et al., 1992]. The 1 4 C measurements of daily production are used to tune the optical production model through comparison with temporally corresponding estim ates of daily The optical production model used to estimate GPP is summarized by the following equation set which includes relations for predicting 0 (C:Chl ratio) and < | > (quantum yield) [Kiefer, 1993]. GPP. GPP(z,t)=achi(z,t) Io(z,t) (jK z.t) C h l(z,t) 3.4) achi(z,t) ( j ) m Io(z,t) + P 0K (z,t) < t> m P QicCz.t) 3.5) p feat(Nph,T) D 3.6) 38 +(a t h |(z'b9 'c p < l > m )2 ]1 * 3.7) Calibration of the MVMS's stimulated fluorometers has been previously described [Marra, 1992]. The resulting chlorophyll-fluorescence time series are input as instantaneous chlorophyll concentration (Chi in equation 3.4) to the optical production model (hereafter referred to as the Kiefer model) along with the time series of instantaneous PAR (Io in equations 3.4,3.5, and 3.7). Table 3.1, previously reported in M arra et al., 1992, shows chlorophyll-specific absorption coefficients (achi(z,t)) which were interpolated from profiles taken during the deployment/recovery cruises. Depth (m) OC1 OC2 OC3 OC4 14 0.0080 0.0190 0.0120 0.0080 23 0.0068 0.0170 0.0120 0.0096 43 0.0086 0.0130 0.0130 0.0085 62 0.0080 0.0120 0.0110 0.0094 81 0.0095 0.0130 0.0097 0.0087 101 0.0095 0.0140 0.0120 0.0087 T able 3.1: C hlorophyll specific ab so rp tio n. T hese values w ere o b tain ed d u rin g th e fo u r m ooring deploym ent a n d recovery cru ises (OC1-OC4). U nite The full expression for equation 3.5 was generated by inserting the expressions for instantaneous light-saturated, carbon-specific photosynthetic 39 rate (P) and 0k, (equations 3.6 and 3.7). This reveals < f > to be explicitly dependent upon both I„ and the photoperiod (D), calculated given latitude and Julian day. The constant 0m (minimum C:Chl with a value of 36 g C/g chlorophyll a) was taken from the literature [Sakshaug et al., 1989]. Since nutrient (N p h) information was sparse and the tem perature range at the Biowatt site (18.5‘C to 27.5*C) was considered a minor influence, the light- saturated, carbon-specific cell growth rate (g»*t(Nph,T)) was held constant [Kiefer and Mitchell, 1983; Kiefer, 1993], Error m inimization GPP has been previously estim ated with this data set [Marra et al., 1992]. However, differences between the methods used for the calculation exist and it is illuminating to compare and contrast the results of the two schemes. The previous model implementation, hereafter referred to as the LDGO method, attem pted to emulate the 1 4 C incubations. This consisted of obtaining a dawn value for Chi from the time series and using it as a constant for the day in conjunction with the daily integrated value for PAR Emulating the 1 4 C technique disregards the data's resolution and leads to aliased predictions since it removes high frequency variability within the Chi signal driven by fluctuating light levels. The Kiefer model makes full use of the high resolution measurements, resulting in essentially instantaneous GPP estim ates. Two model constants ( < t > m and be) were chosen based on an error minimization study for which model implementation emulated the 1 4 C incubations, as ju st described for the LDGO 40 method. In addition to reproducing the 1 4 C results, the tuning process was considered a necessity since the original constants [Kiefer, 1993] resulted in no difference between estimates made with and without a nutrient limitation param eterization (Eq. 3.6). The average percent error with respect to the 1 4 C measurements was calculated and the constants, consisting of 4 > m (maximum quantum yield) and be (the empirical param eter in Eq. 3.7), which minimized this error estimate (0.084 mol C/mol photons and 4.66, respectively) were chosen. A comparison of the GPP results from this error minimization process and the 1 4 C values is shown (Fig. 3.1a). The average percent error between the model estim ates and the in-situ data is 39.9%. Comparison of the GPP results using the full resolution time series for both PAR and Chi is shown (Fig. 3.1b). In this case the average percent error is 45.4%. Finally, comparison between the 1 4 C data and model estimates obtained using the LDGO method (Fig. 3.1c) have an average percent error of 50.6%. The fact th at the discrepancy with respect to the in-situ production values increases between applying the Kiefer model with a constant Chi value and using the full resolution Chi raises an interesting issue. Since the moored instrum entation provides a view of production processes which is distinct from th at supplied by the 1 4 C incubations, it is unreasonable to expect a precise correspondence between the two measurements. Thus, applying the model in a m anner which emulates as closely as possible the in-situ technique, in order to provide a mechanism for tuning the equations, is the best th at should be expected (Fig. 3.1a). Once this is accomplished, time series of GPP obtained using the Kiefer model on the mooring data may be used to investigate temporal variations in production processes a t greatly enhanced resolution. 41 M o d e l R esu lt (mgC/m3 /day) A verage E rror = 39.9% o m o 0 5 10 1 5 3.1a C14 D ata (mgC/m3 /day) E \ o O ! E ■ o o s m A verage E rror = 45.4% o in o 0 5 10 1 5 3.1b Cl4 D ota (mgC/m3 /doy) 42 m 3.1c Average Error = 45.4% > s 5 O E 3 i n v tr o 0 5 10 15 C14 Data (m g C /m 3/d ay ) Figure 3.1. Comparison between gross production measurements obtained from ship based incubations and estimates made by optical production models, a) Model estimates made using a dawn value for Chi and daily integrated PAR with the Kiefer model (Eqs. 3.4-3.7). b) Model estimates made using full resolution time series of Chi and PAR with the Kiefer model. Daily values used for this comparison are based on integrating the high resolution GPP estimates, c) Model estimates made using a dawn value for Chi and daily integrated PAR with the LDGO model (Marra et al., 1992). The average percent error, based on comparison to the ship data, is given. 43 Estim ating C.Chl ratio In addition to estimating rates of gross production which are used in calculating GPP, the Kiefer model provides predictions of cellular C:Chl ratio (Eq. 3.7). The importance of 6 k within the context of predicting GPP was emphasized by a series of model calculations designed to investigate the effects of nutrient limitation. Growth rate limitation was parameterized by incorporating a term in equation 3.6 which consisted of a standard Michaelis- Menten type hyperbolic relation. N utrients (i.e., NO3’ concentration) were derived using a tem perature-nitrate relation based on bottle data obtained during the deployment/recovery cruises and GEOSECS cruise data. The latter were included in order to have values representative of deep water concentrations. These limitation calculations brought to light the sensitivity of the production model to the values of < | ) m , which has been observed to vary by a factor of 2-3 due to a variety of ecosystem characteristics [Cleveland et al., 1989; Prezelin et al., 1991; Kolber and Falkowski, 1993], and the empirical constant (be, Eq. 3.7). Laboratory studies have shown th at 6 varies by an order of magnitude and th at this variability is depends strongly on irradiance intensity [Sakshaug et al., 1989]. This physiological characteristic is incorporated within equation 3.7 and provides for predicting C:Chl with the in-situ data set. However, for Np h with values of 0.1-2.0 (J.M , the laboratory based value for be (0.76) resulted in no variation between GPP estim ates made with and without the nutrient lim itation parameterization, suggesting th a t the relative weights of be and P (Eq. 3.7) were incorrect. Changing the value of be to th a t determined during 44 the previously described error minimization process resulted in a slight decrease in GPP with the incorporation of nutrient limitation. Thus, this model can reproduce the phytoplankton's ability to m aintain nominal performance for a range of environmental conditions through the adjustment of physiological traits [Laws and Bannister, 1980; Prezelin et al., 1991; Cullen et al., 1992b], Estimation o f deployment averaeed net production. growth and fra zin f rates Deployment averaged NP was estim ated using the average diel range of C c and a constant for c% (3.92 • 1 0 * 3 m2 / gC [Hobson, 1967; Siegel et al., 1989]). Higher resolution estim ates of net production and particle accumulation rates were also calculated. The temporal resolution of the C c time series was reduced by creating one hour averages. Then, eight days were averaged to form composite diel time series of C c and A ce with one hour resolution. The hourly averages were created to improve the results of the rate calculations while eight day averages were created in order to improve the characteristics of the diel cycle and the NP estim ates made from the observed extremes. Particle specific rates (pc [1/day]) for both light (i.e., accumulation) and dark (i.e., loss) periods were determined (Eq. 3.8). It has been noted that specific particle production rates determined during the light period were overestimates since optical properties and phytoplankton biomass do not directly correspond [Stramski and Reynolds, 1993]. In order to base the particle specific accumulation/loss rate calculations solely on variations in cell C, the diel composite values of C c are reduced (ce, Eq. 3.9) through multiplication by the ratio of cell C:total C (0*, Eq. 3.10a). Values of 0* are 45 based on total C:Chl (6^, Eq. 3.10b) and estimates of cellular C:Chl from the Kiefer model ( 6 k , Eq. 3.7). 6 * ranges from 0.2 through 0.45 and increases with depth (Fig. 3.2). It should be remembered, th at in this context, Pc refers to the apparent rate of carbon fixation and serves as an upper lim it with respect to the rate of cell production. Pc = At ^ cl(t^)} ] ; A t- t„+ i - tn ■ 1 hr 3.8) Ce(tn) = 0*(tn) • C c(t„) 3.9) 0to(tn) = C h lfe ^ c* c 3'10b) Hourly particle specific rate estim ates, partitioned into light and dark rates based on the PAR signal, are determined for each day. Values within each category are summed and the mean accumulation rate (< pa >) and loss rate (< pi >) are determined for each diel composite. The accumulation and loss rates may be incorporated into the relation for NP (Eq. 3.3) which now takes the form NP = < pa > C 3.11a) where 46 <Pa> =g-(r + h) 3.11b) and <Pi> = r + h. 3.11c) Additionally, light period averaged gross carbon fixation rate (< g >) may be estimated with the Kiefer model using the following relation obtained by combining equations 3.1 and 3.4, where < > denotes mean carbon fixation rate over the light period, identical to the averaging scheme used to determine < pa>. Thus, a budget may be formed using equations 3.11 b-d which allows for a comparison between gross production rates determined by the Kiefer model This relation may be used as an indicator as to the degree to which closure is achieved between the independently derived rate estimates. Additionally, assuming that phytoplankton respiration is constant over the diel cycle, the loss rate may be partitioned into contributions due to respiration and grazing using the laboratory determined relation between phytoplankton 3.11d) and the cell C accumulation and loss rates determined using the ct data (Eq. 3. lie ). < g > = H +r + h = <pfc>+ <p,> 3. lie ) 47 respiration and specific growth (Eq. 3.2), where br is taken to be 0.15. Combining this respiration rate estim ate with the estimates for POC accumulation rate (Eq. 3.11b) and gross carbon fixation rate (Eq. 3.11d) results in the following relation for estimating the grazing rate. h = <g>-<pi>*(l + br)-ar 3.11£) Some estim ates of < p* > and < pi > were culled from the obtained data set based on strong deviation from closure (Eq. 3.lie ). These values coincided with particulate signatures associated with advective features. Contributions to this rate budget due to sinking, advection and diffusion have not been included since the degree to which they influence these data is difficult to ascertain [Harris, 1986]. R esu lts Description of GPP estimates and comparison to previous results High-resolution time series of GPP have been calculated. These were integrated to form time series of daily estim ates (Fig. 3.3). A contour map of these GPP time series was also created (Fig. 3.4a). It indicates elevated production, within the upper 40m, starting around JD 85 and lasting through JD 130. This is accompanied by a subsurface (i.e., below 40m) production feature which begins around JD 120 and extends deeper into the water column as time passes. After JD 240 deep production decreases while near-surface production shows two episodic increases, one around JD 220 and the other 48 2 1.9 • u.e ■ I 21.9 14.9 7.3 21.9 14.6 21.9 14.6 7.3 21.9 14.6 ( 0 I •>21.9 ■ 14.6 • 7.3 13.6 21.9 14.6 7.3 23.3 21.9 14.6 7.3 42.6 2 1.9 14.6 7.3 62.1 21.9 14.6 7.3 61.2 21.9 14.6 7.3 I Julian Day 1987 Figure 3.2. Time series of daily integrated gross primary productivity. These estimates are based on the full resolution application of the Kiefer model (Eqs. 3.4-3.7). The time series shown here are daily integrated values. S O 60 60 90 1 8 0 Julian Day 1987 Figure 3.3. Contour map of 6* from 50 to 100 meters. This represents the ratio of predicted phytoplankton cell carbon (Eq. 3.7) to estimated total particulate carbon within the water column based on the in-situ measurements (Eq. 3.10b). 49 around JD 310. Thus the seasonal cycle of prim ary production a t this site typifies an environment where elevated stratification through the summer causes a deepening of the maximum production region by restricting the flow of nutrients into the surface waters. Comparison to results from the LDGO method (Fig. 3.4b), reveals th at near surface values are around 50% higher than those obtained from the Kiefer model while below 40m, the production estim ates of the Kiefer model are 2-3 times higher. This illustrates th at in comparison to the LDGO model, production estim ates from the Kiefer model are higher at low irradiance and lower at high irradiance. The time series of water column integrated gross production is shown (Fig. 3.5). As seen in the contour map, this tim e series shows a springtime increase in gross production. Extremes through the first part of the summer (JD 170) range from 400 to 800 mg C/m2 /day but values are generally constant (~ 575 mg C/m2 /day). Thus, although the surface m anifestation of the spring bloom is relatively short-lived, the development of the subsurface production maximum results in continuing elevated production within the euphotic zone for an extended period. The average value over the entire time series is 439 mg C/m2 /day. This compares favorably with the estim ate for yearly production first reported (419 mg C/m2 /day, [Marra et al., 1992]) and is indistinguishable from an adjusted estimate o f449 mg C/m2 /day reported recently [Waters et al., 1994], This adjustm ent took the form of increasing the first deployment estim ate of areal (i.e., vertically integrated) production from 125 to 188 mg C/m2 /day [Waters et al., 1994]. This was deemed necessary in order to rectify the discrepancy observed between production estim ates based on natural fluorescence data obtained with the BOMS packages [Smith et al., 50 0 ) SO 40 a a « Q 60 80 1 00 80 SOO 200 520 Ju lia n Day 1987 X 7 3.4 b SO 40 60 0 0 100 8 0 160 SOO S80 Ju lia n Day 1987 Figure 3.4. Contour maps of GPP. a) Production estimates based on applying the Kiefer model to the full resolution data and performing daily integrations, b) Production estimates based on applying dawn values of Chi and daily integrated PAR to the LDGO model. 1991] and the simultaneously obtained stim ulated fluorescence data obtained with the MVMS packages [M arra et al., 1992]. A visual comparison between the depth integrated first deployment results of the Kiefer (Fig. 3.5) and LDGO (Fig. 13 in M arra et al., 1992) methods show that over the first 30 days of the experiment, prior to the spring bloom, the Kiefer values are about twice as high as the LDGO values. However, the entire first deployment contributes 99 mg C/m2 /day to seasonal production which is lower than both of the previously reported values just discussed. This discrepancy further emphasizes the differences between the two formulations and the manner in which they utilize the available time series. However, without further analysis, it is difficult to partition the causes of the observed discrepancy between the differing methods for estim ating quantum yield and the increased resolution of the input data. Estimated N et Production Composite signals of A ce for each deployment have been calculated for all depths down to 101m (Fig. 3.6). These show the general characteristic witnessed in previous studies consisting of daily minima and maxima occurring near sunrise and sunset, respectively. The spring season has the largest peak to peak difference for the signals. This is especially pronounced in the surface w aters where the 14m signal has a difference greater than .02/m between extrema. The average A ce signal, for all seasons, at 101m, does not show the daily cycle witnessed in the upper depths. This agrees with previous results in which no statistically significant trend was found in A C c signals below 95m 52 o o o > s o T ) \ N E \ o ? § t in o 100 200 150 250 300 Julian Day 1987 Figure 3.5. Time series of areal production based on the Kiefer model results. The average daily value has been calculated (438.5 mgC/m2/day). This implies annual production at the site to be 13.3 Mol C/m2/yr. This value may be inflated by the poor representation of wintertime production. 53 - 0.01 Fraction 0/ Day Figure 3.6. Deployment averaged diel cycles of Ac. Units are (1/m). These are obtained from the beam transmissometer data and consist of averaging together all 24 hour subsets contained within the full record for each deployment. The mean has been removed in order to emphasize the amplitude of the diel signal. © i [Siegel et al., 1989], generally just above the 1% light level (Fig. 2.1g). Peak to peak values for each signal have been transform ed to estimates of deployment averaged NP. These estim ates have been plotted as water column profiles (Fig. 3.7). Deployment averaged profiles of GPP (Eqs. 3.4-3.7) were also created to provide a comparison between the two estim ates. The spring profile of NP is the most productive over the three deployments and shows values which are 30*50% higher than those of the GPP profile, with the areal value showing a difference of 25%. The NP profiles from the remaining two deployments show consistent areal values (278 and 274 mgC/m2 /day) which are lower than the corresponding values for the GPP profiles. Indeed, the gross production values are consistently higher in the profiles from the last two deployments except where Chi concentrations are consistently low (23m in the second deployment) and near the base of the euphotic zone in the third deployment. The contour map of NP (Fig. 3.8), made using the composite estimates, shows features sim ilar to those indicated by the deployment averaged profiles (Fig. 3.7). Values during the first deployment are consistently higher than those found in the GPP map (Fig. 3.4a). With the development of the subsurface chlorophyll/GPP maximum (JD 120) the estim ates of NP are consistently lower save for a brief period around JD 240 a t 40m during which they are essentially identical. This feature is temporally similar (JD 220-240) to the local maxima which appears at the surface in the GPP map and, based on its shape, is indicative of a downward particle flux. This occurs during a period of fluctuating MLD (Fig. 2.1e) which precedes the onset of increasing 55 40 (622 mgC/m'/dny) ft 4) Q 80 Spring 100 10 0 2 4 6 8 20 a w •a ■ M 40 §* 60 a 80 S u m m e r 100 10 0 2 4 6 8 20 a ja 80 100 0 2 4 6 8 10 Production (mgC/m3 /day) Figure 3.7. Comparison between GPP (dashed) and NP (solid) profiles. The GPP profiles are based on deployment averaged production estimates obtained using the Kiefer model (Fig. 3.2). The NP profiles are based on the deployment averaged diel cycles of A c (Fig. 3.6). 56 D epth (m) 8 0 40 60 80 1 1 8 0 180 8 4 0 Julian Day 1987 Figure 3.8. Contour map of net production (NP). The data shown here are the eight day composites. These were formed by hourly binning the full resolution Cc data, averaging eight 24 hour cycles then multiplying the peak to peak difference of ech diel cycle by C*c. 57 vertical mixing driven by thermal convection. Estimated, productivity, respiration and erazins rates The deployment averaged rate estim ates of productivity (< g >, p), respiration (r) and grazing (h) are presented (Table 3.2) along with estim ates of the rate budget (Eq. 3.lie ) and the fraction of < g > allocated to h in order to fulfill the assumed steady state balance. Negative values for h and h/< g >, corresponding to negative values for the rate budget, are not reported since this is an artifact of being at or below the base of the euphotic zone (Fig. 2.1g). The results in this table show th at < g > is well-defined (i.e., estimated error is £ ±37.5%) whereas for the C c based rates (Eqs. 3.11 b, c), the estim ated error ranges up to ±500% for the worst case although it is generally < ±100%. The values for estim ated specific growth rate (p) range up to 0.29/day but do not approach the corresponding estimates for < g > within the euphotic zone. This sharply contrasts the indications provided by the comparison between the NP estim ates based on A ce and GPP. Taken together, values of p+h range from 0.16/day up to 2.16/day. The budget is made using the directly estimated values and provides insight into the viability of using these calculations to determine rates of respiration and grazing. Due to the uncertainties associated with the C e based values, the budgets achieve statistical closure in only half of the cases. The dark respiration value estimated using equation 3.2 is then used in this budget to obtain an estim ate of h. These values range from 0.09/day up to 0.98/day over all depths and deployments contained within Table 3.3. Additionally, the 58 Gross Light Period Dark Period Rate Budget Respiration Grazing Percent Accumulation Loss Utilization <g> I <pa> <pi> < g > - <pa> - <pi> r h h/<g> 1 Eq. 3.lid | Eq. 3.11b Eq. 3.11c Eq. 3.11e Eq. 3.2 Eq. 3.11f D ep loy: 23 m 62 m 101 m 0.82 t 0.26 0.46 ± 0.06 0.10 ± 0.02 0.29 ± 0.12 0.37 ± 0.14 0.03 ± 0.09 0.09 ±0.19 0.25 ± 0.19 0.48 ± 0.16 D ep loy 2 1 62 m 0.79 ± 0.09 0.15 ±0.10 0.35 ± 0.20 0.16 ±0.39 0.09 ±0.02 0.44 ± 0.36 0.47 ± 0.28 0.34 ± 0.22 0.05 ± 0.01 0.39 ± 0.09 0.84 ± 0.24 -0.63 ± 0.32 0.08 ± 0.03 0.29 ± 0.34 0.07 ± 0.01 0.57 ± 0.17 0.71 ± 0.16 81m 0.32 ±0.12 0.15 ± 0.08 0.46 ± 0.44 101m 0.12 ±0.02 0.12 ±0.09 0.21 ± 0.13 D ep loy 3 1 43 m 1.12 ± 0.20 0.08 ± 0.09 0.11 ± 0.08 62 m 0.58 ±0.09 0.08 ±0.09 0.10 ± 0.11 81m 0.22 ±0.03 0.02 ±0.10 0.06 ± 0.13 101m 0.22 ±0.03 0.07 ±0.12 0.07 ± 0.19 -0.28 ± 0.38 0.07 ± 0.01 0.10 ± 0.16 0.24 ± 0.43 -0.22 ± 0.20 0.06 ± 0.01 0.93 ± 0.24 0.06 ± 0.01 0.98 ± 0.21 0.87 ± 0.09 0.41 ±0.20 0.06 ± 0.01 0.45 ±0.14 0.76 ± 0.20 0.15 ±0.19 0.05 ±0.01 0.16 ±0.10 0.74 ± 0.50 0.08 ± 0.27 0.06 ± 0.02 0.09 ± 0.13 0.42 ± 0.61 Table 3.2: Primary productivity growth and loss rates determined using the Kiefer model and the transmissometer data. These are combined to form a rate budget in each deployment for depths where sufficient data is available. The error estimates are based on the standard deviation for the averaged values recorded here. These are averages of the eight day composites described in the text. All units are (1/day) except for h/<g> which is non-dimensional. cn to percentage of < g > removed by heterotrophic grazing (i.e., h/< g >) ranges from 0.24 to 0.87. Areal estim ates of h using the tabulated data presented here results in values of 0.42/day, 0.51/day and 0.75/day for the three successive deployments. Areal estim ates of h/< g > provide values of fractional removal of 0.66,0.65 and 0.83. Diacuaaion The in-situ time series of Chi, PAR and C e have been utilized to calculate estim ates of gross primary production (GPP) and net production (NP). The GPP estim ates have been compared to previous estim ates determined using different formulations [M arra et al., 1992; Waters e t al., 1994]. GPP calculated herein have also been compared to NP estim ates based on the amplitude of diel changes in the A c e signal. In addition, rates of gross productivity (< g >) and rates of POC accumulation (< pa >) and loss (< pi >), for both light and dark periods, have been estimated. These rates have in turn been used to form a budget which compares rates determined using two fundamentally different methods. Additionally, estim ates of phytoplankton respiration rate (r) and zooplankton grazing rate (h) were calculated. The latter estimation was made with the assumption th a t the phytoplanktonic system was operating at steady state. Additionally, no assumptions were made concerning gains/losses resulting from vertical mixing, vertical diffusion and horizontal advection, although it is quite apparent in the data set that these processes, especially the latter, are quite prominent. Indeed, a small num ber of rate estim ates were culled from the data set because they appeared to be affected by the passage of large scale advective events and their values were not representative of the local processes which were the focus of this investigation. The GPP estim ates made with the optical production model [Kiefer, 1993] indicate this region to be characterized by a spring bloom followed by the development of a subsurface chlorophyll maximum which appears as a subsurface GPP maximum (Fig. 3.4a). These features are also represented in the results obtained using the LDGO method (Fig. 3.4b) which were featured in a previous work [Marra et al., 1992]. Quantitative comparison of the yearly production values obtained using the two methods reveals good agreement between the estim ate obtained herein and the revised estimate obtained using the LDGO method (439 mg C/m2 /day vs. 449 mg C/m2 /day, [W aters et al., 1994]). A more detailed intercomparison reveals th at the Kiefer model predicts lower values near the surface and higher values below 35-40m. Both methods rely on the basic equation for absorption-based estim ates of gross production (Eq. 3.4). However, the only consistently applied param eter among the four term s in this relation is chlorophyll specific absorption (achi), for which values are based on water samples obtained during the four cruises necessary for deploying/recovering the mooring (Table 3.1). In addition to a different formulation for quantum yield, the LDGO technique applied daily integrated PAR and a dawn value of Chi from the moored time series instead of the full resolution time series. Thus, there are several potential sources for the observed dissimilarities. The inherent differences between various param eteri zations generally used when modeling photosynthesis have been examined previously [Jassby and P latt, 1976] and 61 the form used in the LDGO method [Steele, 1962] was shown to be consistently higher than the Michaelis-Menten form on which the present calculations are based. However, due to the values chosen for < ) > „ , th is attribute is reversed. Moreover, using the productivity model presented herein, a direct comparison between < | » / < | > m reveals higher values for light levels above ~ 10 Ein/m2 /day in the present model (Fig. 3.9). The difference between applying daily integrated and full resolution PAR on daily integrated production is minimal except for its effect on calculations of $. Thus, the difference afforded between applying a constant, dawn value of Chi and applying the full resolution time series must be the sole cause for the vertical distinction between the models. Stimulated fluorescence data have revealed th a t over a diel cycle, Chi concentrations increase during low light regimes and decrease during high light regimes. Indeed, Chi has been observed to vary by a factor of two or more over a diel cycle [Cullen et al., 1988; Hamilton e t al., 1990], This emphasizes the ability of these time series to resolve biological processes a t higher resolution than conventional ship-based techniques and stresses th e dichotomy involved in calibrating the former with the latter. The estim ates of deployment averaged NP based on A q provide values during the first deployment which are 25% higher than the GPP estim ates for the same period. This is an unexpected result (Eqs. 3.1, 3.3). Comparison between the statistical confidence of the GPP estimates with C e based findings (Table 3.2), indicate th at the error is contained within the NP results. The estim ates of areal NP for the second and third deployments are both less than the corresponding estimates of areal GPP. Additionally, they are essentially 62 o o o < £ > 0 E ■ s - ■ « . o CM o o 0 10 20 30 40 50 PAR (E in/m 2/doy) Figure 3.9. Comparison of the quantum yield ( < | > ) function used in the LDGO model (solid line) and the daily integrated values predicted by the Kiefer model (Eq. 3.5). Values for < | > have been normalized by the values of < |> m used in the respective formulations. 63 equivalent to each other, indicating that the ecosystem overall is operating a t steady state. These NP estim ates have relied on constant transform ations of SPM = 1020 • C c [mgC/m3 ] and POC = 0.25 SPM in order to develop an estim ate of carbon specific beam attenuation coefficient (c% ) [Siegel et al., 1989; Cullen et al., 1992a]. It has been previously documented th at both of these constants are subject to considerable variability. Comparison of simultaneous in-situ SPM and C c profiles in surface oligotrophic waters have provided a relation between the two which ranges from 830*1570 [mgC/m2 ] [Bishop, 1986]. Additionally, simultaneously obtained samples show th at the ratio of POC:SPM ranges from 0.12-0.70 but is generally constrained to a range of 0.20-0.30 [Hobson, 1967]. Direct measurements of c*c indicate th at the value used for these production estim ates is reasonable but th at the potential variability of this param eter is an issue which has yet to be resolved [Ackelson et al., 1993]. The moored data and the model generated phytoplankton carbon:chlorophyll ratio ( 6 k , Eq. 3.7) have been combined to generate estim ates of POCrSPM (6*, Eq. 3.10a) which range from 0.25-0.425 (Fig. 3.2). These have been used to isolate variability in the C c data to th at resulting from changes in POC. These reduced C c data (i.e., eg Eq. 3.9) have been used to estimate specific growth rate (p) during the light period and loss rates during the dark period (Eq. 3.8). In addition, the productivity model (Eqs. 3.4-3.7) has been used to estim ate gross production rates (Eq. 3.1 Id). The resulting rate estim ates are listed in Table 3.3. The estim ates for gross production rate 64 agree well with those reported in the literature [Kiefer and Mitchell, 1983] and range from 0.22/day-2.12/day over the light regimes within the euphotic zone represented on the mooring string. The rate budget is based on three independently estimated param eters (i.e., gross growth, specific growth and dark loss rates) and achieves reasonable statistical closure within the euphotic zone except for the third deployment. It was therefore deemed acceptable to proceed with making estimations of dark respiration based on p using an empirical relationship (Eq. 3.2) in order to derive estim ates of heterotrophic grazing rate (h). In order to accomplish this, it was assumed th at the local rate budget achieved closure (i.e., th at the planktonic ecosystem was operating a t steady state) and th a t discrepancies were due to advecting w ater masses to which these measurements (i.e., C c ) are highly sensitive. This resulted in estim ates of h which ranged from 0.09/day- 0.98/day. The upper lim it is derived from the third deployment at 43m and is an overestimate caused by the unreasonably low value of p. In addition, estim ates of the fraction of gross production consumed by the heterotrophic community have been made. These range from 0.24-0.87. These estim ates of h and h/< g > have been integrated over the upper 80m, resulting in water column grazing rates for the three deployments of 0.42/day, 0.51/day and 0.75/day and fractional grazing values with a range of0.65-0.83. These agree well with sim ilar estimates obtained during the SUPER experiment. In-situ incubations over the upper 80m were used to obtain an estim ate of areal grazing rate of 0.45/day while the areal fraction of grazing consumption was determined to be 0.7-0.8 [Frost, 1991]. 65 fih a n tw 4; S p ectral A nalysis o f th e M oored Tim e Series: T he S earch fo r L in e ar In tera ctio n M echanising Introduction to PhyaicaUBioloeiccd Interaction* A number of investigations into the mechanisms involved in interactions between physical and biological characteristics of an oceanographic system have been conducted over the past 25 years. Several early in-situ studies focused on simultaneously acquired tem perature and chlorophyll spatial data in the Gulf of St. Lawrence and the St. Lawrence river's maritime estuary. These data were obtained using a towed package and analyzed using fast fourier transform s (FFTs). Auto- and cross-spectra were created to investigate phytoplankton patch scales [Platt, 1972; Denman and P latt, 1975; Denman, 1976]. For a given depth, the shape of the tem perature and chlorophyll auto-spectra showed good correspondence, including a change in spectral slope occurring a t 101m. For length scales greater than this, coinciding tem perature and chlorophyll signals showed significant coherence, indicating this to be the characteristic transition scale between physically and biologically dominated variability within the local ecosystem [Denman and Platt, 1975; Denman, 1976]. A similar study in the Bay of Biscay discerned analogous evidence for such a transition length scale, this time ranging from 200-400 m. This was observed during a phytoplankton bloom period. A follow- up, post-bloom study in the same region, during a physical regime characterized by permanent stratification, revealed no indication of such a transition scale and lower coherence between the physical and biological signals [Fasham and Pugh, 1976]. The maximum horizontal wavelength associated with an internal gravity wave (IGW) field is C K 102-103 m) [Apel, 1987; de Angelis and Lee, 1994]. Thus, the observed transition between physically and biologically determined variability appears to coincide with a transition from mesoscale to higher frequency fluid motions. Analysis of the G ulf of St. Lawerence data also revealed th at the phase relationship between the two variables changed between depths and insignificant coherence was reported between depths for data obtained by towing two instrum ent packages spaced 4 meters apart in the vertical. These characteristics were attributed to a change in relation between the slopes of the vertical profiles of the two variables and the introduction of vertical oscillations within the fluid medium. That is, slopes with the same sense result in a positive phase relationship when subjected to vertical motions while slopes of opposing sense result in a negative phase relationship. A common oceanographic feature which results in such a reversal in profile relationships is the presence of a subsurface chlorophyll maximum within a stably stratified w ater column. This emphasizes the fact that oceanographic data are rarely ideal with respect to the requirements of analytical techniques and the presence of such a feature creates difficulty when interpreting the results [Star and Cullen, 1981; Cullen et al., 1983]. Denman concluded the Gulf of St. Lawerence studies by stating: "In all cases, the highly stratified nature of both chlorophyll and tem perature together with the ubiquitous nature of the internal waves made the unambiguous determination of spatial scales of horizontal patchiness impossible." [Denman and P latt, 1975]. 67 It is dear th a t the interaction of internal waves with phytoplankton biomass is typical of marine environments. However, the resulting impact of this interaction on biomass and rates of primary production needs further darification. Coherence between tem perature and chlorophyll within the IGW band ( / i g w » bounded by the inertial frequemy (ft) and the Brunt-Vfiisfilft frequency ( / b v )) varies and appears to be in some p art related to the strength of the IGW field. Theory predicts a peak ju st below / b v during periods of elevated IGW activity [Cairns, 1975]. A number of investigators have observed this peak in physical data sets and within the context of physical/biological interactions, it has been noted th a t elevated coherence between tem perature and chlorophyll within / ig w coinrides with the presence of a peak near / b v in the tem perature spectra [Fasham, 1978]. Increased coherence may be due to phytoplankton response to nutrient injection driven by wave breaking associated with the dissipation of the perturbed IGW field or increased amplitude vertical oscillations. The semi-diurnal tide is often the dominant component of / i g w . accounting for up to one third of the total variance within / ig w , depending on location [Gregg and Briscoe, 1979]. Vertical displacements of 15m due to internal tides have been observed within the upper 50m of the w ater column over the California continental shelf. In addition, internal wave driven vertical displacements of 10-20m for wave periods ranging from 12 hours (.08 cph) down to 10 m inutes (6 cph) have been observed [Denman and Gargett, 1983]. Vertical motions of this magnitude, within this range of frequencies, can significantly affect phytoplankton light history and their subsequent growth characteristics. This has been illustrated using a model to modulate the phase 68 relationship between the semi-diurnal tide and the solar cycle [Kamykowski, 1974]. Additional investigations indicated th at vertical oscillations facilitate the development of patchiness as phytoplankton are moved from the mixed layer through the shear region directly below. Introducing phytoplankton sinking into this scheme further enhanced the tendency toward patch formation [Kamykowski, 1976]. Phytoplankton sinking velocity is known to be dependent on in-situ environmental param eters including temperature, light level and nutrient concentration. Reduced sinking velocity corresponds to elevated nutrient concentrations and lower tem perature and light levels [Lande and Wood, 1987; Kamykowski et al., 1988]. The ability to adjust sinking velocity may be a survival trait designed to keep phytoplankton within the mixed layer and allow for repeated transitions between environments characterized by growth inhibiting light intensities or nutrient concentrations. Continual vertical cycling may allow phytoplankton cells to reach higher growth rates than could be attained in either environment separately [Lande and Wood, 1987]. A theoretical analysis, designed to investigate this question, found th at vertical fluctuations reduce primary production near the surface while enhancing it below a level termed the "crossover" depth. This has been estim ated to be between 10 and 30 meters and is principally dependent upon the coefficient for light attenuation [Holloway, 1984; Holloway and Denman, 1989]. Enhancement of primary productivity through this positive interaction between vertical oscillations and phytoplankton biomass is referred to as internal wave pumping and is considered to be a significant source of biological variability [Kahru, 1983]. The towed studies revealed the existence of a critical length scale which marked a transition between biologically and physically determined biological variability with physical processes showing greater influence as characteristic lengths tended toward the mesoscale [Denman and Platt, 1975; Denman, 1976; Fasham and Pugh, 1976]. In the Sargasso Sea, mesoscale phenomena are commonplace occurrences due to the proximity of their source, the Gulf Stream. Warm and cold-core rings may be spawned by the unstable meanderings of the Gulf Stream’ s path and the further interaction of one of these mesoscale eddies can result in the entrainm ent and transport of so-called warm outbreaks [Cornillon et al., 1986]. CZCS images have provided a means of observing surface pigment variability over entire ocean basins and it is evident th at mesoscale eddies are strong contributors to this variability. Some of the most interesting mesoscale features with regard to open ocean physical/biological interactions are warm-core rings. The physical characteristics of these phenomena consist of tem peratures of - 16 ® C and salinities of 36%c down to -400 m. Upwelling velocities within these anti cyclonic rings are around 1 m/day. The weak stratification within these rings results in additional vertical transport when surface wind stress becomes elevated. The upwelled waters transport nutrients into the euphotic zone and allow these eddies to sustain distinct ecosystems. Differences between species of phytoplankton, picoplankton and bacteria existing within these rings and those of the surrounding waters have been observed [Nelson et al., 1985; Nelson et al., 1989]. Further observations reveal th at the phytoplankton bloom which occurs within the rings occurs in the early part of the summer 70 (May/June) which contrasts the timing of the Sargasso Sea's spring bloom (March/April). Finally, total gross production within the eddies extrapolated over a six month period is estimated to be 126 gC/m3 which is nearly half the reported annual production for the continental slope region (280 gC/m2 ) [Hitchcock et al., 1985]. However, the eddy estimated production includes a bloom period which would overestimate annual production without the inclusion of additional samples from periods of lower production. Estimated annual gross production from the oligotrophic Sargasso Sea has been reported with a range of 120-160 gC/m2 /yr [Menzel and Ryther, 1960; Menzel and Ryther, 1961; P latt and Subba Rao, 1975; M arra et al., 1992; W aters et al., 1994]. A direct comparison between the periods for which data are available for the estim ates of ring production [Hitchcock et al., 1985] and the high resolution GPP estim ates made with the Biowatt tim e series (see Chapter 3) is provided in Table 4.1. A verage D a ily G ross P rim ary P ro d u ctio n (mgC/m2 /day) Tim e P erio d (JD) W arm -core R ing B iow att % D ifference 60-130 817 503 63.4 135-240 619 456 35.7 T able 4.1: C om parison o f B iow att p ro d u ctiv ity to estim ates from a w arm -core rin g These data emphasize the fact th at gross production within these features is distinctly higher than what is observed in the area's oligotrophic waters, although it has been noted th at the moored tim e series show extensive mesoscale interactions [Dickey et al., 1993]. The presence of autonomous 71 ecosystems w ithin distinct water masses leads to elevated coherence between physical and biological variables a t mesoscale frequencies. Indeed, when considered in cory unction with results from in-situ studies and modeling investigations, CZCS images emphasize the range of spatial and temporal scales and the myriad processes involved in determining rates and distributions of oceanic primary production. Partitioning individual contributions to biological variability among the potential interaction mechanisms has only recently been achievable using a combination of emerging sampling techniques [McClain et al., 1990]. The unique feature of the Biowatt experiment was the initial acquisition of simultaneous, high resolution, extended physical and bio-optical time series within the photic zone [Dickey, 1991]. This allowed for sampling a variety of phenomena w ith an extensive range of characteristic time scales, encompassing mesoscale 0(2-10 days) through IGW 0(10°-103 m in) signatures which appear in the physical data. Biological processes corresponding to these time scales include episodic blooms and the oscillation, via internal waves, of a vertical gradient in the bio-optical characteristics. Application of FFTs to this data set have previously revealed characteristic frequencies associated with physical (e.g., semidiurnal tide and inertial response to synoptic forcing) and biological (e.g., diel photosynthesis) phenomena sampled by the MVMS packages [Dickey et al., 1991]. The section immediately following this introduction describes the analytical methods and the presentation of the results. The analyses which follow include comparison, over depth and deployment, of the magnitude of the characteristic signals just mentioned. This is followed by the presentation of contour maps of normalized spectral variance and coherence between param eters. These maps reveal spatial (i.e., vertical) or temporal variations which are analyzed with the intent of discerning characteristics indicative of physical/biological interactions. 73 Analytical methods a n d presentation technique Fourier analysis of the Biowatt time series data was performed in order to investigate variance levels over a wide frequency spectrum. FFTs were performed on data ensembles of 8192 points (i.e., 23.8 days). Each ensemble overlapped its predecessor by 75%. In other words, there is a 5.7 day displacement between the mid-point of adjacent ensembles. Before being passed to the FFT algorithm, the ensemble's m ean was subtracted and the resulting tim e series was detrended (with a second order polynomial) and cosine tapered to insure periodicity. Equation 4.1 provides the analytical definition of Transformed data are used to create one-sided auto-spectral density functions using the following relation; an FFT. oo 4.1) tX*(/)X(/)] 4.2) where * denotes the complex conjugate, N„ is the number of data ensembles and Nd is the number of data points within an ensemble. It is understood th at / has been discretized over k values following equation 4.3. For each deployment, a number of autospectra were formed (Table 4.2). Investigations into phenomena with time scales ranging from 11 days down to 8 minutes are possible. These temporal lim its are a result of the Nyquist sampling theorem as it applies to the ensemble size and data resolution [Bendat and Piersol, 1986]. Thus, mesoscale activity seen in the current and tem perature data may be resolved. Episodic biological events are also revealed within the mesoscale band. Due to the resolution of the tim e series, / i g w i s not completely represented, since the high frequency cutoff of the power spectra is 7.5 cycles per hour (cph). This is the same order as the minimum estim ated value of / b v for the experiment which occasionally occurs in the first deployment but is generally up to an order of magnitude below the estim ated values (Table 4.3). Thus, these power spectra do not normally provide information about highest frequency IGW modes and do not enter the realm of turbulence. Several representations have been created with the power spectra presented here. The commonly used representation (Figs. 4.1-4.5) consists of the average of all ensembles from a deployment for a given variable. Two depths are presented for each deployment. Generally, these are 23 and 101 m eters when the data are available. These have been used to illustrate the dominant characteristic frequencies present in each signal. In addition, comparisons between the magnitude of these characteristic frequencies may be made between depths and deployments. Finally, comparisons between the 75 D uration Ensem bles Deploy 1 14 m 70 9 23 m 70 9 43 m 70 9 62 m 70 9 81 m 70 9 101 m 70 9 120 m ND ND 160 m 70 9 Deploy 2 14 m 105 15 23 m 105 15 43 m 78 10 62 m 105 15 81m 105 15 101 m 105 15 120 m 105 15 160 m 105 15 Deploy 3 14 m 72 9 23 m 80 11 43 m 80 11 62 m 80 11 81m 80 11 101m 80 11 120 m 80 11 160 m 80 11 Table 4.2 a: Currents. D uration Ensem bles Deploy 1 14 m 70 9 23 m 70 9 43m 70 9 62 m 70 9 81m 70 9 101m 70 9 120 m ND ND 160 m 70 9 Deploy 2 14 m 105 15 23 m 105 15 43m 78 10 62 m 105 15 81 m 105 15 101 m 105 15 120 m 105 15 160 m 105 15 Deploy 3 14 m 72 9 23 m 80 11 43 m 80 11 62 m 80 11 81 m 80 11 101m 80 11 120 m 80 11 160 m 80 11 Table 4.2 b: Temperature. -a o > D uration Ensem bles Deploy 1 14 m 70 9 23 m 70 9 43 m ND ND 62 m ND ND 8 1m ND ND 101m 70 9 120 m ND ND 160 m 27.3 ND Deploy 2 14 m 105 15 23 m 105 15 43 m ND ND 62 m 105 15 81 m 58 7 101 m ND ND 120 m 105 15 160 m 82 ?? Deploy 3 14 m ND ND 23 m 80 11 43 m 29 ND 62 m ND ND 8 1m 70 9 101m 80 11 120 m ND ND 160 m ND ND Table 4.2 c: Dissolved Oxygen. D uration Ensem bles Deploy 1 14 m 39 3 23 m 70 9 43 m 52 6 62 m 42 4 81 m ND ND 101m 70 9 120 m ND ND 160 m 70 9 Deploy 2 14 m 105 ND 23 m 105 13 43 m 78 6 62 m 105 15 81 m 105 15 101 m 105 15 120 m 105 15 160 m 105 15 Deploy 3 14 m 72 9 23 m ND ND 43 m 80 11 62 m 80 11 81 m 80 10 101m 80 11 120 m ND ND 160 m 80 11 D uration Ensem bles Deploy 1 14 m ND ND 23 m 70 9 43 m ND ND 62 m 70 9 81 m 70 9 101m 70 9 120 m ND ND 160 m 70 9 Deploy 2 14 m 100 14 23 m 104 15 43 m ND ND 62 m 105 15 81 m 105 15 101m 76 10 120 m 105 15 160 m ND ND Deploy 3 14 m 72 9 23 m 58 8 43 m 72 9 62 m 60 8 81 m 75 10 101m 66 8 120 m ND ND 160 m 61 7 Table 4.2 d. Cc. Table 4.2 e. Chi. Table 4.2. Data available for spectral analysis. The duration of the time series used in the FFTs and the number of ensembles obtained. Each ensemble consisted of 8192 points (23.8 days). Each succeeding ensemble was shifted by 25% resulting in a 5.7 day difference between ensemble midpoints. The fourier transforms of these were then used to form the contour maps of G n and y. ND indicates no data. The maximum possible ensembles for the three deployments are 9, 15 and 11 respectively. 77 J u lia n D a y 14 m 23 m 43 m 62 m 8 1 m 101m 120 m 160 m 70.0 13.3 13.6 12.1 12.0 14.2 13.9 12.3 75.7 5.5 6.8 6.8 7.7 10.2 10.0 9.4 81.4 13.4 11.9 8.1 7.8 7.5 7.2 7.0 87.1 19.0 17.0 9.3 8.2 10.2 10.2 10.5 92.8 22.8 20.0 12.8 14.2 14.2 12.9 11.6 98.4 22.7 23.4 24.3 21.3 17.1 15.1 12.9 104.1 30.8 28.1 24.6 24.4 18.8 15.4 13.7 109.8 54.1 46.4 20.1 14.4 12.3 11.3 11.7 115.5 57.3 49.9 24.7 18.4 15.7 13.1 11.8 121.2 30.8 32.3 32.1 24.9 18.6 15.8 14.2 145.0 54.3 52.5 41.5 39.6 40.2 27.1 17.5 17.7 150.7 38.8 44.8 46.8 42.4 41.9 28.2 17.4 16.4 156.4 53.8 55.0 49.5 43.7 41.7 27.8 17.3 17.5 162.1 36.3 48.5 57.8 47.1 41.5 27.0 15.5 15.7 167.8 36.4 47.6 56.8 48.0 43.0 28.7 17.6 17.2 173.4 44.4 53.8 59.8 49.1 43.9 29.1 17.6 17.3 179.1 58.6 62.0 60.8 52.0 46.2 31.2 19.1 17.1 184.8 70.0 69.7 61.4 52.7 47.0 31.7 19.5 17.9 190.5 89.7 83.5 61.2 52.2 46.7 32.5 21.5 19.2 196.2 82.1 80.2 66.9 54.8 48.1 34.2 23.7 22.1 201.9 18.0 33.4 52.0 54.4 51.9 44.9 41.8 38.2 207.6 45.9 52.0 61.0 62.0 51.4 33.1 31.0 25.2 213.3 33.0 41.9 54.3 58.5 49.3 30.0 29.6 23.1 219.0 49.2 50.5 51.7 53.3 41.4 22.6 26.2 22.0 224.6 23.8 32.8 44.7 50.0 40.7 21.9 26.3 22.8 230.3 20.6 32.6 47.2 50.0 40.8 25.1 27.8 24.7 236.0 22.0 36.1 53.0 53.9 44.5 29.7 30.3 27.4 255.0 47.1 44.6 56.4 60.4 46.2 42.6 37.6 23.4 260.7 47.8 50.3 62.2 60.5 46.0 39.5 33.3 14.8 266.4 46.0 47.0 58.2 59.2 47.5 42.6 36.3 19.1 272.1 44.6 39.5 50.0 59.0 48.8 45.5 39.0 23.1 277.8 43.1 43.3 44.0 51.1 52.5 48.0 43.0 28.8 283.4 45.0 42.0 29.2 40.4 52.9 50.5 44.3 35.9 289.1 42.3 39.3 26.2 38.5 49.7 47.9 42.9 30.7 298.4 41.7 40.1 33.9 45.5 51.0 44.4 38.5 23.2 300.5 40.5 38.8 32.1 46.4 51.6 42.7 36.8 21.4 306.2 40.6 37.5 23.3 30.8 47.2 50.0 42.2 29.3 311.9 39.9 36.7 22.2 38.0 50.1 46.1 39.4 26.9 317.6 36.6 33.6 19.3 19.2 33.2 44.6 44.5 32.6 Table 4.3. Brunt-Vaisala frequency estimated from the in-situ temperature and current time series. The unreported values at 120m during the first deployment correspondto the failed MVMS package. Units are in cph. 78 I04 X 1000 5 ,00 < 1 0 E w 1 0 .1 0.01 I04 x io o o I , 0 0 < 1 0 E £ I 0 . 1 0 .0 1 DEPLOY I 101m < / > n o _ -r - 'N \ - p—951fcU«|-| V > t------------ r - < / > DEPLOY 1 - | 23m 2 TOTAL*------ ' cw«------- ccw«------ ■ V - - 1 1 1 — DEPLOY 2 23 m —1" - o - T------------ 1 ---------- « / > DEPLOY 2 | 1 0 1 m r\) V V ■ V - :-----9 9 % L » v f|- . . . 1 i DEPLOY 3 23 m x 1000 C L P 100 E u 0 .1 0 .0 1 0 .0 1 0 .1 DEPLOY 3 1 0 1 m V.V 0 .0 1 0.1 CPH CPH Figure 4.1. Deployment averaged rotary spectra for currents at 23 and 101 meters. A number of 23.8 day ensembles are incorporated within each deployment average (Table 4.2). Diurnal (D), inertial (I) and semi- dirunal (SD) frequencies are marked. The 95% confidence levels are shown. These rotary spectra are partitioned into clockwise (dashed) and counter-clockwise (dashed-dot) components. The total spectra (solid) is also shown. D EPLO Y I 23 m DEPLOY I 1 0 1 m N £ 0.1 O 0 01 ? I0_s I0 "4 - 10-5 : X 0. o c^ 1 0 I 0 .1 0 .0 1 10-5 I0-4 10-5 1 ------ D EPL O Y 2 DEPLOY 2 1 0 1 m DEPLO Y 3 23 m DEPLOY 3 1 0 1 m £ 0.1 u 0.01 I0"5 0 .0 1 0 . 1 1.0 0 .01 C PH C PH Figure 4.2. Deployment averaged spectra for temperature at 23 and 101 meters. A number of 23.8 day ensembles are incorporated within each deployment average (Table 4.2). Diurnal (D), inertial (I) and semi- dirunal (SD) frequencies are marked. The 95% confidence levels are shown. 80 D EPLO Y I 23 m DEPLOY I 1 0 1 m < / > I M X Q . O I 0 .1 0. 01 I0- 3 to-4 10-5 I0- 6 I0-7 10-8 ------ 1------- T ------------ 1---------- o « D EPLO Y 2 ;_ _ 5 23m ; ^ “ V IN) r-r95%L»v»|- T i DEPLOY 3 62m 9 4 7 . L « * tl DEPLOY 2 1 0 1 m 9 5 7 . L«y«< 1 o - 1 1 g DEPLOY 3 ' - 101 m w ; ^ ----- 9 5 7 .L « # I- ” i I i 0 .0 1 0 .1 CPH 1.0 Figure 4.3. Deployment averaged spectra for Cc at 23 and 101 meters (where available). A number of 23.8 day ensembles are incorporated within each deployment average (Table 4.2). Diurnal (D), inertial (I) and semi-dirunal (SD) frequencies are marked. The 95% confidence levels are shown. 81 DEPLOY I 1 0 1 m DEPLOY I 23 m to ro > O .O I 10-3 q < X o. o CM o> X a o C M I 0 .1 0 .0 1 10-3 1 0 -* lO -9 I 0 .1 0 .0 1 10-3 10- * to-5 D EPLO Y 2 14m DEPLOY 2 tOlm C O a to 95% Level - 1 o 1 1 w D EPLOY 3 - | 23 m V jo - r v , . - — - 9 5 % L tv tl- l .1_____________i _ 0 .0 1 0 .1 CPH 1 .0 DEPLOY 3 1 0 1 m 95% Ltvtl Figure 4.4. Deployment averaged spectra for Chi at 23 and 101 meters (where available). A number of 23.8 day ensembles are incorporated within each deployment average (Table 4.2). Diurnal (D), inertial (I) and semi-dirunal (SD) frequencies are marked. The 95% confidence levels are shown. 82 T ------------ r g DEPLOY I £ 23m D EPLO Y I £ 101m O .O I 0 .1 CPH 1.0 0.01 0.1 CPH 1 .0 DEPLO Y 2 23m D EPLO Y 2 1 2 0 m u i n 1000 100 X Q . o 5 5 * . ,4 DEPLOY 3 23 m u > 1000 1 0 0 0 .1 0 .0 1 10-3 — Wftlmi Figure 4.5. Deployment averaged spectra for DO at 23 and 101 meters (where available). A number of 23.8 day ensembles are incorporated within each deployment average (Table 4.2). Diurnal (D), inertial (I) and semi-dirunal (SD) frequencies are marked. The 95% confidence levels are shown. 83 dominant frequencies characteristic of the physical versus the bio-optical signals may be made. The spectra for the horizontal currents (Fig. 4.1) show three curves instead of the single curve normally associated w ith the power spectra of scalar quantities. These power spectra combine the amplitude estim ates from the meridional and zonal velocity components in order to generate rotary spectra [Gonella, 1972]. The three curves represent the total, clockwise (CW) and counterclockwise (CCW) components of the power spectra. The 95% confidence intervals are shown here in order to provide a context for the expected range in values of spectral amplitude. This range is small since the spectra were averaged over all ensembles for a deployment. There is a minimum of nine ensembles for a complete time series within a given deployment (Table 4.2). These FFT results are also presented in contour maps consisting of normalized spectral density (Gn if, Z or t)) as a function of frequency and depth or time. Normalization consists of dividing each individual autospectra within a contour map by the average spectrum for all those represented within the map. This normalization process may be described using equation 4.4; 4- 4> where Gn if) is the auto-spectral density function (see eq. 4.2) for a given ensemble and <Gx* (/)> is defined as; 84 Creating maps of normalized spectral density was motivated by the desire to identify spatial (i.e., vertical) or temporal maxima and minima within frequency bands. In order to create a linear scale, values along the frequency axis are transformed logarithmically. This transformation is summarized in Table 4.4. Gn (/, Z or t) maps have values which range around 1.0 (i.e., the average value of the map a t th at frequency). Values greater than 1.0 correspond to events with above average spectral energy (i.e., variance) within a given frequency band. Thus, over a three decade range in time scales, temporal and spatial (i.e., vertical) distribution of spectral density has been determined for a number of physical and bio-optical signals. These m ay be used to define the temporal and spatial boundaries of processes associated with the physical or biological systems and mechanisms associated with interactions between these systems. It should be emphasized that this normalization removes the "red” spectral behavior normally associated with oceanographic data. In addition, spectral peaks indicative of diurnal and semi-diurnal tidal activity, inertial oscillations and diel photosynthetic cycles, which appear in the standard power spectra, are not expected to appear in the Gn (f) maps. Since variance at each frequency is normalized by the average variance a t each frequency, information regarding differences in variance between adjacent frequency bands is lost. 85 F req u en cy A xis F req u en cy F req u en cy Period (1/d) (1/h) (hr) -2.44 0.09 0.0037 272.90 -2.34 0.11 0.0046 216.77 -2.24 0.14 0.0058 172.19 -2.14 0.18 0.0073 136.77 -2.04 0.22 0.0092 108.64 -1.94 0.28 0.0116 86.30 -1.84 0.35 0.0146 68.55 -1.74 0.44 0.0184 54.45 -1.64 0.55 0.0231 43.25 -1.54 0.70 0.0291 34.36 -1.44 0.88 0.0366 27.29 -1.34 1.11 0.0461 21.68 -1.24 1.39 0.0581 17.22 -1.14 1.75 0.0731 13.68 -1.04 2.21 0.0920 10.86 -0.94 2.78 0.1159 8.63 -0.84 3.50 0.1459 6.85 -0.74 4.41 0.1837 5.45 -0.64 5.55 0.2312 4.33 -0.54 6.99 0.2911 3.44 -0.44 8.79 0.3664 2.73 -0.34 11.07 0.4613 2.17 -0.24 13.94 0.5808 1.72 -0.14 17.55 0.7311 1.37 -0.04 22.09 0.9204 1.09 0.06 27.81 1.1588 0.86 0.16 35.01 1.4588 0.69 0.26 44.08 1.8365 0.54 0.36 55.49 2.3121 0.43 0.46 69.86 2.9107 0.34 0.56 87.95 3.6644 0.27 Diurnal Inertial Semi-Diurnal Table 4.4. Transformation between frequency in standard units and the frequency axis used in the contour maps of Gn and y. The corresponding period is also provided. The specific transformations for the diurnal, inertial and semi-diurnal frequencies are also given. -1.38 1.00 0.0417 24.00 (/d) -1.33 1.12 0.0466 21.46 (/i) -1.09 1.93 0.0805 12.42 (fad) 86 In order to compare spectra over depth or time, WKB scaling has been applied to / i g w in the Gn (f) maps. This is designed to compensate for the natural amplification of current and vertical displacement which takes place when waves travel through a non-uniform media [Eriksen, 1988]. In practice, this consists of scaling Gn ( / i g w ), by the value of / b v associated with its data ensemble. Thus, current spectra use the factor / B V < / / B v ( z ,t ) where / b v 0 is a reference frequency taken as 1 cph. Temperature, bio-optical and dissolved oxygen spectra are scaled by the inverse of this ratio. /Bv(z,t) estim ates based on the mooring data are crude due to the 10-20 m eter vertical displacement between MVMS packages. In-situ tem perature time series and a constant value of salinity (36.5%o, taken from in-situ profiles) were used with the UNESCO equation of state to create density time series [Millero and Poisson, 1981]. These in turn are used for calculating density gradients. /Bv(z,t) was estimated with the following equation; where g=9.81 m/s2, Ap=pL-pu, Az=zl-zu, p*=(pL+pu)/2.0 and z*=(zl+zuV2.0 and the subscripts L and U denote the lower and upper MVMS packages, respectively. These values of / b v are for depths corresponding to the average depth of the two time series used to perform the calculation described in eq. 4.5. / B v ( z , t ) values for depths corresponding to those of the MVMS packages are listed in Table 4.3. These were generated by interpolating between the depths 4.5) 87 of the /Bv(z*,t) time series and taking six day averages around the temporal midpoint of each ensemble used to create the power spectra. Cross-spectral density functions (G,y (/)) have been formed from the fourier transformed time series using equation 4.6 in order to investigate the strength of linear inter-relationships between two time series. These functions were used to determine coherence between two records using equation 4.7. Contour maps of coherence as a function of frequency and depth have been created for each deployment and for the whole experiment by averaging all available ensembles a t each depth. The number of available ensembles for a given param eter at each depth and deployment is reported in Tables A4.1 a-e. Contour maps of coherence as a function of tim e and frequency have been created using data from two in-situ param eters at a given depth. Although the actual significance level depends on the amount of usable data, the significant coherence threshold for the frequency/depth maps is a t least 0.4 while for the frequency/time maps it is 0.6. Details concerning these coherence thresholds are contained in Appendix 4.1. [X*Cf)Y(/)]. 4.6) YxyCf) = G x x ( f ) G y y ( f ) 4.7) 88 Deploym ent averaged autospectra Horizontal Currents For all depths and deployments, the dominant spectral peaks in the current spectra correspond to the inertial ( / i ) and semidiurnal ( / s d ) frequencies (Fig. 4.1). The spectral magnitudes of f \ and / s d for a given depth are essentially constant between deployments and decreasing with depth for a given deployment The drop is not large a t / s d b u t is a half decade at f\. As expected in the northern hemisphere, both peaks have a clockwise orientation. The first deployment is characterized by the greatest mesoscale energies of the experiment which overwhelm the characteristic peaks a t / i and / s d (Fig- 4.1 a,b). The other two deployments have mesoscale energies which are an order of magnitude lower than / i (Figs. 4.1 c-f). Temperature The first deployment’ s 23m tem perature spectrum shows mesoscale energy ju st below /d to be a t least a half decade higher than all other spectra presented (Fig. 4.2). The second deployment spectra show depth consistent energy a t the lowest mesoscale frequencies represented (Fig. 4.2 c, d). The third deployment's 101m spectra reveals elevated energy levels at the lowest mesoscale frequencies which do not appear at 23m, possibly due to the passage of advective events with weak surface expressions. The / i and / s d peaks in the tem perature spectra are less prominent and occur only at 101m during the second and third deployments where stratification is present (Figs. 89 4.2 d,f). These two deep spectra also show values of G W /igw) a decade higher than those seen in the first deployment's 101m spectrum (Fig. 4.2b), again due to the reduced stratification observed during the first deployment. niftsnlued Oxygen (DO) The upper DO spectra reveal no pronounced peaks in the mesoscale band. However, the second deployment spectrum shows Gx x (/m) a t the lowest frequencies to be a full decade greater th an that observed in the surrounding deployments (Figs. 4.3 a, c, e). The deep spectra all show G ^ / m) which are elevated in comparison to their near-surface counterparts (Figs. 4.3 b, d, f). The dominant feature in all of the 23m spectra is G «(/d)- This m ust be due to photosynthetic activity since the corresponding tem perature spectra do not show a similar feature which eliminates diurnal heating as a potential source of variability. The diumal/diel peak is not present in the deep spectra due to minimal photosynthetic activity at depth. The second and third deployment near-surface spectra also reveal prominent peaks a t / s d caused by tidally driven vertical oscillations. This feature does not appear during the first deployment, due to a reduced vertical gradient in DO characteristic of recent w ater column ventilation. The peaks a t / s d in the deep spectra of the second and third deployments are present but do not contain as much energy. G x x ( / i g w ) at 23m during the second deployment is at least one decade greater than during the other two deployments. The deep spectra from the first deployment is consistent with the near-surface spectra. This corresponds well with the previously noted characteristics of the tem perature spectra and 90 is consistent with the presence of minimal stratification through much of this deployment and interm ittent periods of water column ventilation. G u ( / i g w ) at 101m in the second deployment is greater than th at observed a t 23m, consistent with the presence of a surface mixed layer. This pattern of depth variation in G u ( / ig w ) i s also shown in the third deployment spectra. Bio-optical spectra (c- and Chi) The characteristics of the C c and Chi spectra (Figs. 4.4,4.5) will be described concurrently and referred to as the bio-optical spectra. Their spectral characteristics are sim ilar enough th at this may be done without loss of information and will reduce repetition. All of the upper bio-optical spectra show mesoscale signatures and shape comparisons between the two upper bio- optical spectra for a given deployment show reasonable correspondence. The first deployment comparison between mesoscale shapes is not strong but the peak observed in the C c spectrum is within the frequency range of the hump in the Chi spectrum (Figs. 4.4a and 4.5a). The second deployment spectra both exhibit spectral humps (Figs. 4.4c and 4.5c) while the third deployment spectra show distinct peaks at the same frequency (Figs. 4.4e and 4.5e). The only significant mesoscale band features in the deep bio-optical spectra occur during the first deployment (Figs. 4.4b and 4.5b). The hump in these spectra is a t a somewhat higher frequency than th at observed at 23m. All of the upper bio-optical spectra show distinct peaks a t / d > corresponding to photosynthetic activity. This peak is not as prominent in the 101m spectra. The deep spectra have more distinct peaks a t / s d and at /i, which was difficult to discern in the upper spectra due to the dominance of /d- 91 Chronology o f Observed Phxsical/Bioloeical Interactions The maps of G n (/, t) show a number of instances when elevated variance in one variable coincides with th a t of another (see also Appendix 4.2). These are used to provide direction in investigating inter-relationships between variables when examining coherence m aps created from cross-spectra. A variety of physical/biological interactions are represented in these time series. One of these consists of the horizontal advection of w ater masses and their associated ecosystems due to mesoscale advective activity. Due to this experiment’ s proximity to the path of the Gulf Stream, this mechanism prevails for a significant portion of the experiment and is a strong contributor to ecosystem variability [Hitchcock et al., 1985; Nelson et al., 1989]. Additionally, interactions based on water column stratification directly controlling nutrient concentration and both the instantaneous and daily- integrated light fields available to phytoplankton populations, via modulation of their vertical excursion length, are seen [Lande and Wood, 1987]. Finally, water column stratification, through establishm ent of the physical environment necessary for support of an internal wave field, provides an indirect means of modulating the nutrient and light regimes experienced by the autotrophic population [Denman and Gargett, 1983; Kahru, 1983; Holloway and Denman, 1989]. The following presentation of observed interactions is partitioned into three distinct periods which cover the first two deployments and summarizes more extensive observations presented in Appendix 4.2. 92 D g n lo x m z n L l Physical/biological interactions due to isolated ecosystems within specific w ater masses are present throughout the experiment but dominate the first deployment. Elevated advective activity is well-documented by the time series of current velocity vectors and the synoptic maps (Figs. 2.2 and 2.8). The average current speed dining this deployment is 41 cm/s at 81m (i.e., the midpoint of the positions occupied by the mooring’ s MVMS packages). This contrasts the corresponding average speeds from the second and third deployments which are 16 and 26 cm/s [Dickey et al., 1990]. At all depths of the entire first deployment, currents show mesoscale variance up to three times greater than later deployments and elevated variance extending over the entire IGW band (Fig. 4.6). This is well-supported by the first deployment power spectra for which the mesoscale band is elevated to the point of masking the inertial and semi-diurnal peaks (Fig. 4.1 a, b). The tem perature field does not em ulate this mesoscale activity below 40m and reveals minimal variance in the IGW band until the appearance of stratification (Figs. 2.15,4.7). This deployment may be divided into three periods based on the advective field. The first period (JD 60-78) is characterized by high current velocities which are predominantly meridional and the episodic passage of stratified mesoscale structures which appear as a succession of warming events (Figs. 2.2 and 2.3). The attributes of the second period (JD 78-90) consist of uninterrupted deep convection through JD 87 (i.e., 18s mode waters) followed by the onset of permanent stratification and extremely low current velocities. Concomitant with the development of stratification is a sharp 93 Julian Da; 19 87 e o e e a o o Julian Day 1 9 8 7 S ! Julian Day 1 9 8 7 S 8 0 > J u lia n D a y 198? l MMMcato B »ad I m Lot Frequency 1 Q W 8 s • 3 Figure 4.6. GN(f,t) for current speed. Figures a-f, from the indicated depths, cover the entire experiment, to Figures g (20m) and h (100m) emphasize dampening of high frequency motions by developing stratification. O l J u lia n D a y 1 9 6 7 J u lia n D a y 1M7 14 o . i i 0.4 o .a 0 B ra n t 100 too B ran t'V a 62 m I I 1 4.7 c 0 100 B ra n t J u lia n D iy 1987 O 100 - * • * ■ - * 0 - * o 100 B runt-V aisala Fi«q. (cph) L g f Fi»m » u cy B nm t-V aiaal* F n q . (cph) Figure 4.7. GN(f,t) for temperature with associated BV time series. Figures a-f, for the indicated depths, cover <0 the entire experiment. Figures g (10m) and h (20m) indicate development of IGW activity during springtime. increase in the C c and Chi signals which last about two days (Figs. 2.5 and 2.6). The third period (JD 90-130) shows the highest current speeds of the experiment, a strong directional veering due to interaction with a cold-core ring (Fig. 2.8 c-f) and the onset of perm anent stratification (Figs. 2.3 and 2.15). It will also be observed th at mesoscale coherence between bio-optical variables decreases during energetic horizontal advection while mesoscale coherence between the physical and bio-optical signals becomes significant during these periods. Period 1: (JD 60-78) Extremely high mesoscale variance in the velocity field appears throughout the water column during this period. Elevated variance also extends into / i g w (Fig. 4.6). This contrasts the tem perature field which only shows elevated mesoscale variance above 40m due to the episodic passage of stratified waters and minimal values in the IGW band (Fig. 4.7). The particle (C c ) field a t 101 and 160 m eters reproduces the characteristics of the velocity field where broad band elevated variance is seen (Figs. 4.8 d, e). This contrasts the pigment (Chi) field which contains elevated mesoscale variance only a t 23m and shows minimal variance in the IGW field at this tim e (Fig. 4.9a). Coherence between the individual current directions and temperature generally increases with depth below 60m (Figs. 4.10 a,b). However, coherence maps for zonal currents and tem perature indicate a mesoscale feature with depth decreasing coherence which has significant values for a short period (JD 70-80) in the near-surface maps (Figs. 4.10 c, d). Significant mesoscale 98 Julian D a y 1 9 8 7 Julian D a y 1987 IG W B an d M esoscale B and SD 2 3 m g 1 1 90 .0 4 .8 a 0 .8 Log F req u en c y IG W B an d M esoscale B and SD 500 . 0 6 1 m 870. 0 * .5 4.8 b 0 .4 Log F req u en cy 99 Julian D a y 1 9 8 7 Julian D a y 1987 IGW B an d M esoscale B and SD 81 m 2 7 0 . 0 O 2 4 0 .0 1 8 0 . 0 1 50 . 0 4.8 c Log F req u e n c y MeBOBcale B and IG W B an d SD 5 0 0 .0 101 m 2 7 0 .0 1 5 0 . 0 0 .6 120.0 4.8 d 0 .4 Log F req u en c y 100 IG W Band M eaoscale B a n d SD 160 m 270 .0 840 .0 21 0.0 8 •S- ISO .0 4.8 e L o g F re q u e n c y fG W Band SD 1 2 0 .0 101 m O) I 80.0 .2 * ° 4 .8 f L o g F re q u e n c y Figure 4.8. GN(f,t) for C c. Figures a-e present results of the available spectral analyses. Thus, except for the 100m (d) and 160m (e) maps, the time period represented does not cover the entire experiment. Figure f represents the first deployment subsection of the 100m map and is used to illustrate the response of the particle field to the development of permanent stratification. 101 Ju lian D a y 1987 IG W B an d M eaoscale B and SD SO 23 m 80 1 0 00 90 O.s 90 4.9 a 70 L og Fre q ue ncy M esoscale B and S D IG W B and 61 m o .0 0 •fi.O. 4 .9 b Log F req u en cy 102 1QW B an d SD 00 .0 81 m 7 0 .0 1 0.0 eo.o > e .o . SO .0 ao .o 90 .0 4 .9 c Log F re q ue ncy IG W B and SD 10.0 101 m eo.o so .0 ao .o 90 .0 4 .9 d Log F re q u e n c y Figure 4.9. GN(f,t) for Chi. Figure a (23m) reveals coincidence of high frequency variability with permanent stratification. Figures b-d (62, 81 and 101 meters) indicate link between internal waves and episodic blooms. 103 D epth (ml D epth (ml IQ W B and M esoacale B an d SD D e p l 85 .0 4.10 a 1 00 .0 L og Frequency IQ W B and Meeoacale B an d S D D e p l 85 .0 50 .0 4.10 b 0 .6 Log F req u en cy 104 Ju lian D a y 1 9 6 7 Ju lian D a y 1987 IQW B and M eaoacale B an d SD 13 m o o > » 0 0 0 4.10 c 0 L og F requency M esoecale B an d IG W B and SD 23 m 4.10 d 0 Log F req u en cy 105 IQ W B an d Meaoacale B a n d S D 160 10.0 00 .0 90 .0 80 .0 4.10 e 70 .0 L og F requency 1G W B a n d S D 1 6 0 m 10.0 00 .0 90 .0 90 .0 4.10 f 70 .0 L o g F requency Figure 4.10. 7(f) between temperature and individual current directions. Maps a and b show 7<f,Z) during the first deployment for zonal and meridional directions respectively. Maps c-e show y(f,t) with zonal current at 14, 23 and 160 meters during the first deployment. Map f shows 7(f,t) with meridional current at 160m during the first deployment. 106 coherence occurs between tem perature and DO a t 23 and 101 m eters (Fig. 4.11 c, d), both bio-optical variables and DO at 23m (not shown) and tem perature and Chi at 101m (Fig. 4.12a). The strongest coherence occurs between C e and Chi and appears consistently at all depths, extending into th e IGW band (Figs. 4.13 a-c). There are several other instances of significant coherence between the physical and bio-optical variables. These are not shown due to these figure’ s limited utility, but indude zonal current and Chi a t 23m, tem perature and C e a t 23m, and tem perature and Chi a t 160m. These coherence maps indude a number of manifestations indicating the passage of a distinct w ater mass and ecosystem. The current direction a t this tim e is essentially meridional (directed toward the south), maintains speeds of at least 60 cm/s for about half of this time period and attains speeds greater than 1 m /s for an entire day (JD 69) near the surface. The onset of the most prominent tem perature excursion (JD 69-72) occurs coincidentally w ith this period of maximum current velocity and shows good temporal correspondence w ith the high coherence features ju st summarized. Period 2: (JD 78-90) Although the entire deployment is characterized by mesoscale variance above the experimental norm, this period exhibits values which are lower th an the remainder of the deployment in the current field (Figs. 4.6 g, h). This also holds in the mesoscale variance of the tem perature field where values decrease below the experimental average at this time (Fig. 4.7g). This decrease in mesoscale variance in the physical fields coincides with increased mesoscale 107 ^ p I ^SD T i i Log Frequency 110 •0 • 0. Log Frequency o 0 0 J u lia n D a y 1 9 8 7 J u lia n D a y 1987 Log Frequency IQW Bend Meeoocale Bend ^ D I ^ j ,0 .0 .0 .0 .0 .0 '0.0 Log Frequency Jo li* n D ty l9 S 7 Log FnqufBcy Log FnqiwDty •© , i M wc m h I* Band D I ^ s n •o.« -o.* Log Frequency 0.0 o.« ^ Figure 4.11. 7(f,t) between temperature and DOX. Map a (23m) covers the entire experiment. Maps b-d (14, 23 g and 101 meters) cover the first deployment. Maps e-h (14, 23, 62 and 120 meters) cover the second deployment. •t.4 •* .0 1.« -1.8 -0.8 Log Frequency •0.* 0.0 o.« o.a ^ Meeeecele Bend k. D I .-SD IGW Bend } 1 J ^ Log Frequency « 0.# o J u lia n D a y 1 9 8 7 Ju lia n D a y 1987 8B0.0 0 •1 0 00 0 0 0 o 0 0 0 0 0 0 Log Frequency ISO ICO 100 Log Frequency Log Frequency Log F requency Figure 4.12. 7(t,f) between temperature and Chi. Map a (101m) cover the first deployment. Maps b (62m) and c (120m) cover the second deployment. Maps d-f (62, 81 and 101 meters) cover the transition between the highly advective regime at the end of the first deployment and the quiescent physical regime at the beginning of the second deployment. Zll Julian Da; 19 87 Julian Day 19 87 Julian Day 19 87 $ 'O .SO' 1 I < 3 ■o.os. •O .O S- Julian Day 19 87 •0.9S. f I i ■ 0 . 8 S I e Julian Day 1 9 8 7 Julian Day 1 9 6 7 Julian Day 1 9 8 7 Julian Day 1 9 8 7 Figure 4.13. Y (f,t) between C c and Chi. Maps a-c (23,101 and 160 meters) cover the first deployment. Maps d-g (62, 81,101 and 120 meters) show the available data from the second deployment for each depth. Map h (81m) covers the third deployment. Maps i and j (62 and 81 meters) span the second third deployments (JD 200-280), illustrating the transition from quiescent conditions to increasing horizontal advection. variance in the C c and Chi fields at 23m (Figs. 4.8a, 4.9a). In fact, although it is not characterized by elevated productivity, this period of extremely low current speeds and tem perature variability shows the highest levels of coherence between the two bio-optical variables, with values ranging up to 0.9 (Figs. 4.13 a-c). Additionally, coherence between physical and bio-optical signals are not significant (e.g., Fig. 4.12a). The last three days of this period show continuing low current velocities and the initialization of perm anent seasonal stratification at the site. Additionally, an episodic increase in C c , less prominently featured in the Chi time series, appears through the upper 100 m eters (Figs. 2.5 and 2.6). Since incident PAR is extremely low at this time, it would appear th at this bloom is adverted through the site, despite the extremely low current velocities. A closer view of the coherence map for C c and Chi a t 23m (not shown) shows the maximal values (up to 0.86) for the deployment coinciding with the passage of this ecosystem which is apparent in C e through 100 meters (Fig. 2.5). Following this event, mesoscale coherence between the two bio-optical variables drops precipitously as the euphotic zone Chi time series transition to a regime characterized by a series of low frequency, high magnitude fluctuations which extend into the second deployment a t and below 60m. This is not duplicated in the second deployment C e time series and is unobservable in the first deployment due to incomplete data return. 115 Period 3: (JD 90-130) This period has already been described as being dominated by strong advection driven by heavy mesoscale activity, represented by the consistent presence of a cold-core eddy ju st northwest of the mooring (Figs. 2.8 c-f). The maximum current speed for the experiment, 125 cm/s through 40m, is attained as the currents undergo a rapid directional veering during this period and the energy overall is greater than during the first period. Depth increasing coherence between individual velocity components and tem perature (Figs. 4.10 a, b) is emphasized by the two 160m maps which show coherent mesoscale activity starting after JD 90 (Figs. 4.10 e, f). In the zonal case (Fig. 4.10e), significant coherence lasts through JD 105 and ceases in concert with the establishment of essentially meridional flow (Fig. 2.2). This characteristic, including the temporal bounds, is also observed in the 23m map of coherence between zonal current and DO (Fig. 4.14a). Coherence between meridional velocity and tem perature is significant throughout this period (Fig. 4.10f). Further evidence of the passage of distinct water masses during this period is shown in maximal mesoscale coherence between tem perature and DO a t 14 and 23 m eters (Figs. 4.11 b, c) and a t 101m between tem perature and Chi (Fig. 4.12a). However, it is interesting to note th at mesoscale Chi variance a t this depth is below the experimental average throughout the first deployment (Fig. 4.9d). A difference more critical than the relative magnitude of horizontal energy between this period and the first is the presence of stratification and the resultant IGW field. This is apparent in the maps of Gn ( / i g w ) for 116 Ju lia n D a , 1 9 8 7 J u lia n D u , 19 67 Log Frequency i n d ^ ^ D > 2 0 . 0 • 0 0 .0 1 7 0 .0 1SO.O Log Frequency Log Frequency i a . 3 Log Frequency Figure 4.14. 7<f,t) between DOX and individual current directions. Map a shows ?(f,t) with zonal current at 23m during die first deployment. Maps b-d show y(f,t) with meridional current at 23, 62 and 120 meters. tem perature. IGW activity, appearing after JD 90 a t 14m and JD 105 a t 23m (Figs. 4.7 g, h), corresponds well with the development of stratification (Fig. 2.15). The C e maps (Fig. 4.8) also show elevated IGW variability a t all depths during this period and correspond well with the characteristics of the velocity field (Fig. 4.6). The characteristics of the Chi field are distinctly different from those seen in the physical and particle fields. The G n ( / i g w ) maps show maximum variability for low frequencies within / i g w a t 23m and later in time with increasing depth, apparently corresponding to a threshold level of stratification (Fig. 4.9). Indeed, the presence of a minimal level of stratification seems responsible for a transition in IGW as well as mesoscale band characteristics in Gh(J) and Y x y C f) maps in several fields. Over the course of this deployment, O n ( / i g w ) in the velocity field decreases by an order of magnitude at 23m and several fold a t 101m (Figs. 4.6 g, h) indicating that stratification restricts higher frequency horizontal motions. In addition, coherence between tem perature and Chi a t 62,81 and 101 m eters exhibits an inverse relationship between / m and / i g w . whereby decreasing Y xy(fM ) coincides with increasing YxyCflcw) (Figs. 4.12 d-f). This is best illustrated a t 62m where, between JD 100 and JD 110, coherence £ 0.4 occurs in conjunction with a local maximum in / b v at JD 105 (Fig. 4.7c). Following this temporary peak in / b v . significant mesoscale coherence reemerges briefly (JD 115-125). Thus, stratification triggers a transition in the pigment field from mesoscale variability caused by advection to variability resulting from elevated phytoplankton growth rates. Thus, a transition between physically and biologically dominated mesoscale 118 variability is suggested. Further evidence for this is presented in the following section since this transition continues into the second deployment. Deployment 2 The second deployment will also be presented in three distinct time periods, again based on the presence or absence of advective activity. The second of these periods (JD 190 to 220) is distinguished by the presence of the major mesoscale event of the deployment and is illustrated by the synoptic maps (Figs. 2.9 c, d). This feature appears prominently in the tem perature time series and contour map (Figs. 2.3 and 2.11) and as a pronounced veering of the current vectors (Fig. 2.2). The associated current speeds are the highest attained during the deployment. The following period (JD 220 to 240) consists of low current velocities and increasing stratification below 80m (Fig. 2.15). The synoptic maps reveal a lack of activity around the mooring site. The deployment’ s initial period (JD 135-190) has sim ilar characteristics as those of the final period, low current velocities and minimal advective phenomena. In addition, stratification is intensifying over the entire w ater column. Period 4: (JD 135-190) The drop in current energy between the first and second deployments is dramatic and the magnitude remains low throughout this period (Fig. 2.2). Additionally, this tim e frame is characterized by mesoscale energies as low as 20% of the experimental average (Figs. 4.6,4.7). This is well-represented in the deployment averaged current spectra and the low mesoscale energies serve to 119 emphasize the spectral peaks a t f\ and fso (Figs- 4.1 c, d). The sub-mixed layer tem perature spectra differ only in th at the mesoscale band is not as low (Fig. 4.2 d), due to the more pronounced manifestation of advective activity in th e tem perature time series. An interesting coherence feature, generally between physical variables, appears in the mesoscale band from JD 150 to JD 170. The lim ited bandwidth and temporal placement provide a distinctive shape of which manifestations are found in coherence maps ranging from 14 to 120 meters. Maps revealing th is feature include coherence between meridional current and tem perature from 14,62 and 81 meters (Figs. A4.2.3 d-f) and coherence between meridional currents and DO a t 120m (Fig. 4.14d). Coherence between tem perature and C e a t 23m also shows this feature (Fig. 4.15a). Finally, tem perature and DO at 14, 23 and 62 meters show lowered coherence compared to elevated, significant values in the surrounding temporal/frequency space (Figs. 4.11 c-e). It appears that, although current velocities are extremely low during this period, a series of directional reversals in the meridional currents cause this feature (Fig. 2.2). Visual inspection of the time series does not reveal obviously coincident events during this period. However, the synoptic maps indicate the presence of several mesoscale eddies. Although there is no sign of direct interaction in the tim e series, it is likely th at the eddy positions relative to the mooring drive the meridional current directional reversals through the far entrainm ent field of the eddies. The time scale indicated in the coherence maps (~ 5 days) and the magnitude of the meridional velocity (~ 5 cm/s) suggest a meridional advection 120 Julian D a y 1M 7 s i 1 " f f t i s & i g Figure 4.15. 7(f,t) between temperature and Cc. Maps a (23m) and b (120m) cover the second deployment. * - * Map c (101m) covers the first deployment. length scale (~ 200 km). It is felt th at this feature, which does not have a strong manifestation outside of the various coherence maps presented here, is due to meridional variation in mixed layer and seasonal thermocline characteristics. Corresponding meridional ecosystem variation is also indicated a t 23m (Fig. 4.15a), but there is not a depth consistent pattern for C c and a following paragraph will suggest Chi mesoscale variability to be biologically derived. The second deployment map of depth varying coherence between tem perature and DO shows maximal values in the upper 30m a t fsD (Fig. A4.2.8) which manifest as distinctive peaks in the 23m power spectra (Figs. 4.2c and 4.3c). The 23m temporal map shows these features lasting through the entire deployment. This tidal feature and its summertime prominence are due to the 23m MVMS being at the top of the seasonal thermocline where stratification is extremely elevated (Fig. 2.15). W hat is of more interest for the first period is the extension of significant coherence well into the IGW band (Fig. 4.11a). This dissipates with the onset of increased mesoscale activity. This high frequency feature appears to be due to the creation of a strong vertical DO gradient resulting from a combination of atmospherically ventilated DO values within the mixed layer and productivity modulated DO values within the seasonal thermocline, the site of the deep chlorophyll maximum (Fig. 2.14). Evidence for elevated biological activity within the thermocline during this period is observed in the pigment field and presented in the following paragraph. 122 Elevated mesoscale Chi variance initiates toward the end of the first deployment, continues through the first period of the second deployment (Fig. 4.9) and dissipates with the onset of the second period’ s mesoscale activity (Figs. 4.6 a, b and 4.7). Thus, it appears during minimal advective activity in the physical variables. Additionally, it corresponds temporally to elevated variance in the IGW band which in turn corresponds to increasing IGW activity in the tem perature maps. Corresponding mesoscale and IGW features in the Chi maps appear later in time with depth (i.e., JD 90 a t 62m, JD 115 a t 81m and JD 125 a t 101m). It was previously noted th at a t 23m during the first deployment, elevated IGW variance in Chi shows a strong correspondence to elevated IGW activity indicated by the maps of tem perature Gn. This is not evident deeper in the w ater column since the dominant IGW fields occur later in the second deployment corresponding to elevated / b v (Figs. 4.9 b-d, 4.7 c-e). However, the onset of elevated IGW variance in the Chi maps shows excellent correspondence to the establishm ent of stratification a t depth as indicated by the .0025 ‘C/m isoline in the stratification map (Fig. 2.15). The observed mesoscale activity in the Chi field reflects a series of large magnitude fluctuations, on the order of 1 pg Chi a/L (Fig. 2.6). These fluctuations are most likely the product of locally driven ecosystem dynamics since they coincide with increasing stratification and minimal horizontal advection [Sverdrup, 1953; Siegel et al., 1990b]. Of the C c maps from 62,81 and 101 meters (Figs. 4.8 b-d), the 81m case shows elevated mesoscale variance during the fourth period while the 101m map shows a mesoscale expression (JD 165-195) prior to the major advective feature of the second period. The 81m map of coherence between C c and Chi 123 also reflects a strong coupling between the signals in the mesoscale band during the first period through JD 190 (Fig. 4.13 e). The deep C c maps (Figs. 4.8 d, e from 101 and 160 meters) strongly resemble the current field since they show maximal variance across the mesoscale and IGW bands throughout the first deployment followed by a sharp decrease into the second deployment. However, the 101m map does indicate elevated variance in low IGW band frequencies, well into the second deployment, coinciding with maximal G n at these frequencies in the Chi map (Fig. 4.9c). The bio-optical power spectra (Figs. 4.4d, 4.5d) indicate th at these dominant frequencies are a melding of signals a t / d and / i in addition to / s d - Both / i and / s d are prominent in the corresponding power spectra for the physical variables (Figs. 4. Id, 4.2d). Coherence between the bio-optical variables is strongest over this frequency range and most dominant at / s d - In addition, the strength of this coherence increases with depth from an expression a t 62m which is ju st below statistical significance to values greater than 0.85 a t 120m (Figs. 4.13 d-g). Significant coherence is also observed a t / s d between tem perature and Cc a t 120m and at 62 and 120 m eters between tem perature and Chi (Figs. 4.12 b, c and 4.15b). Values in the latter case are exceedingly high, again indicating the tendency for coherence a t / s d to increase with depth. These features indicate the presence of vertical gradients in both bio-optical variables as well as tem perature and th at tidal activity is responsible for moving these scalar fields vertically past the moored instrum ent array. 124 Period 5: (JD 190-220) The onset of the second period's mesoscale activity is apparent in a number of ways in the temporal Gn maps. During this time, the current maps reveal mesoscale expressions which decrease in magnitude with depth and never reach values of Gn >2.4 (Fig. 4.6). In contrast, the tem perature maps show the strongest expression of this advective feature at 62m (Fig. A4.2.2c), where values of Gn reach 4.0, and the weakest a t 14m, where values are less th an the experimental average due to the mesoscale activity of the first deployment (Fig. 4.7). Contour maps of the cross-spectral analysis results consistently show elevated coherence between the physical variables, including DO. It is interesting to note that, although the current vectors show the two components to be equally prominent (Fig. 2.2), significant coherence between velocity and the other two variables only occurs for the meridional component. This indicates zonal homogeneity and a strong meridional gradient for both tem perature and DO. Coherence estim ates corresponding to this mesoscale event are generally greatest between JD 195 and JD 205. This appears a t 23, 62 and 81 m eters in the maps for the meridional and tem perature case (Figs. A4.2.3 e-g) and a t 23 and 62 meters in the meridional and DO case (Figs. 4.14 b, c). It should be noted for the latter combination th at this feature does not appear at 120m (Fig. 4.14d) and th at coherence is insignificant below 62m (figure not shown). Coherence between tem perature and DO is generally strong. This is especially the case w ith the passage of the advective feature where coherence is > 0.9 a t 23 and 62 meters (Figs. 4.11 a-c). 125 The C c maps from this period show significantly decreased Gn in the mesoscale band with values less than half the experimental average a t 23,62 and 81 m eters (Figs. 4.8 a-e). This is also characteristic of the Chi maps (Figs. 4.9 b, c) although the values of Gn are not as low as observed in the C c maps. The maps of coherence between the bio-optical measurements all show low values during this time and generally show the surrounding periods to be characterized by a stronger linear relationship between the two variables (Figs. 4.13 d-g). It is also noteworthy th at coherence at / s d between tem perature and the two bio-optical variables decreases precipitously during this period. This attribute is prominent a t 120m for C c (Fig. 4.15b) and a t 23 and 120 meters for Chi (Figs. 4.12 b, c). The characteristics of this period emphasize the inverse dependence between the degree of linearity relating the physical processes and the degree of linearity relating the biological processes. In other words, during highly advective regimes, the strongest inter relationship occurs between the physical variables (i.e., currents, tem perature and DO) whereas for regimes characterized by local physical forcing and minimal advection, the strongest inter-relationships occur between C c and Chi and between the two bio-optical variables and tem perature. The strength of the latter relation is primarily based on vertical oscillations driven by tidal and/or internal wave activity the influence of which diminishes during the presence of significant advection. Period 6: (JD 220-240) The current field during this period is characterized by low magnitude, generally northward flow (Fig. 2.2). This may be associated with the far 126 entrainm ent field of the eddy shown in the synoptic maps (Fig. 2.9e). The current maps show minimal Gn throughout this period in all frequency bands (Fig. 4.6). The tem perature maps show this period to be characterized by elevated IGW activity and minimal mesoscale variability (Fig. 4.7) indicating th a t this is a regime for which bio-optical variability will be the result of biological rath er than physical processes. The m ost notable characteristic during this period in the time series is the peak which occurs simultaneously in the C e and, more prominently, in the Chi time series a t 62m from JD 230 to JD 234 (Figs. 2.5 and 2.6). These episodic peaks (i.e., phytoplankton blooms) indicate a divergence from steady state growth. This growth regime is characterized by sufficient, but not overabundant, nutrient concentrations obtained via recycling processes and/or constant vertical transport via turbulent diffusion. This transition from steady state production to an episodic bloom implies an influx of nitrate. These peaks appear as mesoscale features in the Gn m aps for the two variables (Figs. 4.8b and 4.9b). Strong mesoscale coherence a t 62m is seen. At both 62 and 81 meters, elevated coherence appears ranging over /p , f \ and /sd and the lower IGW band. (Figs. 4.13 j, k). Another aspect of this event is the coincidentally occurring IGW band maxima which appear in both 62m maps. The tem perature map shows the IGW field at 62m to be at its maximum during this period (Fig. 4.7c). However, reduced values appear below 62m (Figs. 4.7 d-f). These are preceded by a decrease in stratification which is visible below 101m in the contour map from JD 210 to 240, a period of elevated shear down to 90m around JD 215, and a local minimum in /b v a t 101m from JD 220 to JD 225 (Figs. 2.15,2.16,4.7e). 127 This combination of reduced stratification and elevated shear provides a mechanism for the ii\jection of nitrate into the lower euphotic zone, setting the stage for the observed phytoplankton bloom. However, it is interesting to note th at there is little evidence for corresponding blooms in the 81 or 101 m eter bio-optical time series. At 101m this is likely due to light limitation as this is near the base of the euphotic zone (Fig. 2.1g). At 81m a weak bloom is indicated in the c « time series which m anifests as a mesoscale feature accompanied by elevated IGW band variance as was true a t 62m (Fig. 4.8c). D iscussion Spectral analyses of these physical and bio-optical time series have revealed specific characteristics within the signals for individual param eters and a number of relationships between physical and biological processes. The deployment averaged power spectra show the presence or absence of spectral peaks associated with specific processes. Seasonal and depth variations are also indicated. In the velocity data these include depth consistent variance associated with the semi-diurnal tide, a half decade decrease in inertial energy between 23 and 101 meters and order of magnitude differences in mesoscale energy between deployments. The tem perature spectra only show peaks at Si and S s d below the mixed layer, underscoring the need for a vertical gradient in order to detect vertical oscillations. The dissolved oxygen (DO) spectra show a strong resemblance to the tem perature spectra but include a diurnal peak associated with the growth and respiration cycles of photosynthesis. This peak dominates the near-surface bio-optical (ct and Chi) spectra but peaks at 128 f l and fsD» which become more prominent with depth as the diurnal signal weakens, are also present. The consistency of this feature (i.e., depth increasing Gxx(/l> J sd) between tem perature and the hio-optical variables is a harbinger of the observed dependence of the biological system on the thermal field. The Gn maps do not show these spectral peaks since they contain normalized data. However, they emphasize the temporal or spatial distribution of spectral energy. Comparing Gn distributions between different variables provides insight into how the measured param eters are interrelated. Direct quantification of these interrelationships, especially the paramount interest in physical/biological interaction, is provided by coherence maps. Significant coherence indicates a strong linear relation between two param eters. The first two deployments of this mooring experiment alternate between physical regimes which are highly advective in nature and those which are not so [Dickey et al., 1993]. Additionally, the advective and non-advective regimes have been partitioned based on whether homogeneous or stratified waters are present. Six periods exhibiting combinations of these general characteristics have been defined over these deployments. The first and third periods consist of large magnitude, barotropic currents which appear as elevated mesoscale Gn at all depths in the current field. Elevated mesoscale Gn is also represented in the near-surface tem perature field. These two high energy regimes are separated by a period characterized by low current speed and deep convection indicated by the presence of 18s waters. The third period also shows a uniform 129 veering of the currents, indicating a strong interaction with a mesoscale feature, and the onset of perm anent seasonal stratification. The latter feature coincides with decreasing current Gn and increasing tem perature Gn in the IGW band. The fourth and sixth periods are characterized by low current speeds and elevated stratification. The intervening period is dominated by the passage of a mesoscale feature which appears strongly in the tem perature data and as a veering in the current vectors, although the current magnitude is not strong in comparison to the levels seen during the first deployment. This event is strongly represented at all depths in the mesoscale band of the tem perature Gn maps and in the near-surface current maps. The presence of these various regimes around the mooring allows for the observation of bio- optical processes under a variety of physical conditions and provides insight into several physical/biological interaction mechanisms. Significant coherence between oceanic currents and tem perature as well as currents and DO suggests the passage of distinct water masses. There are indications of several separate events within the mesoscale band of the temporal maps and coherence in the spatial maps increases w ith depth. Generally, passage of distinct water masses occurs during highly advective periods such as the near-surface feature from JD 70 to 80 and the deep feature from JD 90 to 115 during the first deployment, the major second deployment tem perature feature and two third deployment features which are split in time and depth, simila r to the first deployment. However, during the experiment's most quiescent current regime (JD 150 to 170), a change in w ater mass is indicated at all depths in coherence maps between currents and both 130 tem perature and DO. This corresponds in time to a reversal in the meridional component of the current velocity. Coherence between tem perature and DO is prominent due to their physical-chemical relationship [Weiss, 1970]. Thus, this combination is an excellent indicator of distinct water m ass passages which appear in the mesoscale band and the vertical oscillation of a stratified water column which appear at tidal and higher IGW band frequencies. A depth comparison of first deployment mesoscale coherence in the temporal m aps reveals a brief period of significance a t 14m whereas the 23m map shows significant values throughout. This indicates th at processes related to gas exchange across the air-sea interface have a significant influence on the DO signal a t 14m. However, it is also interesting to note th a t coherence between these variables is only significant during highly advective regimes (i.e., the first and third periods) at 101m. The dominance of 23m coherence is due to th is depth being relatively isolated from the influences of air-sea interaction while being strongly affected by episodic appearance of surface stratification associated with advective activity during the first period. The fourth period shows extremely high coherence a t tidal and higher IGW band frequencies near the surface which decreases w ith depth. T his is due to th e presence of depth decreasing stratification, which has maximal values a t 14m. During the fifth period, mesoscale coherence attains maximal values (-0.8) due to the passage of the tem perature feature and coincides with a decrease in IGW band coherence. 131 The spatio-temporal distribution of coherence between tem perature and the two bio-optical variables is sim ilar for both cases. Coherence between tem perature and C c a t 101m is highly significant during the third period while coherence between tem perature and Chi at 101m is significant during both the first and third periods. Significant mesoscale coherence is disrupted during both the second period, characterized by minimal advection and deep convection, and the fourth period, characterized by low advection and emerging stratification. Thus, significant mesoscale coherence between ecosystem param eters and tem perature appears during periods of elevated advective activity. Over the first two deployments, both bio-optical cases show coherence levels a t fsD which change due to the degree of stratification and advection. During the period characterized by low advection and low stratification (JD 78-90), coherence is minimal (~0.3), while during the period of high stratification and low advection (JD 135-190), coherence in this band attains maximum values (-0.6-0.7). Obviously, stratification provides the framework for the appearance of this tidal/IGW signal. Finally, during the heavily advective periods, values of coherence are observed to be between the extremes ju st discussed. Coherence a t low frequencies between the bio-optical variables also display such an inverse relation with the presence of elevated advection. Through the first two deployments, maximum mesoscale coherence is attained during both periods of minimal advection. These changes in coherence m ark a transition between physically and biologically dominated low frequency ecosystem variability a t the site [Cullen et al., 1983; Denman and Powell, 1984]. In addition, coherence in the band ranging from / d to / sd increases 132 w ith depth as it increases in frequency. This indicates that near the surface, significant coherence is due to photosynthetic processes. These become less influential a t depth due to reduced light levels and the increased prominence of the tidal signal below the homogeneous surface mixed layer. This shift in the dominant coherence frequency indicates another transition between biologically and physically dominated variability. Stratification is a vital environmental component as it provides the framework which supports the phytoplankton community w ithin the euphotic zone and elevates the daily dose of PAR as the vertical excursion length is reduced, resulting in increased growth rates [Sverdrup, 1953]. With the development of therm al stratification and vertical gradients in the bio-optical fields, increased variance within the IGW band appears, especially at fso- Previous works have estim ated that one third of the vertical variance occurring within the IGW band is accounted for by the semi-diurnal tide [Gregg and Briscoe, 1979; Denman and Gargett, 1983] and it has been estimated that vertical motions such as these have a net positive effect on integrated w ater column production [Holloway, 1984; Holloway and Denman, 1989]. The combined presence of stratification and elevated biomass recorded in the C e time series through the upper 101m, which occurs toward the end of the second period, illustrates the ecosystem's dependence on this characteristic. Although this is not a highly advective period, this "bloom" does not appear to be locally generated since it coincides with low levels of PAR Estim ated current speeds over this period range from 15 to 18 cm/s and have an essentially constant direction while the duration of the bloom passage 133 ranges from 2.5 to 2.9 days. The range in these estim ates is due to differences observed over the upper 40m. This results in an estim ated patch scale of 23- 30 km which decreases deeper in the w ater column. Without coincident synoptic pigment measurements, determining what portion of the patch passed the mooring is impossible. This makes precisely determining this length scale problematic although this estimate is within previously observed spatial bounds O(50km) [Denman and Abbott, 1988]. Due to minimal advective activity, the fourth period provides an ideal setting for observing local ecosystem dynamics in the presence of emerging stratification. Although this period does not coincide with maximum IGW activity seen in the tem perature Gn maps, a developing IGW field is indicated in the tem perature maps and elevated IGW variance is present in the Chi maps. The latter feature coincides with the development of a threshold level of stratification which penetrates into the water column, causing this IGW feature to appear later in time with depth (i.e„ JD 90 at 62m, JD 115 at 81m and JD 125 a t 101m). Due to a series of episodic phytoplankton blooms, the mesoscale band also contains elevated variance in concert with this IGW variability. This is not reflected in Gn maps of C c , although coherence between the two bio-optical variables attains values > 0.85 a t depth (i.e., 120m) in the fsD band during this period. As mentioned earlier, the blooms indicated in the Chi data coincide with the development of minimal stratification and follow winter’ s deep convective mixing which transports nutrients to the surface w aters. Therefore, the observed IGW variance is caused by the development of a vertical gradient in 134 the Chi field interacting with the evolving IGW field and the blooms appearing as mesoscale variability are the result of the trapping of phytoplankton within the euphotic zone [Sverdrup, 1953], although IGW pumping may have some influence a t this depth [Kahru, 1983; Sandstrom and Elliott, 1984]. These observations emphasize stratification’ s role in supporting the ocean’ s primary ecosystem [Stram ska and Dickey, 1994] and the onset of episodic phytoplankton blooms, with depth varying temporal placement, is reminiscent of the meridional evolution of the Sargasso Sea’ s spring bloom [Siegel et al., 1990b]. The sixth period shows minimal advection in both the current and tem perature fields and elevated IGW variance. Indeed, the tem perature Gn maps show this period to contain the highest levels of IGW activity for the entire experiment. Thus, it has the proper characteristics for the observation of locally generated ecosystem dynamics. However, a t this tim e of year, primary productivity is generally characterized by steady state growth since it depends upon nutrients derived from recycling m aterials available within the euphotic zone. Thus, the observation of a phytoplankton bloom a t 62m in both the C c and Chi time series reveals a deviation from steady state growth and indicates the occurrence of a nutrient iqjection. This bloom appears as a mesoscale event in the 62m Gn maps for both variables and has a significant signature in the mesoscale band as well as the diel band in the coherence map. Further crucial characteristics which may be observed a t this time are the presence of elevated IGW Gn in the 62m tem perature map, a weakening of deep stratification (i.e., below 100m) which is also reflected as reduced values of / by and an increase in current shear at this depth. These characteristics provide a mechanism for introducing nitrate into the base of the euphotic zone. From there, it is hypothesized th a t the resultant bloom is generated by a combination of vertical oscillations associated with tidal and low frequency internal waves, leading to an apparent vertical flux of nitrate, and direct physical mixing due to breaking internal waves [Kamykowslti, 1981; Denman and Gargett, 1983; Kahru, 1983; Denman and Powell, 1984]. 136 C h a n t e r K ; A one dim ensional in terd iscip lin ary m odel Introduction Objectives behind the development of a 1-D interdisciplinary model The upper ocean is an extremely active region and is the focus of a variety of processes encompassing a considerable range of today’ s oceanographic disciplines. One-dimensional studies of the pelagic ecosystem in the upper ocean are faced with quantifying: 1) variations in water column therm al structure as it relates to the development of the seasonal mixed layer and thermodine [Siegel et al., 1990b]; 2) the magnitude and distribution of diapycnal diffusivities responsible for vertically distributing the thermo hydrodynamic, planktonic and nutrient fields [Gregg, 1987; Gargett, 1989]; 3) attenuation, via w ater and phytoplankton biomass, of photosynthetically available radiation (PAR) [Baker and Frouin, 1987; Morel, 1988]; 4) growth characteristics, including light adaptation and the kinetics of nutrient preference/uptake, of the autotrophic representative^) within the ecosystem [Evans, 1988; Dortch, 1990; Kiefer, 1993]; 5) grazing characteristics of the resident zooplankton population [Frost, 1975; Totterdell et al., 1993] and; 6) other trophic interactions (e.g., inclusion of the microbial loop and detrital fluxes) which are not as well quantified as the planktonic processes [Betzer et al., 1984; Cho and Azam, 1988]. Interdisciplinary models encompassing both physical and ecological aspects of the pelagic environment are useful for studyingfemulating observed physical/biological interaction mechanisms on seasonal [Menzel and Ryther, 1960; Kiefer and Kremer, 1981; Deuser, 1986] and episodic [Klein and Coste, 1984; M arra et al., 1990] temporal scales, biological feedback into the therm al field through biomass specific irradiance attenuation [Morel and Antoine, 1994; Stram ska and Dickey, 1994] and the ocean’ s potential role in mitigating anthropogenic emissions via carbon sequestration [Longhurst, 1991; Sarmiento and Siegenthaler, 1992], The pelagic ecosystem as referenced in the following review of previous modeling endeavors, and as defined in the present approach, consists of a number of compartments and processes which have been summarized in the flow chart shown in Fig. 5.1. Because of the dearth of quantitative information available for some of these compartments, most notably the magnitude of the dissolved organic m atter (DOM) pool and the specifics of the uptake and rem ineralization processes associated with the bacterial compartment, a number of interdisciplinary modeling studies have chosen not to explicitly include all portions of the pictured ecosystem. However, in the present formulation, the ecosystem will be as presented in the flow chart in order to retain the ability to investigate vertical fluxes of organic m atter which allows for m aking some estim ates of the experiment site’ s importance to carbon cycling. Additionally, as future oceanographic studies are performed on the problematic components of the ecosystem as represented herein, greater confidence that individual compartment’ s roles are accurately portrayed will develop and their inclusion in the presently extant schemes may provide insights into how in-situ investigations should proceed. Finally, interdisciplinary models "calibrated" with presently available data, may be used as tools for further investigations into interaction mechanisms and 138 Phytoplankton Nitrate DON A m m o n iu m Zooplankton Bacteria Detritus F ig u re 5.1: In te rc o m p a rtm e n ta l flo w s fo r th e e c o sy ste m m o d el T his flow ch art illu stra te s nitrogen flow pathw ays betw een individual com partm ents and th e deep ocean as param eterized w ith in th e ecosystem m odel (Fasham e t al., 1990). ecosystem adjustm ent strategies under the influence of parameterized physical forcing. Previous interdisciplinary modeling studies Investigation of biogeochemical cycles through the use of modeling has been approached in a variety of ways. In order to isolate specific interactions within the ecosystem and to determine the most suitable means of param eterizing individual components of the ecosystem, modeling studies using reduced equation sets and/or imposed physical fields have been employed. For example, an implicit parameterization for higher order grazing pressure is necessary to provide closure in the highest explicitly modeled trophic level (i.e., zooplankton). Studies using a reduced equation set found th at incorporating a quadratic form for this term resulted in more robust solutions [Steele and Henderson, 1992]. Another approach has combined basic physical forcing, through the imposition of a seasonal time series of mixed layer depth, with a phytoplankton- zooplankton-nutrient (PZN) model to reproduce observed seasonal cycles such as the spring bloom in the planktonic system [Evans and Parslow, 1985]. Finally, a more complex set of ecosystem equations which included explicit relations for the microbial loop and sinking particulate m aterial was similarly coupled with a mixed layer tim e series taken from the Sargasso Sea [Fasham et al., 1990]. This modeling experiment was compared w ith some success to in-situ observations acquired over an extended period a t Bermuda Station “S” (32*N, 65*W), near Bermuda [Menzel and Ryther, 1961; Jenkins and Goldman, 1985; Altabet, 1989]. However, the utilization of a 140 mixed layer time series with an ad hoc parameterization for mixing between the surface layer and the deep ocean proved to be overly restricting. Incorporating more realistic physical forcing with pelagic models has resulted in the ability to emulate more precisely the observed development and evolution of such characteristics of the prim ary ecosystem as the spring and fall blooms and the development of a summertime deep chlorophyll maximum (DCM). In an interdisciplinary model where both the mean and turbulent fields are predicted using a turbulent closure scheme [Mellor and Yamada, 1974], the position of the DCM is shown to be sensitive to the magnitude of the predicted vertical difiusivities [Varela et al., 1992]. Another one-dimensional interdisciplinary model using a planetary boundary layer model employing a non-local turbulent mixing scheme [Large et al., 1994] has been developed using the biogeochemical time series acquired at Bermuda Station “S” [Lohrenz et al., 1992; Michaels et al., 1994b] as a comparative data set for the model results [Doney et al., 1995]. This formulation also shows reasonable replication of the DCM and spring bloom with a more reasonable representation of nutrient levels and f-ratio, defined as the ratio of new production (P P ncm) to total production (i.e., P P no3 + regenerated (P P niw)) [Dugdale and Goering, 1967], than was predicted in the previously described Bermuda Station “S” based formulation [Fasham et al., 1990]. However, the high levels of surface production which are observed to continue throughout the summer [Michaels et al., 1994b] are not well represented. This is most likely due to a combination of the model’ s excessive export of organic m aterial from the surface layer via particle flux in conjunction with a non-parameterized 141 spedation transition which results in an in-situ primary production regime characterized by highly efficient, nutrient recycling [Doney et al., 1995]. Further insights into areas needing additional investigation have been indicated through three-dimensional modeling experiments. The seven compartment ecosystem developed using the Bermuda Station “S” data [Fasham et al., 1990] has been successfully incorporated into a seasonally forced general circulation model (GCM) of the North Atlantic a t Princeton's Geophysical Fluid Dynamics Laboratory (GFDL) [Sarmiento et al., 1989]. Results from this series of modeling experiments have been compared to synoptic, annual mean, and seasonal chlorophyll distributions based on CZCS images and in-situ data from both Bermuda Station “ S” and Ocean W eather Station “India” (59s N, 19s W). Overall, correspondence between the annual m ean chlorophyll map and the model results is decent. Additionally, discrepancies are as much due to poorly predicted physical transports, such as a downwelling region in the Gulf of Guinea where the CZCS reveals high biomass characteristic of upwelling, as to poorly parameterized ecosystem characteristics for a particular region [Sarmiento et al., 1993]. Comparison w ith the in-situ time series revealed th at the amplitude of the spring bloom was far too high and th at this was a t least partially due to the overwintering zooplankton stock being too low. Additionally, the model emphasized the fact th a t there is a need for additional quantitative studies of the sinking rate of detritus, its degradation to DOM and the role of the latter in the microbial loop [Fasham et al., 1993]. 142 These questions were addressed in another modeling endeavor designed to investigate a recent controversy regarding the relative importance of particle flux and DON in the downward transport of organic m atter, in light of recent measurements which revealed much higher concentrations of DON than were previously observed [Suzuki et al., 1986]. Differentiating between these two transport mechanisms is of interest since it relates to the degree to which the surface euphotic zone is coupled to the aphotic zone directly below in term s of the regeneration of nutrients since detrital fluxes are essentially unaffected by horizontal currents. Varying the relative contributions of the two mechanisms for organic m atter flux in the modeling experiments resulted in the observance of enhanced nutrient trapping in upwelling regions due to reduced sinking rates in a "particles-only" experiment whereas "DOM-only" experiments resulted in more realistic nutrient distributions in these areas [Najjar et al., 1992]. Development of the interdisciplinary model The interdisciplinary model developed in this work is based on combining two separate models consisting of: 1) a 1-D physical model used to predict the seasonal evolution of the oceanic mixed layer and; 2 ) a pelagic ecosystem model used to predict m aterial fluxes between various trophic and nutrient compartments. As previously mentioned, the ecosystem model is comprised of seven compartments, consisting of phytoplankton, zooplankton, bacteria, dissolved organic nitrogen (DON), ammonium (NH*), nitrate (NO3) and detritus. The inter-compartment fluxes are quantified as flows of nitrogen (Fig. 5.1). The previous 1-D application of this ecosystem incorporated a mixed 143 layer time series which served to demarcate the surface boundary layer from the deep ocean. By coupling the ecosystem with realistic, locally forced dynamics, the effect of variations in the physical system on the pelagic system w ithin the seasonal therm odine as well as the mixed layer, a t temporal scales ranging from seasonal through episodic and diurnal, may be observed. The physical model The physical model utilized for these numerical experiments is the level 2.5 Mellor-Yamada second-moment turbulence closure model. This description of the model as a turbulence closure scheme refers to the fact th at it solves the turbulent, Reynold's averaged Navier-Stokes equations which have been reduced through a number of assumptions and for which closure is provided by param eterizing the higher order term s through observations made in turbulence experiments carried out in the laboratory [Mellor and Yamada, 1977; Mellor and Yamada, 1982]. The reference to second-moment closure is indicative of the practice of obtaining an equation for the Reynold's stress tensor through the application of a moment with the fluctuating velocities to the fluctuating momentum equation [Mellor and Herring, 1972]. The assumptions utilized to reduce the equations to the form applied by the numerical scheme indude Rotta's energy redistribution hypothesis, the Boussinesq approximation, low Rossby number, Kolmogorov's local small-scale isotropy hypothesis, no horizontal or mean vertical advection, and dosure assumptions for the diffusion term s whose magnitudes make them minimally im portant [Mellor, 1973; Mellor and Durbin, 1975; Blumberg and Mellor, 1980]. The final equation set (Appendix 5.1) appears to be no different than traditional 144 K theory format, whereby the Reynold's stress term s are param eterized as functions of vertical gradients of the mean flow field (Eq. A5.1.11). However, in this case, values for K are dependent on predicted values for turbulent kinetic energy (q, Eq. A5.1.7), turbulent length scale ( 1 » Eq. A5.1.8) and derived stability factors (S) which depend upon the flux Richardson num ber (cf. Fig. 1 in [Mellor and Durbin, 1975]). These derivations contain empirical constants which are taken from laboratory results of neutral turbulence experiments [Mellor and Yamada, 1974]. Two basic types of mixed layer schemes have been presented in the literature; 1) the differential or turbulence closure type described above and 2 ) the bulk or integral type [Kraus and Turner, 1967; Price et al., 1986]. The latter type integrates the primitive equations over depth which is varied to meet specified bulk stability criteria. One advantage of differential schemes over integral models is th at the development of the surface mixed layer occurs naturally and is not assumed a priori. Additionally, the integral models are less universal in their application in that once they are "tuned" to a specific site, their predictive capability is not assured for a different oceanic region [Martin, 1985]. Indeed, one of the original goals behind the development of the differential schemes was the hope th at there would be no need for reparam eterizing closure between applications [Mellor and Herring, 1972]. Differential schemes are designed to predict mixing within fully turbulent flow regimes. Thus, limitations to their predictive capability are expected w ithin the seasonal thermocline where turbulent events, whose mechanisms may indude double diffusion, internal wave breaking and mixing via shear 145 instability [Gregg, 1987; Gargett, 1989], are interm ittent. Param eterization of these processes is generally accomplished by setting threshold values to diffusion of heat and momentum which are modified by the bulk Richardson number during unstable flow regimes [Peters et al., 1988]. Previous parameterization8 of turbulent mixing below the surface layer in this model [Mellor, 1989] have been shown to be too low by recent studies [Large et al., 1994] and adjustm ents have been implemented which provide more realistic heat transport into the therm odine [Peters et al., 1988; Kantha and Clayson, 1994]. The interior difiusivity parameterization (Eqs. A5.1.13 a-d) used in this application of the mixed layer Scheme was chosen after a series of trials for which model results were compared to in-situ measurements. Components of the ecosystem model The ecosystem model employed here consists of a seven compartment nitrogen based pelagic system which has already been described (Fig. 5.1). Using nitrogen to monitor m aterial flows between compartments stems naturally from the fact th at it is used primarily as a structural m aterial whereas carbon and phosphorus are more directly involved in cellular metabolic activity [Roman, 1983]. Furthermore, this allows for the inclusion of both N species (i.e., NO3' and NH4*) within the compartment definitions. This provides the opportunity to emulate and observe the transition between the nitrate fueled spring bloom and the ammonium fueled steady state regime which follows. Thus, both new and regenerated primary productivity are distinctly incorporated and their relative contributions to carbon sequestration may be investigated. The equations used to describe the ecosystem are listed 146 in Appendix 5.2 and will be referenced in the ensuing qualitative descriptions. Additionally, a list of definitions for parameterized functions used in the equations is given in Appendix 5.3. Finally, values for constants employed in these functions are listed in Appendix 5.4. Prim ary production resulting from the autotrophs’ function as fixers of carbon through photosynthesis is the basis of the entire m arine ecosystem. Of all the components of the pelagic ecosystem, photosynthetic growth has been subjected to, by far, the most quantitative analysis. Numerous investigations have resulted in a robust understanding of photosynthetic chemical processes [Kirk, 1983]. Additionally, adjustment to varying environmental conditions such as tem perature [Eppley, 1972], nutrient [Dugdale and Goering, 1967; Laws and Bannister, 1980; Cullen et al., 1992b] or trace m etal [Martin and Fitzwater, 1988] concentration and the magnitude [Marra and Heinemann, 1982; Falkowski, 1984; Langdon, 1988] or spectral quality [Sathyendranath et al., 1987; Morel, 1991] of incident irradiance may manifest as a physiological response or a taxonomic shift. The phytoplankton equation indudes term s for photosynthetic growth, zooplankton grazing, density dependent mortality and turbulent diffusion (Eq. A5.2.1a). Phytoplankton growth employs the equation set (Eqs. A5.2.1 d-g) utilized for the gross production calculations in Chapter 3 (Eqs. 3.4-3.7). The one change involves the maximum spedfic phytoplankton growth rate (gV Eq. A5.2.1g) which is parameterized with a tem perature dependence [Eppley, 1972]. This allows for higher growth rates which is desirable since phytoplankton growth characteristics determined in-situ necessarily 147 incorporate losses due to heterotrophic activity [Larsson and HagstrOm, 1979]. In addition to adjusted empirical constants obtained from comparison with the 1 4 C production estim ates, the set of growth equations incorporates a function which allows for irradianee dependent modulation of cellular carbon:chlorophyll ratio (Eq. A5.2.1f) which has been shown to be im portant in the development of the subsurface chlorophyll maximum [Doney et al., 1995]. A previously utilized parameterization for nutrient switching (Eq. A5.2.1c) has been applied [Sarmiento et al., 1993] w ith a slight modification of the uptake kinetics (i.e., Ki, K2) pertaining to ammonium preference [Dortch, 1990]. The zooplankton equation (Eq. A5.2.2a) consists of three uptake term s, corresponding to each of the three food sources and a density dependent m ortality term. The zooplankton grazing parameterization is as originally developed for the Fasham ecosystem. The grazing rate formulations (Eq. A5.2.2b) are Michaelis-Menten based and allow the zooplankton to choose between food groups. This implies a succession of species w ithin an aggregated zooplankton community as is observed in the Sargasso Sea where a spring to summer transition from macrozooplankton to microzooplankton dominated grazing occurs [Roman et al., 1993]. This transition also implies a change in zooplankton feeding efficiency which has a substantial range between size classes and which has been shown to have a substantial im pact on correctly predicting seasonal production [Steele and Henderson, 1993], Food preference param eters are not well-quantified by observations, especially in the sense of seasonal cycles and species succession. Previous numerical experiments proved to be highly sensitive to the values chosen when employing constant preferences [Evans, 1988]. The food preference param eterization chosen here 148 is dynamically based on the food group's concentration [Hutson, 1984] and increases the likelihood of realizing an overwintering zooplankton population. The density based mortality term [Hofmann and Ambler, 1988; Steele and Henderson, 1992] is used to encompass zooplankton excretion of mineralized and dissolved forms of nitrogen [Bidigare, 1983; Verify, 1985] and loss to higher trophic levels [Slater et al., 1993] and provides further assurance of surviving the deeply convective winter regime by its transition to a quadratic form a t low concentrations [Totterdell et al., 1993]. The rem aining ecosystem equations (Eqs. A5.2.3-7) encompass the concentration of inorganic nutrients over the water column including nitrification within the aphotic zone (Eqs. A5.2.5 and A5.2.6) and remineralization and transformation of organic m aterials (i.e., particulate and dissolved forms of nitrogen) to inorganic forms utilizable by phytoplankton (Eqs. A5.2.3, A5.2.4, A5.2.7). The set of biogeochemical reactions involved in these processes includes a variety of phenomena which are not well-quantified a t the present time. Nitrification occurs within the aphotic zone and involves a two-stage transformation consisting of oxidation of ammonium to nitrite to nitrate with each stage being accomplished by separate genera of bacteria [Kaplan, 1983; Najjar, 1992]. This is parameterized using first-order kinetics and is incorporated within the ecosystem equations by forcing the nitrification rate (m in Eqs. A5.2.5 and A5.2.6) to zero above the .2% light level, then phasing the parameterization into the equations using a Michaelis-Menten function with a half-saturation constant corresponding to the .07% light-level. 149 This, in addition to the incorporation of a constant excretion rate (m in Eqs. A5.2.4 and A5.2.7) for the hypersolubilization of sinking particulate m atter to dissolved m atter by attached bacteria [Cho and Azam, 1988], removes the need for a distinct set of equations below an arbitrarily defined aphotic zone depth [Sarmiento e t al., 1993]. For the present simulations, detrital flux has been partitioned into fast and slow sinking particles (v in Eq. A6.2.4). Fast sinking PON is immediately removed from the upper w ater column, emulating the combined effects of zooplankton’ s dense detrital m aterial and diurnal migration. Slow sinking particles have a vertical velocity (wg in Eq. A5.2.7) which is taken to be 12.5 m/day. This partitioning and arbitrary definition for w, are based on the particle flux data obtained from the 200m sediment trap s associated with the BATS program. However, it m ust be remembered th a t sediment trap findings themselves are subject to serious scrutiny due to; 1) problems associated with trap geometry, near trap flow fields and captured versus expected particle composition [McCave, 1975; Staresiiiic et al., 1978] and 2) ill-defined magnitudes of, and vertical gradients in, horizontal advection [Martin et al., 1987; Michaels et al., 1994a]. Defining detrital sinking rates in past interdisciplinary modeling experiments has always been an arbitrary undertaking, spanning 1-100 m/day [Fasham et al., 1990; Fasham et al., 1993], due to the range of values reported in the literature due to seasonal variation [Deuser, 1986; Altabet, 1989] and particle composition [Komar et al., 1981; Michaels and Silver, 1988]. In addition to these remineralization parameterizations, mineralization of DOM is accomplished directly with the bacteria equation (Eq. A5.2.3a) where the DOM and ammonium uptake parameterizations (Eqs. A5.2.3b and 150 A5.2.3c) are based on the assumption th at relative utilization depends on regulation o f gross growth efficiency by substrate C:N ratio [Goldman e t al., 1987]. This portion of the so-called "microbial loop", which obviously consists of a plethora of processes, is interchangeably referred to as secondary or heterotrophic production and can significantly exceed the magnitude of gross primary production since it is a recycling mechanism [Hollibaugh et al., 1980; Harrison, 1992]. Sources of DOM include the previously described hypersolubilization of sinking particulate m aterial, the efficiency of zooplankton feeding [Eppley et al., 1981], plankton exudation [Larsson and Hagstrdm, 1982; Bidigare, 1983] and bacterial cell lysis resulting from viral infection [Fuhrman, 1992]. The extent of the DOM pool is currently a topic of intense in te rest Due to the advent of new sampling methods [Suzuki et al., 1985; Sugimura and Suzuki, 1988], order of magnitude differences in values determined for in-situ DOM concentration presently exist and modeling experiments have revealed that accurately defining these concentrations is im portant for correctly emulating nutrient remineralization within the water column [Nsyjar et al., 1992], Coupling between the physical and ecological systems Coupling between the physical and ecological systems is accomplished in several ways (Fig. 5.2). The dominant mechanisms pertain to the physical field's effect on the ecosystem through water column stratification and vertical mixing. At th is site, the former process is the overriding means by which the physical environment dictates the development of the ecosystem. The latter process is im portant in capturing the vertical redistribution of plankton and 151 Mixed Layer Model (Mellor-Yamada) 58% 42% Eco-System Model (Fasham) Net Heat Flux Momentum Flux Short Wave Insolation Penetrative Component of Heat Flux (dlo/dz) Initial Conditions from BIOWATT (N03, Chi) Initial Conditions from BIOWATT (U, V, T) Mean/Turbulent Heat and Momentum Fields (U, V, T) (q, 1, Km, Kh, Kq, Ps, Pb, E ) Surface Boundary Conditions from BIOWATT (T*, Xy) (Q se , Q la , Q lw , Q aw ) Vertically Redistribute Eco-System Components via Turbulent Diffusion Coefficients (K b ) ( I(z)=Io exp(-Kz)) ( K=Kw<z)+KcChl(z)) PAR Attenuation via Vertical Phytoplankton Distribution F igu re 5.2: C ou pling m ech an ism s w ith in th e in terd iscip lin a ry m odel bacteria as population expansion is mitigated by turbulent diffusion. Also, under suitable meteorological forcing, nutrients may be entrained into the euphoric zone [Klein and Coste, 1984], resulting in an episodic pulse in primary production. This diffusion of passive (i.e., non-motile) tracers is accomplished using the salinity conservation relation (Eq. A5.2.6). This is applied to all ecosystem components save zooplankton and detritus since they are characterized by larger scale vertical motions. Zooplankton have been observed to migrate over one hundred m eters or more over the course of a diurnal period [Longhurst and G., 1988; Longhurst et al., 1989]. Additionally, detrital sinking velocities are such th at vertical motions induced by turbulent diffusion are generally insignificant [McCave, 1975]. Further coupling between the physical and ecological systems occurs via the inclusion of the penetrative component of short wave radiation in the heat equation (Eq. A5.2.5). Time series of short wave insolation gathered by the surface meteorological buoy (Fig. 3.3b), after a 5% loss due to surface albedo, is partitioned into infrared (45%) and visible (55%) components [Dickey and Simpson, 1983] and allowed to penetrate into the water column with separate attenuation schemes. The infrared component dm) is attenuated using the solar zenith angle (£, Eq. A5.2.5b). Since £ is a function of time of day, julian day and latitude [Kirk, 1983], its noon value ranges from 0.174 to 0.866 between summer and winter resulting in 99% attenuation within the upper 1-4 m. Attenuation of downwelling irradiance G o) includes terms for both d ear w ater attenuation and attenuation due to phytoplankton biomass (Eq. A5.2.1h). A profile of K*(z), spectrally integrated over the PAR band, is obtained from an empirical fit (Eq. A5.2.1i) to d ear water irradiance profile data presented in the literature [Baker and Frouin, 1987]. Values for spectrally integrated chlorophyll specific attenuation coefficient ( K c ) are obtained using an empirical relation (Eq. A5.2.1j) taken from the literature [Morel, 1988]. This incorporation of biomass based irradiance attenuation provides for potential feedback from the ecological system to the physical system [Morel and Antoine, 1994]. This is generally a minor effect unless plankton concentrations are extremely elevated and/or water column stratification is weak. Thus, given the correct physical conditions, phytoplankton may have limited control over their environment, providing for the ability to influence their daily light dosage resulting in a series of minor blooms [Stram ska and Dickey, 1994], Application of the interdisciplinary model Applied boundary conditions and model initialization In the numerical experiments presented here, surface forcing containing realistic variance is provided by the Biowatt meteorological data [Dickey et al., 1990]. These consist of the heat and momentum flux time series used as boundary conditions for the physical model (Eqs. A5.2.14). The vertical grid consists of 61 levels in the upper 200m with logarithmic spacing near the surface to provide more precise propagation of downwelling irradiance where the vertical gradient in attenuation is highest. The model is initialized with current and tem perature profiles derived from the moored time series. In addition, chlorophyll concentrations from these data, after transforming to nitrogen units using a constant C:N ratio and predicted C:Chl values (Eq. 154 A5.2.1k), are used to initialize the ecosystem model's phytoplankton profile. An initial nitrate profile is derived from the initial temperature profile and an empirical nitrate-tem perature relation (Appendix 5.5). The remaining ecosystem variables are initialized using constants based on typical values for the region and particular time of year as reported in the literature. In addition, all ecosystem variables which undergo vertical diffusion have a specified deep concentration which serves as a minimum threshold (except in the case of NOs). The physical fields predicted by the model may be validated through comparison with the in-situ physical fields (i.e., distribution of tem perature, currents and derived vertical difiusivities, and tim e series of mixed layer depth and SST) in the data sets obtained by the MVMS packages and the surface buoy. Additionally, several biological fields predicted by the ecosystem model (i.e., distribution of PAR, phytoplankton biomass, gross prim ary productivity and nitrate concentration and time series of 1% light and areal production) may be compared to corresponding in-situ fields obtained from the moored experiment and associated deployment/recovery cruises. Finally, predicted ecosystem characteristics, not m easured during Biowatt, m ay be compared to in-situ data obtained by the BATS/JGOFS sampling program and to other in- situ sampling experiments which have been carried out a t Station S approximately 670 km southeast of Biowatt's location in the Sargasso Sea. 155 Synopsis o f the site’ s physical and. biological regimes The seasonal and episodic features of the physical and biological fields have been presented in the previous chapters and are briefly reviewed here as a refresher. The overall therm al stratification follows a typical seasonal cycle of deep convective mixing and thermodine ventilation in late winter, mixed layer shoaling and the onset of perm anent stratification during the spring, increasing stratification and elevated internal wave variance through the summer followed by the fall deepening driven by decreasing surface heat flux and increasing wind stress. The period of deep convection shows some of the minimal values in areal production (down to 200-260 mgC/m2 /day, Fig. 3.5). The precise tuning of the development of perm anent w ater column stratification is difficult to ascertain in the tem perature map (Fig. 2.11) due to the predominant mesoscale features of the first deployment but the minimum isoline in th e stratification map (Fig. 2.15) and the onset of chlorophyll NSV>1 in the IGW band (Figs 4.9 a-d) show spatio-temporal correspondence. These indicate th a t perm anent stratification propagates into the water column with time, starting around JD 90 down to 75m, corresponding with the spring bloom observed in tim e series of Chi, C c and areal production (Figs. 2.5,2.6 and 3.5), and after JD 150 below 100m. The onset of perm anent stratification, and its continued development into the sum m er (Fig. 2.15), provides the framework for supporting the IGW field observed in the tem perature NSV maps (Figs. 4.7 a-f). These maps indicate th a t maximal variance occurs later in the experiment with increasing depth, sim ilar to the magnitude distribution observed in the stratification map. 156 Increased IGW activity and the concomitant increase in wave breaking, provides a mechanism for increasing turbulent vertical difiusivities. Summertime, in the tropical and subtropical regions, is also characterized by surface conditions (i.e., evaporation>precipitation, [Isemer and Hasse, 1987]) suitable for the development of vertical m iring due to double diffusion. These mixing mechanisms generally manifest within the seasonal thermocline, below the surface mixed layer, providing a source of inorganic nutrients suitable for the development and persistence of the DCM, a dominant feature within the observed Chi distribution (Fig. 2.14). The fall deepening which follows this period is driven by reduced net heat flux and increased momentum flux at the air-sea interface (Figs. 2.1 c, d). This provides a mechanism for greater mixing over the entire w ater column as stratification dissipates, resulting in a second phytoplankton bloom (Figs. 2.14,3.5) triggered by the entrainm ent of nutrients into the upper photic zone. This synopsis of the observed in-situ physical and biological seasonal cycles m ust be viewed as a one-dimensional idealization which precludes the site’ s prevalent mesoscale activity. These advective events m anifest within the moored data as a clearly discernible veering in current direction (Fig. 2.2), distinct tem perature features (Figs. 2.3 and 2.11), and periods of elevated mesoscale band current (Figs. 4.6 a-f) and tem perature (Figs. 4.7 a-f) NSV. Further evidence of their presence appears in SST maps (Figs. 2.8-2.10) and SSH maps (not shown, [Dickey et al., 1993]) which reveal them as being essentially ubiquitous, except during the summer. 157 Observations o f the coupled system based on the locally derived heat budget The advective activity has been estim ated quantitatively by calculating a local heat budget within a surface layer. This is based on surface fluxes from the meteorological time series, water column heat content derived from in-situ tem perature profiles and radiative flux from the base of the surface layer. The inferred quantities from this calculation include vertically and horizontally advected heat flux and upwelling velocity (see Appendix 5.6 for details). The results from this analysis reveal large magnitude values for advected heat flux (~ 100-275 W/m2) during the first deployment through JD 110 w ith the maximum values occurring - JD 100 (Fig. A5.6.2 d). This analysis also shows th at even during the "quiescent" period (JD 130-190) values of advective heat flux range from -100 to 100 W/m2. These values, during what may be considered a 'locally" generated physical regime, are five times higher than those determined in a sim ilar study a t Ocean W eather Station P in the North Pacific [Davis et al., 1981]. These results underscore the problematic nature of attem pting to emulate the region's observed characteristics using a 1-D modeling scheme. The vertical component of heat transport in the heat budget (Fig. A5.6.2 c) indicates the periods of elevated mesoscale activity during the first and second deployments via the associated upwelling velocities. In addition, maximum vertical heat flux occurs between JD 165-175 within a 40 day period characterized by a series of similar, albeit lower magnitude, events. These are indicative of a vertical transport of nutrients and coincide with elevated mesoscale band Chi variability (Fig. 4.9 b-d) corresponding to the relatively 158 large amplitude oscillations in the Chi time series. Indeed, a period of elevated biomass which occurs a t both 62 and 81 m eters in the w ater column (Fig. 2.6) coincides with the onset of the vertical flux maximum ju st described. Since the upwelling velocities associated with these vertical flux events are likely not large in comparison to other upwelling events, the relative prominence of these flux events m ust be due to elevated values of the tem perature gradient (Eq. A5.6.3) and the ongoing development of water column stratification occurring dining this period. This implies the presence of a strong nitrate gradient and injection of nutrients into the photic zone. Comparison o f simulated physical fields with in-situ measurements Obtaining accurate simulations of the upper ocean's physical fields are vital to accurately predicting the dynamics of the pelagic system due to the numerous physical/biological coupling mechanisms. The spring bloom paradigm [Sverdrup, 1953] provides an illustration of coupling (i.e., initiation of the phytoplankton bloom by the shoaling mixed layer) and the necessity for reproducing the therm al structure of the water column. A comparison of mixed layer depth (Z^d) between th at determined from the in-situ data and that predicted by the model shows reasonable agreement except for the first deployment's advective regime (Fig. 5.3). The in-situ Z m id time series for this period shows a series of shoalings and deepenings corresponding to the passage of mesoscale features. The modeled time series indicates the onset of locally driven stratification to occur late in the first deployment (~ JD 122) and to become perm anent coincident with the start of the second deployment (JD 135). After this occurs, the model consistently predicts Zm id to be slightly 159 o I v ‘ S E > _✓ * D V X 2 o o o C M 150 1 0 0 200 250 300 Julian Day 1987 Figure 5.3. Comparison of MLD based on the in-situ measurements (dashed) and the model results (solid). Both use a 0.5 C criterion for estimating MLD. 100 150 200 250 300 Julian Day 1987 Figure 5.4. Comparison of measured (solid) and modeled (dashed) SST. The in-situ data were measured at 3m by a thermistor attached to the surface buoy. 160 shallower (<5m) than th at calculated from the moored data except during advective periods. The timing of the fall deepening is well-reproduced. Comparison of measured and modeled SST also shows reasonable agreement (Fig. 5.4). Again, the worst discrepancies occur during advective events and period of deep convection which occur prior to JD 100. The Zm id time series indicate th at the model distributes heat in the upper ocean differently than is observed in the data set. Comparison of the contoured tem perature maps (Figs. 2.11 and 5.5) provides further evidence of th is and it can be seen th at the model traps too much heat near the ocean surface. The timing of maximum near-surface tem perature (i.e., the 27s isotherm) is well-reproduced. However, both its spatial and temporal extent are significantly reduced in the model. The surface manifestation of the 24s isotherm shows a much stronger temporal correspondence but its maximum simulated depth is 40m while in the moored data it extends beyond 70m. Excessive trapping of heat near the surface in the simulations is also indicated when comparing the stratification maps (Figs. 2.15 and 5.6). Indeed, the simulated field shows magnitudes which reach levels twice those observed in- situ. However, it is of interest to note th at in a general quantitative sense, the structure of the 0.12 'C/m simulated isoline is sim ilar to the .025 ® C /m observed isoline in both the timing of the surface representation and the development of a subsurface, constant depth, feature. The in-situ map shows th is feature to be a subsurface maximum. Maps of current shear also show definite dissimilarities between the in- situ and simulated cases (Figs. 2.16 and 5.7). Indeed, the contrast between 161 so 60 60 160 SOO 220 240 260 260 300 100 120 160 Julian Day Figure 5.5. Contour map of seasonal temperature cycle predicted by physical model. Isotherms are in degrees C. 2 0 40 a 60 60 140 180 200 220 240 260 260 300 320 100 120 160 Julian Day Figure 5.6. Contour map of seasonal stratification cycle predicted by physical model. Isolines are in degrees C/m. 162 0 .0 0 6 SO o.oos • 0 . 0 0 * ' 0.001 'o . 6 0 8 0 140 160 SOO SSO S40 2 60 200 100 160 Julian Day Figure 5.7. Contour map of seasonal cycle in current shear predicted by physical model. Isolines have units of 1/sec. 163 them is striking. Maximum values correspond well between the two cases. However, the temporal evolution differs significantly. The in-situ case shows maximal near surface values during periods of elevated advection whereas simulations show a seasonal progression sim ilar to the stratification with increasing values associated with surface wanning. Additionally, the moored data show values of interior shear which are generally twice as high. This provides an explanation for the insufficient transport of heat into the interior, thus reinforcing the observed deficiencies in the simulated thermal fields. Difficulty with simulating observed shear distributions has been previously reported in mixed layer model applications [Halpem et al„ 1995]. Comparison o f simulated ecosystem fields with in-situ measurements The interdisciplinary simulations provide predictions for both the ecosystem param eters within the model compartments (Fig. 5.1) and derived characteristics based on these predictions, irradiance conditions and vertical transport processes. These fields are compared to measurements from the moored experiment and for cases where comparable data are not available from Biowatt, time series data taken during the BATS/JGOFS sampling program are utilized [Knap et al., 1992; Knap et al., 1993]. Comparison between in-situ measurements and the sim ulated fields provides the means to ascertain how well the model predicts the ecosystem's evolution and to assess how accurately trophic interactions are parameterized. Temporal evolution of phytoplankton biomass in the model simulations (Fig. 5.8a) shows characteristics which are generally sim ilar to oceanic 164 so s I Q 60 1 1 0 0 130 160 180 300 330 340 360 380 500 530 140 Julian Day Figure 5.8a. Contour map of chlorophyll concentration as predicted by the coupled interdisciplinary model. The isolines have units of mg Chl/m3. 30 40 60 O “ a 80 a § . 1 0 0 Q 130 140 0 .05 160 180 300 460 480 S00 530 540 560 580 600 630 640 660 690 Julian Day Figure 5.8b. Contour map of chlorophyll concentration from 1990 BATS data. The time line covers the same calendar days as Fig. 5.8a but was shifted to include two sampling years (BATS 2 and 3). Data down to 200m are shown to illustrate the DCM. 1 6 5 observations (Fig. 2.15). These include the development of a spring bloom with the onset of perm anent stratification and the evolution of a DCM as surface nutrients become limiting. However, the simulation predicts biomass concentrations which are four times higher during the surface bloom than those observed in the time series data. In addition, the observed manifestation of the DCM occurs deeper in the w ater column (80-100m as opposed to 30- 40m) with concentrations which are twice as high. Biomass concentrations from the BATS surveys also indicate the presence of the DCM to be around 100m (Fig. 5.8b). The discrepancy between these characteristics of biomass concentration is verified by comparing time series of daily water colum n production (Fig. 5.9). These show th at maximum estim ated water column production is twice as high in the simulations and is generally elevated in comparison to the in-situ estimates from the triggering of the spring bloom through the early fall (JD 260). However, the timing of the simulated bloom corresponds precisely with that observed in-situ though the contour m aps of gross production (Figs. 3.4a and 5.10) show th at the areal value is derived from different depths. The phytoplankton bloom appears to dissipate due to a reduction in available nutrients and losses due to mortality and exudation. It m ust also be mitigated by heterotrophic grazing to some degree but th is is not necessarily the dominant mechanism, a t least initially. The latter observation is based on the map of zooplankton concentration which indicates a lag of twenty days between the phytoplankton bloom and the maximal zooplankton values (Fig. 5.11). These reach concentrations of 0.5 mMolN/m3 and greater which extend down to 30m for about a month (JD 140-178). 166 u <J o 200 250 100 150 300 Julian Day 1987 Figure 5.9. Comparison of areal production calculated using the Kiefer model with the in-situ data (dark) and as predicted by the interdisciplinary model (light). Estimates of annual areal production are 143 and 190 gC/m2/yr respectively. SO a 60 60 1 00 1 8 0 140 160 160 800 Julian Day 860 880 500 Figure 5.10. Contour map of gross primary production as predicted by the interdisciplinary model. Isolines are in mgC/m2/day. 167 D epth (m) 80 40 60 B O 1 00 180 0 .05 140 160 180 180 8 0 0 880 Julian Day 8 8 0 1 00 180 140 160 8 4 0 860 2 8 0 Figure 5.11. Contour map of zooplankton concentration as predicted by the interdisciplinary model. The isolines are in mMol N/m3. 168 5.12 a .00 80 50 /—s S 40 A 50 a « 0 ) Q 60 70 90 1 100 180 140 160 180 800 880 840 860 860 500 Julian Day 50 a 4 0 . 0 0 ' 3 a a > Q 6 0 • 0 0 . 80 90 1 100 180 140 160 180 800 880 840 860 88Q 500 5.12 b Julian Day Figure 5.12. Contour maps of modeled (a) and measured (b) PAR. The isolines are in Ein/m2 /day. 169 These production maps both indicate a near-surface increase after JD 220 due to an increase in PAR which appears both in the simulations, where vertically propagating PAR was obtained by partitioning the surface time series of short wave insolation (Fig. 5.12a) and in-situ (Fig. 5.12b). Indeed, the correspondences which may be observed between the in-situ and modeled time series of areal production (Fig. 5.9) are derived from fluctuations in solar irradiance. The light field is further quantified by calculating the 1% light depth which shows a good correspondence between the simulations and the in-situ data (Fig. 5.13). The contour map of nitrate shows a typical seasonal evolution whereby the spring bloom essentially exhausts the surface supply and minimal concentrations slowly extend into the water column (Fig. 5.14a). These features mimic those of the nitrate map formed from the Biowatt tem perature time series (Fig. 5.14b) using the empirical nitrate-tem perature relation (Appendix 5.5). Further comparison reveals th at in the simulation the 0.1 pM isoline has a greater temporal extent, slightly lower extension in depth and deep values which are 3-4 times higher. The contour map of simulated ammonium shows the development of a subsurface maximum toward the end of the summer which extends through the end of the simulation while increasing in magnitude to 0.3 pM at 100m (Fig. 5.14c). The f-ratio, based on uptake of inorganic nutrients (i.e., Q1AQ1+Q2) from eq. A5.2.1c), shows values S 0.24 in the upper 15m following the spring bloom (Fig. 5.14d). Additionally, values less than 0.5, indicative of a balance between new and regenerated production, extend to 40m following the bloom. Processes associated with the microbial loop are represented in the contour maps of DON, bacteria and detritus. The simulated fields are strongly 170 o o m ~ o o. o 4> «- o 2 : _1 o m o o C M 200 1 0 0 150 250 J00 Julian Day 1987 Figure 5.13. Comparison of 1% light depth between estimates based on in-situ PAR measurements (dashed) and the irradiance field as predicted by the interdisciplinary model (solid). 171 Q i < V 40 6 0 6 0 1 1 0 0 120 140 160 1 8 0 2 0 0 2 2 0 2 4 0 2 6 0 2 8 0 5 0 0 5 20 5.14 a Julian Day 2 0 Q 6 0 8 0 1 0 0 140 160 16 0 2 0 0 2 2 0 2 4 0 2 6 0 2 8 0 5 0 0 5 20 5.14 b Julian Day 172 so 0.100 40 O .tS O 60 •O.SSO 6 0 ^ 100 O < 0 ) Q ISO ■ 3 S 0 . ■ 6 0 0 . •»S 0 140 0 .1 0 0 160 180 1S0 140 1 60 180 SOO SSO 8 4 0 8 6 0 S 8 0 3 0 0 3S0 100 Julian Day ,o . e « ' SO 0 .4 0 40 a 0 ) Q 80 1 0 0 100 150 SSO SOO 5 0 0 Julian Day Figure 5.14. Model (a) and in-situ (b) nitrate distributions (pM N). In-situ nitrate is estimated using in-situ temperature and an empirical relation (Appendix 5.5). Ammonium distribution (pM N) predicted by the model (c) is used, in conjunction with modeled nitrate, to calculate f-ratio (d). 173 coupled to the distribution of phytoplankton biomass (Fig. 5.8a). DON shows the strongest resemblance w ith maximal values te 0.1 pM) extending down to 23m around JD 130 while values 2 0.5 pM consistently reach 40m for a two m onth period (Fig. 5.15). Following these elevated values corresponding to the phytoplankton bloom, DON concentrations slowly decrease in th e near surface w aters but do not drop to background levels above 40m by the end of the simulation. There is presently no regional in-situ d ata with which these results m ay be compared. The bacteria map (Fig. 5.16a) also reveals maximal values (2 0.48 mMol N/m3 ) near th e surface, extending to 30m. These do not correspond to the DON feature but are temporally associated w ith the maximum in zooplankton biomass (Fig. 5.11). Following the rapid development of the near surface maximum, the concentration slowly declines through the summer and fall. Additionally, there is a consistent gradient w ith depth whereby concentrations are monotonically decreasing. This contrasts the observed spatio-temporal distribution which shows relatively constant values over time and a subsurface maximum which generally ranges from 40 to 80 m eters except around JD 470 where maximal values have a surface expression (Fig. 5.16b). The simulated concentration values are representative of those in th e observations although the maximal values are twice as high as observed. The distribution of detritus or PON in the simulation results (Fig. 5.17a) reveals a near surface maxima which is temporally associated w ith that observed in the zooplankton and bacteria (Figs. 5.11 and 5.16a). The maximum predicted concentrations (5 0.5 mMol N/m3) show good agreement w ith the observed concentrations (Fig. 5.17b). However, the observations 174 D epth (m) 2 0 40 60 80 1 180 2 0 0 220 120 140 2 4 0 260 2 8 0 3 2 0 Julian Day Figure 5.15. Distribution of DON predicted by the interdisciplinary model. Isolines are in pM N. 175 80 40 60 /—s a 90 140 160 190 SOI 100 180 140 160 160 800 880 840 860 890 500 580 Julian Day 80 40 60 0. 6 0 5 100 ft V Q 120 >0.08' M O 180 200 460 480 500 520 540 S60 580 600 620 640 660 680 Julian Day Figure 5.16. Model (a) and in-situ (b) distribution of bacteria (mMol N/m3). The in-situ data were acquired during BATS 2 and 3 and are from 1990. The time line represents the same calendar days but is adjusted to account for the consecutive acquisition years needed to show corresponding time periods. 176 Depth (m) a 0 .8 0 S2* O' 0 .4 0 c n < 1 < ^ 3 ^ 3 a- Julian Day Depth (m) S0*0 o < 1 0 5 5.17 c (0 o 1 C V | g • * . r o | k Q ) •• 4 ; § o 1 0 0 150 200 250 300 Julian Day Figure 5.17. Model (a) and in-situ (b) distributions of PON (mMol N/m3). The in-situ data was obtained during BATS 2 and 3. The time line for the BATS data represents the same calendar days but is adjusted to account for the two data acquisition years required to generate the correct time period for comparison with the model result. The time series of particle flux at 200m is shown with points representing sediment trap data (*) from the BATS site superimposed (c). These data from the BATS site were were used to adjust the detrital component of the model was partitioned into quickly and slowly sinking particles. 178 reveal the consistent presence of particulate m atter in the water column while the simulations indicate concentrations of this magnitude to be episodic and tend to dissipate over the seasonal cycle. The tim e series of sim ulated particle flux a t 200m shows reasonable agreement with the observations from the sediment trap data outside of the initial elevated values (-0.6 mMol N /m3 ) associated with the deep convective period (Fig. 5.17c). The magnitude of the partide flux associated with the phytoplankton bloom and the following increase in zooplankton and bacteria is within the range indicated by the in- situ data. Additionally, the monotonically decreasing flux values indicated in the simulation are supported by the trap data. Discussion The results of the interdisciplinary model presented here show a number of promising features but there are also a number of inconsistencies upon comparison with the in-situ measurements. These m ust be addressed before applying the model as a tool to investigate the pelagic system’ s response to the incorporation of varied fordng mechanisms either internal or external to the oceanic boundary layer. Some of the inconsistendes may be attributed to the physical model’ s inability to simulate three dimensional processes which dominate the springtime observations and are also pronounced during the fall period. The summertime simulations show the best correspondence to the in- situ observations due to a lack of mesoscale influences, except for an isolated feature, and the fact th at the ecosystem has transitioned into the post-bloom, regenerated production regime. 179 Prominent discrepancies between the simulations and the observations may be seen when comparing the time series of areal production. These reveal observed production levels prior to the spring bloom to be 2-3 times greater than those predicted by the simulations (Fig. 5.9). However, with the onset of the spring bloom, which shows a precise temporal correspondence to the observed feature, simulated levels of areal production attain values which are nearly twice as high as the in-situ estimates. The low primary production values prior to the bloom are due to the lack of stratification achieved in the modeled physical field (i.e., the vertical mixing length is CKlOOm)) whereas the in-situ tem perature data show episodic stratification due to the passage of mesoscale features (Fig. 2.5). Stratification reduces the magnitude of the vertical excursions experienced by the phytoplankton, providing them with sufficient daily integrated irradiance to achieve net growth. This in turn lowers the near surface concentrations of nitrate, reducing the magnitude of the spring bloom when w ater column stratification becomes permanently established. Production levels following the bloom remain higher than the in- situ estim ates until the beginning of the fall period (JD 250). From th at time until the end of the experiment, both the magnitudes and temporal variability show good correspondence between the two time series, outside of the in-situ fall bloom. The strong temporal correspondence is a result of both production time series being based on similarly varying light fields and may be observed in the summertime values as well. The magnitude of the simulated production during the summer is highly sensitive to the bacterial ammonium excretion rate and the bloom's magnitude as this represents total phytoplankton biomass. Following the bloom, the 180 simulated tim e series does not show significant deviations in magnitude from the observations even during periods of mesoscale activity. This indicates that once quasi-steady state production is achieved in the oligotrophic environment, discriminating between local production regimes and those within advective features is somewhat difficult. Discrepancies between production estim ates are emphasized through comparison of the maps of gross production (Pigs. 3.4a and 5.10). These show simulation estim ates to be monotonically decreasing from the surface while in-situ estim ates consist of a subsurface feature related to the prominent DCM featured in the time series (Fig. 2.14). The model also indicates such a feature (Fig. 5.8a) but the difference in magnitude between surface Chi and Chi a t depth is insufficient to m anifest as a subsurface production maximum due to the strong gradient characteristic of w ater column irradiance (Fig. 5.12a). The lack of a prominent DCM in the simulations is a concern and will certainly be an area of future model refinement. Such a feature is apparent in both the Biowatt (Fig. 3.17) and the BATS (Fig. 5.8b) data sets and capturing it within the simulations is a necessity. Comparison of the PAR field as m easured by the MVMS packages and as it appears in the model (Figs. 5.12 a, b) reveals the modeled light field to be about half as high a t a given depth as the observed field during the second deployment (i.e., the summer months) while during the first deployment (i.e., early spring to early summer) the relative rankings of the two cases are reversed to almost the same degree. Comparison of the Qw tim e series w ith the theoretical daily maximum, determined from orbital mechanics and assuming no cloud cover [Kirk, 1983], show reasonable agreement during the first and third deployments but 181 measured values from the second deployment are consistently 10-20% too low (Fig. 5.18). Thus, summertime irradiance levels at depth may be insufficient for the development of a DCM. Additionally, the slightly elevated simulated irradiance during the first and third deployments may be due, at least in part, to the usage of a constant value for surface reflectance (a). This should, in fact, be a function of the surface wave field which is determined by the wind stress. Average values of surface wind speed for the three deployments are 7.8,5.8 and 6.4 m/s, respectively [Dickey et al., 1990]. Furthermore, simulated near surface nitrate concentrations are higher than those estimated with the nitrate-tem perature relation (Figs. 5.14 a, b), resulting in estimates of the f-ratio (Fig. 5.14d) which are higher than values generally associated with regenerated production regimes [Eppley and Peterson, 1979; Platt, 1984]. W ithout growth limiting near surface nutrient concentrations, the development of a DCM is further inhibited and increased irradiance during the summer months becomes counterproductive since the model’ s value for annual areal production is already 33% higher than the in-situ estimate. The elevated nitrate values may be due to forcing the interior diflusivities within the physical model to too high a level. This prospect is currently under investigation using in-situ tim e series and coincident spatial data from a site in the North Atlantic [Jones et al., 1995] where it has been shown th at phytoplankton growth is extremely sensitive to the formation and decay of daily stratification [Stramska and Dickey, 1994], Indeed, the physical environment in this North Atlantic system is characterized by relatively weak water column stratification, when compared to the Sargasso Sea. This provides for a more pronounced reaction to changes in the parameterized 182 100 150 200 250 300 Julian D ay 1987 Figure 5.18. Comparison between short wave insolation time series acquired by the meteorological package and the empirical formulation developed by Lamb which predicts the maximum sea surface value expected for a given day and position (Kirk, 1983). This indicates that the insolation time series is -10% below the expected values during the summer months. This time series is used by the interdisciplinary model to generate the in-situ irradiance field. 183 interior diffusivities and an opportunity for determining how to best define this param eterization within the interdisciplinary code. A number of such param eterizations have been suggested in recent years [Peters et al., 1988; Mellor, 1989; Kantha and Clayson, 1994; Large et al., 1994]. Except for a temporal displacement of about 20 days, the simulation's heterotrophic compartments (Figs. 5.11 and 5.16a) correspond strongly to the characteristics of the phytoplankton distribution (Fig. 5.8a). This delayed, yet strong, correspondence between autotrophs and heterotrophs is also represented in the BATS data between the Chi and bacteria distributions (Fig. 5.8b and 5.16b). Additionally, the simulations indicate a direct correspondence, both spatial and temporal, between concentrations of Chi and detritus (Figs. 5.8a and 5.17a), which is reflected as the maxima at JD 145 in the particle flux time series (Fig. 5.17c). Again, the modeling results provide a reasonable emulation of in-situ measurements in th at these two param eters track one another directly. However, the relationship between PON and Chi in the BATS data also suggests a 3-4 fold variation in C:Chl which creates an apparent discrepancy between the two fields. This is not reproduced in the simulations, although light varying C:Chl is included in the autotrophic equation set (Eq. A4.2.1f). In-situ bacteria (Fig. 5.16b), which are unaffected by photoadaptation or spedation, show a more direct correspondence w ith PON. The observed correspondence between elements of the pelagic system is not particularly noteworthy but has been presented in order to illustrate the model’ s ability to reproduce these interrelationships and to emphasize the need for accurately emulating the seasonal evolution of the lowest trophic level. 184 In conclusion, the interdisciplinary model presented here exhibits a num ber of positive features. These include the ability to accurately predict the onset of perm anent stratification which triggers the spring phytoplankton bloom and the transition to a regime of regenerated production for which the temporal variability shows decent correspondence to th at of the in-situ areal estim ates. Additionally, interactions between trophic compartments are also representative of observed relationships although presently, the model does not provide an acceptable simulation of the observed evolution of phytoplankton biomass. Some possible means of improving the prediction of this field have been presented (e.g., reducing interior mixing to reduce surface nitrate concentrations and more accurately reproducing the irradiance field) and will be addressed in order to enhance the model such th at it may be utilized as a tool for studying ecosystem response to various superimposed physical stimuli. Finally, several model runs not presented here, indicated the existence of a numerical instability related to the DON param eterization a t elevated concentrations. Given the present controversy over the role and magnitude of dissolved nutrients [Sugimura and Suzuki, 1988; Najjar et al., 1992], discerning the mechanism by which such oscillatory behavior is introduced may be useful in advancing the understanding of this portion of the pelagic system. 185 C onclusion Advancing the understanding of the ocean's contributions to, and interrelations with, the global climate system requires an interdisciplinary approach when developing both in-situ sampling methods and modeling schemes. Applying the observations obtained from in-situ platforms is a necessity for aiding in the development of interdisciplinary schemes and has been the focus of this work. Indeed, the in-situ measurements clearly emphasize the complexities associated with individual components of the physical and pelagic systems and the interactions which occur between them. The in-situ data set has been extremely useful for providing the thermal and current structure a t the site. These reveal the highly advective nature of the region during certain time periods. These advective events have been further quantified with synoptic maps and through the calculation Of a one dimensional heat budget. Spectral analysis results show a correspondence between mesoscale events in the physical fields and mesoscale variability in the bio-optical signals indicating the transport of distinct ecosystems within these mesoscale features. The spectral results have also shown the seasonal evolution of variance within the internal wave field in the tem perature data. Corresponding variability has also been observed in the bio-optical fields, indicating vertical oscillation of a biomass gradient. Additionally, episodic bloom events are observed to coincide w ith this internal wave activity indicating vertical transport of nutrients. Estim ates of gross production have been made which serve to quantify production within the region. In addition, the productivity rates obtained from the optical production model are used, in conjunction with loss rates estimated from the C c time series, in order to estim ate zooplankton grazing rates. The observed physical and ecological fields are used to provide a framework for judging the interdisciplinary model's performance. The model scheme developed here contains provisions for several direct couplings between the physical and biological fields. Thus, interactions and feedback mechanisms are included. Presently, the results of the model show reasonable qualitative agreement with areal production but show definite discrepancies when comparing the spatial distributions of the ecosystem characteristics. The root of these inconsistencies is the distribution of phytoplankton biomass. Thus, the initial focus of future work will be to improve the model's ability to recreate this field. Additionally, interest in refining trophic interaction param eterizations and studying ecosystem response to varied physical forcing will be pursued with the 1-D scheme. 187 Anm>ndi» 4.1; D eterm ination of significant coherence The maximum number of available ensembles was 9,15 and 11 for deployments 1,2 and 3 respectively. The number of available ensembles decreased when time series were shortened by instrum ent failure or bio-fouling and are recorded in Tables 4.2 a-e. The following formulas (Eqs. A4.1-3) were used to estim ate the threshold level of coherence (ft) for a range of confidence intervals (Cl) [Bloomfield, 1976]. It should be noted th at the degrees of freedom include band averaging (minimum of 3 in this case) as well as the number of ensembles. Coherence values below ft (Eq. A4.1) are not considered significant. Threshold coherence as a function of the num ber of ensembles for a range of confidence intervals is shown (Fig. A4.1.1). gk = 2/na nd a # of degrees of freedom p s2 m /n n m points in ensemble m a points within tapering window Yt=l-(1 - 01/100)1-8* Eq. A4.1 Eq. A4.2 U 22= ( l - f ) 2 Eq. A4.3 188 1 0.8 S o 2 0.6 j i H o > u a 0 ) £ 0.4 J i o O 0.2 0 0 5 10 15 20 25 30 E n s e m b le s F ig u re A4.1.1: S ig n ific a n t c o h e re n c e th re s h o ld . Coherence threshold is determ ined for a given confidence interval as a function of the number of ensem bles. j 1 1 1 .............................. .................. ♦ o • 95% Cl X x 90% Cl + ■ A 0 + 85% Cl ■ X • o * 80% Cl ♦ + A X O • 70% Cl - “ + V ■ 75% Cl ♦ . + • a ♦ ■ O ♦ 65% Cl • ♦ x + O ■ x ^ t t X 0 1 i X o 4- X 1 1 . . . . 1 . . . . 1 . . . . 1 . . . . 1 . i i : 1 1 1 1 1 1 ■ 1 1 , Annendlx 4. 2: Presentation of spectral analysis results S pa tia l I Temporal distribution o f norm alized spectral density H orizontal Current Speed G n( / m> Z) for the horizontal current speed (Fig. A4.2.1 a) decreases monotonically with depth during the first and third deployments (Figs. A4.2.1 b, d). The second deployment map (Fig. A4.2.1 c) is consistent with the surrounding periods but has an additional localized maxima a t 75m near / d- The temporal maps (i.e., G n( / m» t)) reveal that the first deployment contains the bulk of the variance over all frequencies (Fig. 4.6). Above average G n( / m» Z) is present from JD 60-125 a t all depths. In addition, the first deployment's dominance in the mesoscale band becomes more prominent w ith depth, peaking a t roughly 60m. All temporal maps reveal a mesoscale event temporally centered around JD 200 although its expression is much stronger near the surface. Prior to the passage of this feature, the second deployment is extremely quiescent. Following this event, mesoscale activity increases and continues into the following deployment. The third deployment is subject to multiple advective features which appear as elevated Gn( / m> t) towards the end of the experiment. However, due to the magnitude of first deployment Gn( / m» t), the 0.8 isoline is used to indicate mesoscale processes during this period. The 14 and 23 m eter maps (Figs. 4.6 a,b) indicate mesoscale activity starting w ith the passage of the mesoscale feature at JD 200 through JD 275. The 62m m ap 190 D e p t h ( m ) D e p t h (■ ) Esp • 0 TS Log Rnquency 1 *.tt- S O TS 1 0 0 Log Frequency cs so T S 100 •s o 0 0 h- Figure A4.2.1. GN(f,Z) for the current field. Map a shows average distribution over the entire experiment. J 2 Maps b-d show average distribution over each of the three deployments. (Fig. 4.6 c) shows elevated Gn(/m» t) from JD 280 until the end of the experiment while the 160m map (Fig. 4.6 0 shows elevated Gn( / m. t) ranging from JD 230-265 with a maximum centered on JD 250. All three deployment snapshot maps reveal a feature centered around f \. This is characterized by a two to three fold decrease in GxxCfl. Z) between the surface and 130m (Figs. A4.2.1 b-d). The background map (Fig. A4.2.1a) shows the surface Gn( / i, Z) maximum extending into the low / igw and a subsurface maximum in G n( / igw, Z). This occurs between 25 and 50 m eters in the first and third deployment maps (Figs. A4.2.1 b, d). This is contrasted by the second deployment map which (Fig. A4.2.1 c) shows Gn( / igw> Z) to be monotonically decreasing through 80m. The temporal maps indicate th at these events a t f \an d in / igw coincide w ith periods of elevated G n( / m» t). Temperature The background map of Gn( /, Z) for tem perature (Fig. A4.2.2 a) shows a maximum in the mesoscale band which ranges from 60-80m. This feature is a product of the subsurface maxima seen in deployments 2 and 3 (Figs. A4.2.2 c, d). G n( / m. Z) in the first deployment snapshot decreases monotonically with depth (Fig. A4.2.2 b). Maps of G n( / m» t) from 14 and 23 m eters show dominant mesoscale activity in the first deployment (Figs. 4.7 a, b). They also indicate a mesoscale feature which occurs around JD 200. Below 23m, the dominant feature in the mesoscale band temporal maps is this latter feature (Fig. 2.11 d,e). Its signature dominates the 62 and 81 meter maps, leading to the subsurface maximum in the second deployment snapshot (Fig. A4.2.2 c). 192 Depth!*) D ep th (ffl) 1 1 I I Exp :i? Log Frequency •o. cs so r s Log Frequency Log Frequency 1 Log Frequency H - * Figure A4.2.2. GN(f,Z) for the temperature field. Map a shows average distribution over the entire experiment, w Maps b-d show average distribution over each of the three deployments. The most prominent IGW feature in the background map is the subsurface maximum (Fig. A4.2.2 a). At lower frequencies of the IGW band this feature is centered a t 35m and extends down to 90m. At higher frequencies, it is still centered at 35m but extends only to 70m. In addition, there is a greater range over depth in G n CTi g w , Z) a t higher frequencies than at near-inertial frequencies. All three deployment snapshots (Figs. A4.2.2 b-d) have subsurface maxima encompassing / i g w * The maxima's centers occur between 25 and 45 meters and are deeper with each succeeding deployment. fsvit) has been plotted next to each depth’ s contour map of G n ( / , t) for the tem perature data (Fig. 4.7) in order to illustrate the potential for supporting an internal wave field. There is a dose temporal correspondence between maxima in fBvit) and maxima in G n ( / i g w , t). This is especially noticeable in the 14 and 23 m eter maps which show maximum G n ( / i g w > t) corresponding to elevated / b v around JD 190-195. The 14m map shows the onset of elevated G n ( / i g w » t) after JD 100 corresponding to a local maxima in / b v due to the establishm ent of perm anent stratification. Since these maps show normalized values, inferences drawn when comparing temporal maps over depth should be made with care. However, elevated G n ( / i g w ) occurs later in time with increasing depth. The 14 and 23 m eter maps both show elevated G n ( / i g w . t) during a time period which corresponds to mixed layer depths of 15-20m. This surface layer begins to deepen following the passage of the mesoscale feature around JD 200 and coinciding decreases in both G n ( / i g w , t) and / b v (t) are present The deeper 194 maps reveal elevated Gn( / igw> t) following the passage of the mesoscale feature a t JD 200. Beam attenuation coefficient (c j The background and deployment maps of Gs(f, Z) for C c all show G n( / mi Z) to be monotonically decreasing with depth (not shown). Only two depths, 101 and 160 meters, have temporal maps covering the entire experiment. These have been supplemented with temporal maps from 23,62 and 81 meters. The 23m map covers the first two deployments while the 62 and 81 m eter m aps cover the last two deployments. The 23,101 and 160 m eter maps (Figs. 4.8 a,d,e) all show maximal Gn( / m » t) during the first deployment. The 23 and 101 m eter maps show elevated values throughout the first deployment while a t 160m they last through JD 95. Mesoscale features are also apparent a t 62 and 81 m eters for time periods ranging from JD 232 to 265 and JD 225 to JD 250 respectively. The temporal maps from 101 and 160 m eters (Figs. 4.8 d,e) both show elevated G n( / igw» t) throughout the first deployment. The 101m maximum encompasses / igw while the 160m maximum includes all but the highest frequencies, / igw in the 23m map starts later in the first deployment (JD 90) and persists into the second deployment, through JD 140. Another interesting feature in several of the temporal maps occurs during the second deployment in the high frequency IGW band. This appears a t 23,81 and 160 meters and generally lasts from JD 180 to JD 225. Corresponding features do not appear in the m aps from 62 and 101 m eters because of elevated G n( / igw> t) which 195 appear during different periods. At 101m this occurs during the first deployment and has already been described. At 62m the dominant IGW event encompasses / ig w and occurs from JD 230 to JD 266. This feature corresponds temporally to elevated Gn( / m* D previously described a t 62 and 81 m eters. Chlorophyll-Fluorescence (Chi) The 23m temporal m ap covers the first deployment and serves to illustrate the fact th at the pigment field does not precisely follow the Current field. This is shown by the fact that, though elevated GW/m. t) occurs early in the deployment, elevated Gn( / igw, t) does not occur until later in the deployment after the onset of stratification (Fig. 4.9a). There are only two depths (62 and 81 meters) for which the usable Chi data were sufficient for the creation of full experiment temporal maps (Figs. 4.9 b,c). Both show elevated Gn(/m» t) beginning in the first deployment albeit w ith slightly different temporal bounds. The 62m map shows elevated Gn(/m. t) throughout the first deployment which continues through JD 150 in the second deployment while the 81m mesoscale feature extends from JD 110 to JD 190. Finally, the 101m map reveals a sim ilar feature with elevated Gn(/, t) occurring across all bands beginning a t the sta rt of the second deployment (Fig. 4.9d). It is interesting to note th a t these features with similar characteristics initiate later in time, deeper in the w ater column. F irst deployment Gn( / igw> Z) is monotonically decreasing with depth (figure not shown), consistent with maximal near-surface Chi a t this time. The 196 temporal m aps from 23 and 62 meters reveal the tim ing of elevated G n ( / i g w » t) (Fig. 4.9a, b). They show elevated values (3 to 9 tim es the deployment average for the 23m case) starting after JD 90 at 23m and after JD 96 a t 62m. In addition, the 81m map (Fig. 4.9c) shows the onset of elevated G n ( / i g w , t) following JD 120. Both the 62 and 81 m eter Chi maps also show maxima within / i g w around JD 230. In the 81m map, elevated G n ( / i g w » t) starts a t the higher frequencies before expanding to encompass all of / i g w and before elevated G n ( / i g w > t) appears in the 62m map. According to the G n ( / i g w , t) values a t these two depths, the impact of this event was much greater a t 62m. 197 Spatial t Temporal distribution of coherence Coherence between oceanic currents and temperature The first deployment shows coherence CfoCfif > Z)) which increases below 60m in the zonal case and which increases with depth for the meridional case (Figs. 4.10a, b). The temporal m aps from 14 and 23 meters for the zonal case both show elevated JxyifUf t) from JD 70 to 80 corresponding to a period of five days in the mesoscale band (Figs. 4.10 c,d). This feature loses prominence and duration with depth, becoming statistically insignificant below 101m. Both current directions show maximal Z) a t 160m. The zonal case shows a feature occurring from JD 90-105 while the meridional case reveals elevated coherence from JD 90-115 (Figs. 4.10 e, f). The second deployment maps show a sub-surface maximum in zonal coherence and monotonically decreasing coherence with depth for the meridional case (Figs. A4.2.3 a,b). The zonal case’ s subsurface maximum occurs between 70 and 80 m. According to the 81 m temporal map, this occurs after JD 190 and continues through the end of the deployment (Fig. A4.2.3 c). The meridional case shows two significant events in Y x y C fM . t). The first appears a t most depths but has statistically significant manifestations at 14, 62 and 81 meters and lasts from JD 150 to 170 (Figs. A4.2.3 d-f). The second ranges from JD 192 to 202, is present a t all depths, and is statistically significant at 14,62 and 81 m (Figs. A4.2.3 d-f). This feature corresponds to the mesoscale event most prominently observed in the Gn( / m> t) m aps for tem perature (Fig. 4.7). 198 661 Julian Day 1 9 8 7 Depth (n) •> m 0 .4 0 . f I I < 3 •pro- •pro* •o.tp. > s ! ■o.tp. •90'O ' > Julian Day 1 9 8 7 Depth (n) ; : © a r 1 s ■O.tp. t S 1 •0 .2 0 . •0 . 1#. •O.tp- •O.U. Ju lia n D a y 1 9 8 7 \ /OfV! 0 i i i i £ i 3 A4.2.3 g L o g Frequency Figure A4.2.3. y(f) between individual current vectors and temperature. Maps a and b show y(f,Z) for the second deployment. Map c (81m) shows 7(f,t) with zonal current. Maps d-g (14, 62, 81 and 23 meters) show 7(f,t) with § meridional current. In the third deployment, the zonal case has a subsurface minimum at 60m while the meridional case shows a subsurface maximum ranging from 60 to 100m (Figs. A4.2.4 a, b). The zonal temporal maps show inconsistent distributions of significant Y yCfM . t) between depths. The 81m (Fig. A4.2.4d), 101m and 120m zonal maps (not shown) all show a significant mesoscale signature lasting from JD 285 to JD 300. The temporal/frequency distribution of significant Y x y C fM . t) a t 14 and 43 meters differs from what is observed below 81m. The Y iy C fM . t) maps for the meridional case also indicate a distinction between two sections of the water column. The 23,43 and 62 meter m aps show mesoscale signatures which range from JD 295-310 (e.g., Fig. A4.2.4c) while, sim ilar to the zonal case, the 81,101 and 120 m eter maps contain mesoscale signatures which range from JD 285-300 (e.g., Fig. A4.2.4e). An increase in Y x y C f * t) at 50 m at / i appears in both the second and third deployment maps (Figs. A4.2.3 and A4.2.4). It has already been seen in the autospectra th at the current data contain an inertial signal throughout the w ater column (Figure 4.1). Therefore, the inertial band feature in these coherence maps corresponds to the inertial signal's strengthened presence in sub-mixed layer tem perature data (Figs. 4.2d and 4.2f). The first deployment coherence maps do not exhibit this feature due to low/intermittent stratification during the late winter period (Fig. 4.10). This is reflected in the lack of an inertial peak in the corresponding auto-spectra (Figs. 4.2 a, b). 201 Depth (b ) p m I S g > to o to XMwnbuj fo-f Depth (m) o « « • o m o w o a e e y .» .« t*'° / > Julian D a y 1967 e s .o s o .e 7S.0 oo.o o I Log Frequency Figure A4.2.4. 7(f) between individual current vectors and temperature during the third deployment. Maps a g (zonal) and b (meridional) show 7(f,Z). 7(f,t) maps (c-e) cover JD 250-310. Map c (62m) is with the meridional 0 3 component. Maps d and e (81m) are with the zonal and meridional components, respectively. Coherence between oceanic currents and dissolved oxygen The deployment maps of coherence between currents and DO show significance in the mesoscale band (not shown). Near the surface, significant Y * y C f M » Z) encompasses fu - The 23 m tem poral map shows significant 7 k y (fM > t) in the meridional case confined to the low mesoscale band from JD 166 to JD 220, w ith the dominant feature appearing from JD 194 to JD 211 (Fig. 4.14a). The 62 and 120 m eter temporal maps for th e meridional case illustrate these depths to have decreasing 'fry(/M > t) for shorter duration features sim ilar to those seen a t 23m (Fig. 4,14b). The dominant feature of the 120m m ap is the maximum which occurs between JD 147 and JD 173. The feature appearing between JD 194 and JD 211 a t 23m, and to a lesser degree throughout the w ater column, is well-documented as a mesoscale interaction in the synoptic maps, current vectors and the tem perature time series (Figs. 2.4,2.5,2.11 c, d). This is not the case for the feature seen in the 120m meridional m ap from JD 147 to JD 173. Coherence between oceanic currents and bio-optical variables In the deployment maps, the zonal case shows maximum Y x y C fM . Z) near the surface for both bio-optical variables (Figs. A4.2.5a, c; A4.2.6a, c; A4.2.7a, c). A near-surface maximum is also present in all deployments for Y*y(/M . Z) between meridional currents and C c (Figs. A4.2.5b, A4.2.6b, A4.2.7b). However, for Y ty C fM . Z) between meridional currents and Cbl, a well-defined subsurface maximum appears a t 81m in all deployments (Figs. A4.2.5d, A4.2.6d, A4.2.7d). During the second deployment, the m aps for meridional 204 D e p th (m ) D ep th (m) Depl t f t n e s .o so.o 7 S .0 Depl Depl a •S i 1 1 Log Frequency Figure A4.2.5. 7(f,Z) between the two current directions and the bio-optical variables for the first deployment. Log Frequency Maps a (zonal) and b (meridional) show 7(f,Z) with Cc. Maps c (zonal) and d (meridional) show 7(f,Z) with Chi. to 0 0 1 ^ a a ■3 i B 3 i 8 6 60 7* 100 Figure A4.2.6. y(f,Z) between the two current directions and the bio-optical variables for the second deployment. Maps a (zonal) and b (meridional) show y(f,Z) with Cc. Maps c (zonal) and d (meridional) show yif,Z) with Chi. a I 1 1 • s s •0 y 1 00 a ■3 i • S S •V , so 100 Figure A4.2.7. y(f,Z) between the two current directions and the bio-optical variables for the third deployment. Maps a (zonal) and b (meridional) show 7(f,Z) with Cc. Maps c (zonal) and d (meridional) show 7(f,Z) with Chi. to o -3 currents and both bio-optical variables show the strongest correspondence between dominant features. The meridional/e^ map includes a subsurface coherence maxima at 101 m to accompany the characteristic near-surface feature while the meridional/Chl map indudes a surface maxima down to 30 m in addition to the standard subsurface maxima. This pattern indicates th a t the partide field has greater meridional variability within the surface mixed layer while the pigment field has greater meridional variability within the seasonal therm odine toward the base of the euphoric zone. Coherence between temperature and DO The dominant second deployment feature occurs in the upper 30 m eters a t fsD and shows values of a t least 0.7, corresponding to the 90% confidence interval (Fig. A4.2.8). In addition, significant coherence is present well into the higher frequencies of the IGW band through the upper 60 m eters. Coherence covering the full experiment is available a t 23m (Fig. 4.11a). This map shows significant YtyCfM.t) throughout the first two deployments, with maximal values occurring between JD 190 and JD 215 coinriding with the tem perature feature’ s passage. Significant YxyCfiGW>t) is confined to the second deployment and shows a broader bandwidth prior to the tem perature event. Maximal coherence appears at / s d during this time. Additionally, a sharp gradient appears in YxyCficw.t) levels between the first and second deployment. 'fo rC f.t) during the third deployment is just below significance over all bands. The temporal maps from each individual deployment reveal depth dependencies of coherence between tem perature and DO. The first deployment 208 D epth (m) Mesoscale Band IGW Band SD o ' < 3 ° > ’ & 5 0 . 0 0 .40 7 5 .0 00.0 a .4 a .1 1 .8 1 .5 i .a 0.9 0.6 0.5 0.0 0.5 0 .6 Log Frequency Figure A4.2.8. -y(f,Z ) between temperature and DOX for the second deployment. maps from 14 and 23 meters show significant coherence only within the mesoscale band and a marked lack of consistency between the two depths (Figs. 4.11 b, c). At 14m, TiyCfM.t) is insignificant except for a five day period (JD 85-90) over a lim ited bandwidth whereas the 23m map indicates strong IbyC fM ’ t) over the entire deployment in the band’ s lower frequencies. The discrepancy between these two m aps is likely due to a stronger manifestation of air-sea interaction in the 14m record. It may also be seen th at at 101m, YxyC/M.t) is significant for about two weeks a t the beginning (JD 60-76) and toward the end (JD 106-120) of the deployment (Fig. 4.11 d). These features agree well with those seen at 23m, thus indicating the passage of distinct w ater masses. However, the 23m map shows a strong Yxy(/M »t) feature (JD 90-100) which does not appear a t 101m. This corresponds to the passage of near-surface waters which are highly stratified (Fig. 2.3). Maximal Yxy(/M»t) in the second deployment maps is apparent through 160 m eters and corresponds to the advective feature around JD 200 (Figs. 4.11 e-h). At 14,23 and 62 meters, significant YxyCfM.t) also appears briefly at the beginning of the deployment (JD 145 to JD 160). YxyCflGW.t) during the second deployment has a depth dependence. Significant coherence a t 14m encompasses the IGW band down to time scales of 0(30 min) and continues through the advective event (Fig. 4 .lie ). At 23m, significant YxyCflGW.t) covers the entire deployment with greater significance but narrower bandwidth (Fig. 4. Ilf). Both of these depths indicate th at decreased Y corresponds with increased advective activity which contrasts w hat is indicated by the 62 and 120 m eter maps (Fig. 4.11 g, h). 210 This pattern of elevated near surface and minimal deep Y x yC flG W .t) persists in the third deployment. The 23m temporal m ap reveals high, but non-significant, coherence (£0.5) throughout the deployment down to time scales of 0(1 hr) with a region of statistically significant coherence occurring over a narrow band from JD 285 through the end of the deployment (Fig. 4.11 i). The 81m map (Fig. 4.11j) reveals insignificant TfryCfiow.t). The only region of significant coherence at 81m occurs in the low mesoscale band ranging from JD 290 through the end of the deployment. This again contrasts the characteristics of the 23m map which reveals significant YxyCfM.t) JD 265 to 285 (Fig. 4.11 i). Coherence between temperature and cc Spatial maps show YxyCfM. Z) to increase with decreasing frequency (at a given depth) and decrease w ith depth (Figs. A4.2.9 a,b). The second deployment temporal maps show significant YxyCfM.t) starting from the beginning of the deployment and lasting through JD 175 at 23 m (Fig. 4.15 a) and through JD 164 at 120 m (Fig. 4.15 b). The second deployment spatial map also shows a subsurface coherence maxima which appears surrounding / sd below 70 m corresponding to tidally driven vertical oscillations (Fig. A4.2.9b). The second deployment temporal maps show coherence features a t / sd which become less prominent in the middle of the deployment during the passage of the mesoscale event around JD 200. However, only the features at 120 m are statistically significant (Fig. 4.15 b), coinciding with the characteristic of the spatial m ap just described. 211 IGW Band Depl •0 .4 0 ' 5 0 .0 7 5 .0 1 0 0 .0 Log Frequency IGW Band Meaoacale Band 5 0 .0 7 S .0 A 4.2.9b 1 0 0 .0 0 .5 Log Frequency Figure A4.2.9. yif,Z) between temperature and Cc. Map a depicts the first deployment. Map b depicts the second deployment. 212 Coherence between temperature and Chi The first and third deployment spatial maps (Figs. A4.2.10 a,c) both show Y xyC f»Z ) which increases with depth and decreases with frequency. This is contrasted by the second deployment map (Fig. A4.2.10 b) which contains two features which deviate from this pattern. The first is the presence of insignificant Yxy(/M»Z) at 101m. The second is the region of significant coherence which extends throughout the water column and is centered on / sd - Two maxima are present, one centered at ~60m and the other reaching depths below 100m. The 101m map from the first deployment shows significant YxyCfM.t) over most of the deployment save for a period ranging from JD 83 to JD 97 (Fig. 4.12 a). Three supplementary maps covering the transition between the first and second deployments a t 62,81 and 101 m eters show significant YxyCfM.t) which dissipates between JD 120 and JD 125 at all depths (Figs. 4.12 d-f) as the current field transitions to a lengthy quiescent period. The second deployment temporal maps from 62 and 120 meters both show significant coherence a t / sd (Figs. 4.12 b,c). In addition, at both depths, maximal YxyCflGW.t) temporally corresponds to maximal coherence a t / sd which range from JD 145 to JD 190, a period of weak advection. Coherence between ce and Chi The first deployment temporal maps from 23 and 62 m eters show elevated YxyCfM.t) with a temporal maximum between JD 78 and JD 90 (Figs. 4.13 a, b). The 101m map shows YxyCfM.t) to be continually increasing w ith 213 D e p th (m) ^ D I ^ S D Depl es S 0 71 100 8S.0 * y so.o 8 8 88 80 7 8 L o| Frequency to I — k 4 ^ Figure A4.2.10. 7 <f,Z) between temperature and Chi. Maps a-c depict the three deployments in order. ^ time during the first deployment (Fig. 4.13 c). The second deployment temporal m aps exhibit significant decreases in ‘ ftyCfM.t) during advective regimes (Figs. 4.13 d-g). Additionally, following the period of advective interaction the 23 and 120 m eter maps show increased ‘ fryCfM.t) toward the end of the deployment (Fig. 4.13 d, g). The third deployment shows reduced TxyCfM 't) compared to the previous deployments. The 81m temporal map contains no significant events while the 101m m ap shows significant Y xyC fM .t) for two interrupted periods, JD 250 to JD 263 and JD 280 to JD 310. The latter period is on the threshold of statistical significance (Figs. 4.13 h, i). Another coherence feature in the first deployment tem poral maps occurs around fo/fso » is present throughout the deployment, and shifts to higher frequency and greater coherence with depth (Figs. 4.13 a-c). The second deployment temporal maps also show significant coherence in this frequency band. At 62m, it does not cover the entire deployment (Fig. 4.32d). All depths show reduced coherence in this band during the deployment's advective period but remain significant throughout the deployment below 62m (Figs. 4.13 d-f). As in the first deployment, these second deployment maps show increasing bandwidth and coherence with depth. After this mesoscale occurrence, significant coherence is reestablished toward the end of the deployment and continues into the third deployment through JD 270 (Figs. 4.13 h, i). Finally, the deployment maps show significant coherence in the lower IGW band. The first deployment temporal maps from 23, 62 and 101 m eters reveal no sharp gradients in YxyCflGW.t) isolines b ut values which range up to 0.9 (Figs. 4.13 a-c). Generally, the temporal maps from the second and third 215 deployment do not show significant YxyCflGW.t). The one exception is the 120m temporal map from the second deployment which shows significant 'fryCflGW.t) over the entire deployment (Fig. 4.13 g). 216 A n n an d it fi.l; E q u atio n s e t fo r th e level 2JS M ellor-Ysunada m ixed lay e r m odel M om entum Eqs.: W - 'fl(V-V« ) - | [ K“ ^ ] A®11) f £ + / l ( U - U ,) - £ [ K « g ] A6.2.1) § ^ = pg A5.1.3) C ontinuity Eq.: 3U 3V . 3x + dy = 0 A5.1.4) H eat C onservation Eq.: Iffi = IiR(0) e * A5.1.5b) G eneralized C onservation (for passive tracers): f - 8 ^ [ Rh 1 ] A6.1.6) T u rb u le n t K inetic E nergy Eq.: If■ s [* i 1 3 + • 2 6 A ® - 1 - 7 ) T u rb u le n t L ength Scale Eq.: a b:i .8) 217 Sh ear P roduction and B uovancv P roduction Terms: _ au d v P. = -<w u>gi --<w v>9i - A5.1.9a) Pb = pg <w8> E nergy D issipation Term : A5.1.9b) A5.1.10) P a ram e te riz a tio n o f E nsem ble A veraged V ertical T u rb u len t F lu x es Sm , h , q are stability functions (cf., [Mellor and Yamada, 19 82]). P a ram eterizatio n fo r In te rio r D iffusivities (mg/s): B ackground D iffusivities fo r In te rn a l W ave B reaking Kmi = 2 * 10 ; K Hi = 1 * 1 0 '3 B ackground D iffusivities fo r S h ear In stab ility K m s= 1*10-6 ; K h s = 1 * 10-3 ForRicO : Km = K Mi + Kms A5.1.13a) K h = K h i + K h s A5.1.13b) F o r R i > 0: [ -<w u>, -<WV>, -<w6> ] = [ Km ( £ ) Km ^ K h (§ £ ) ] A5.1.11) D ianvcnal D iffusivities; K m, h, q = 1 Q S m, H,q A5.1.12) A5.1.13c) 218 Kh = K Hi + K hs[q + sR i^s] Boundary C onditions: W ind S tre ss (x0*, toy) = PaCioUio | Uio | „ au = pw Km 0z it a v T o y = pW KM z -0 l«0 = PaCioVio I Vio I T u rb u len t H eat Flux (H) U XT W H = pw cpK h 0^ (■ 0 ■ (l*tt) Q aw + Qlw + Q ae + Qla A5.1.13d) AS. 1.14a) AS. 1.14b) A5.1.14c) 219 Annendlx S.2: Equation set used for the pelagic ecosystem model Aufa 5 * = (l-Yi) odo,Nn.Nr) N. - Gr mLxN. + ^ [ k h A5.2.1a) ,(Na) o(Io,N„^r) = Q(N„,Nr)A(W L i * : N j l Ks+Na A5.2.1b) N utrient Switching Q(N„,Nr) = Qi(Nn^r) + Q2(Nr) A5.2.1c) (W roblew ski, 1977) Autotrophic Growth Rate Ado) = a’I^ l 'l > 8 chi 0m lo + P®K p= - f L A5.2.1d) A5.2.1e) A5.2.1f) D (Kiefer, 1993) A5.2.1g) Attenuation of Io Io(*n+l) = Io(Zn) 6 * + [Chl(zn)]) dz A5.2.1h) 220 Kw(z) = .079 e -213* - .02 e -0027«. .0 0 7 3 log(z) + .069 A5.2.1i) (empirical fit to data from Baker and Frouin 1987) K c(z)« 0 .0 5 1 8 C h l(z ) 0-872 C:N Na(z) Chl(z) = ■ 0 k(z) (Morel 1988) A5.2.1j) A5.2.1k) Zooplanfctea (N,) dN. (jj. = P iG i + P2G2 + P3G3 * H 5L2N z A5.2.2a) N*_ Ke+Nx Zooplankton growth rate Gi = g * zNz ----- K a+ ^P jC j j= i A5.2.2b) PiCj Pi = XpnCn n=l A5.2.2c) Ci = Na,Nb,Np i = 1...3 (Fasham et aL, 1990) B a c te ria (Nb) A5.2.3a) 221 t t _ — € \> ^ b .^ d — a c o q u 1 “ K4 + S + Nd A5.Z.3W u = = T c f ^ f t w M -2™ S = min (Nr, T iN d) (Fasham e t aL, 1990) C£2£i(Nd) ® * - 7 x Ado) Qi(N„^r) Na + vjuNp + QtfisLjiNi - Ui * s h f 1 ] Ammonium (Nr ) ^ = - Ado) Q2<Nr) Na - U2 + HaNb + fl2M 6LaN* - H«Nr * Bz N itra te (Nn) d.[rr M l! [ K h 1 z . dN, dt A5.2.4) A5.2.5) 1 = ■ Ado) Qi(Nn»N r) Na + HeNr + ^ [ k H^ ] A5.2.6) PO N (NP) = (l*Pi) Gi + (I-P2) G 2 ' P3G3 + |J.iLiNa - mNp - w8 0 zP A5.2.7) 222 A nnendix B.3: Ecosystem model param eter definitions [(tyZ) dep end en ce is im plied] Sym bol Desi A Autotrophic growth rate D Photoperiod (fraction of 24 hours) Gi,2,3 Zooplankton grazing rate (Na, Nb, Np) I0 Photosynthetically available radiation Kc Chlorophyll specific attenuation coefficient Kh Diapycnal diffusivity Kw Attenuation coefficient due to water Na Autotroph (phytoplankton) concentration Nb Bacteria concentration Nd DON concentration Nn N itrate concentration Np PON (detritus) concentration Nr Ammonium concentration Nz Zooplankton concentration P Instantaneous carbon-specific photosynthetic rate Qi,2 N utrient limitation (Nn, Nr) U i,2 Bacterial uptake rate (Nd, Nr) U nits d 1 d-l Ein m2 d m_ 1 m2 see m'1 mMol N mMol N m3 mMol N n mMol N mMolN n? mMol N m3 mMol N - = x - TOr d l mMol N m3 d ♦ eK Quantum yield Carbon:Chl ratio Nutrient-lim ited autotrophic specific growth rate MolC Ein MolC g C h la d l 223 A ppendix S.4; Ecosystem model constant definitions Sym bol D escription V alue A a*chl Chlorophyll specific absorption coefficient 11 m gC hla be Constant param eter from optical production model 4.56 C:N Redfield ratio c a MolC MolN g * b Maximum bacterial growth rate 1.5/day gz Maximum zooplankton growth rate 1.1/day Ki> 2 H alf saturation for nutrient uptake (Nn, Nr) mMolN (.5, .4) ms k 3 H alf saturation constant for zooplankton ingestion _ _ mMolN 10 H alf saturation constant for bacterial uptake 0.5 — in3 Kb H alf saturation constant for phytoplankton m ortality . - mMolN »'2 K« H alf saturation constant for zooplankton m ortality o.2 w8 Detrital sinking rate 12.5 f P i,2,3 Zooplankton assimilation efficiencies (Na, Nb, Np) .5 0m Maximum quantum yield Yi DON exudation fraction 0.1/day Ml,5 Specific plankton m ortality rate (Na , Nz) (0.15, 0.3Vday M ’ S Specific ammonium excretion rate (Nz, Nb) 0.05/day M 4 D etrital loss rate 0.22/day M 6 Nitrification rate 0.03/day V Fraction of D etrital Loss 0.2 Pm Minimum C:Chl ratio . MolC gC hla P i,2,3 Zooplankton feeding preferences (Na, Nb, Np) .5,.25,.25 Ql,2 Fraction of zooplankton m ortality (Na, Nr) 0.5, 0.4 ¥ N itrate uptake inhibition param eter 1.5 mMolN Annftndlx K .K ; N itrate-T em p eratu re rela tio n d eriv ed from in -situ p ro files A nitrate-tem perature relation has been derived based on data taken from the four interdeploymental cruises which took place during the Biowatt experiment. These profiles were taken down to 100m and provide a basis for creating an empirical relation for predicting surface concentrations, especially during the nutrient depleted state characteristic of the summertime production regime which is based on recycled nutrients (i.e., NH*). In order to predict nitrate concentrations typical of the region's characteristic 18s water, GEOSECS data from a station a t 36.6N 70.3W were obtained from the M acintosh program OceanAtlas [Atkinson, 1994]. These profiles reach depths of600m and concentration values for the deepest points reach 5-6 pM N. The empirical relation derived from these data consists of a combination of two curves in order to create a more realistic fit. The transition between the two curves occurs at 19.6s C and is summarized below (Eq. A5.5.1). The data and the resulting curve fit are shown (Fig. A5.5.1). For T < 19.6s C [NOjf ] = - 0.03 T3 + 1.8 T2 -40.4 T + 298.4 A5.5.1a) For T £ 19.6 8 C [NO<f ] = 1.12 * 107 T AS.S.lb) The R2 values for the curve fits are 0.62 and 0.76 respectively. These may be considered somewhat high. However, for both individual fits all available data was used in order to provide a smoother transition between the two curves. 225 to BIOWATT OC CRUISE DATA (*) GEOSECS DATA z C M 18 20 22 24 26 Temperature (’C) Figure A5.5.1. Empirical relation between temperature and nitrate. The data shown here were obtained during the deployment/recovery cruises associated with the Biowatt mooring experiment (*) and the GEOSECS cruises (+). The latter data set was utilized to increase the number of deep nitrate observations which provided a more robust representation of deep convection for the interdisciplinary model. Annftndix S-fe Determination of the local heat budget It has been well documented th at the site of this moored experiment is subjected to strong advective activity due to its proximity to the Gulf Stream (e.g., Chapters 1 and 3). The magnitude of this advective activity significantly hampers the ability of a 1-D mixed layer model to predict the physical field's evolution from locally measured sea surface fluxes of heat and momentum. Generally, current time series presented in the form of vectors which reveal directional veering and magnitudinal shifts may be used as an indicator of elevated advective activity. Another means is through the calculation of a local heat budget which provides a more quantitative perspective. This budget is determined using heat fluxes measured and derived from the meteorological data and in-situ tem perature and PAR sampled by the MVMS packages. The tem perature data are low-pass filtered using a six day gaussian window. The filter width was chosen to improve the estim ate of the advective term as internal wave fluctuations have been shown to severely lim it the ability to achieve closure when calculating local heat budgets [Hebert, 1994]. Hourly profiles of tem perature (T) are extracted from the MVMS time series with SST from the buoy taken a t 3m to provide some representation of the diurnal mixed layer. These are passed to an interpolation scheme in order to enhance the vertical resolution before performing the vertical integrations necessary for the heat budget calculation. The following equation set presents the terms involved in the calculation of the heat budget [Price et al., 1978]. i + U . V T . - ^ + - ^ - § *5.6.1) 0 fuHVH TdZ= A5.6.2) -5W 0 0 - J f £ d Z - <we>(0) + - 1 — [Io(0 ) - I0(-Zr,f)] - J w (Z )||d Z Zref Pw P -Znt (A) (B) (C) © ) (E) 0 I W(Z) ^ dZ = W(-Zref) [T - T(-Zref)] A5.6.3) -W W CZref) = -W(-Z) Zre{/Z A5.6.4) _ 1 0 T = 2 ^ Jt (Z) dZ A5.6.5) Equation A5.6.1 is the formulation for heat conservation for a fluid parcel, describing the temporal change in heat content due to advective and turbulent heat fluxes and short wave radiation. This conservation equation is integrated from the surface down to a reference depth (Z ref)> 6 m eters below the depth of the mixed layer (Zmid), defined as the depth where T(14m) - T(Z) is 0 .1 “ C. Thus, the evolution of Z^f strongly mimics the evolution of Z^d. However, Zn t is held constant while the isotherm originally associated w ith that depth stays within ±6 meters of Z^f. If the isotherm's excursions exceed this lim it, a new isotherm is chosen and Znf is updated. 228 Since the goal of this study is to quantify horizontal advection, the advective term has been partitioned into its horizontal and vertical components and the vertically integrated heat conservation equation (Eq. A5.6.2) has been organized with all term s, save the horizontal component, collected on the right hand side. These have been labeled as they will be referred to individually in the following discussion. The first term (A) represents the temporal change in w ater column heat content indicated by the tem perature profiles. The surface boundary conditions, consisting of turbulent heat flux (term B) and short wave insolation (term C), are taken from the meteorological time series. Surface turbulent heat flux (Fig. A5.6.1a) consists of the summation of the latent (Qia), net long wave (Qiw ) and sensible (Qm) h eat fluxes and represents a loss of heat from the ocean. Short wave insolation shows a distinct seasonal cycle (Fig. A5.6.1b) and represents the mayor heat input at the ocean's surface. The bottom flux condition is obtained using the in-situ PAR profiles to estim ate radiative heat flux leaving the control volume a t Znf (term D). Turbulent fluxes a t Z^f are assumed negligible. The final term (E) represents vertically advected heat flux which takes the form shown in equation A5.6.3 after integration by parts. Up welling velocity (i.e., W(Z)>, at the depth of the isotherm initially defined at Zs-Znf, is estim ated from the rate a t which the isotherm's depth changes and is referenced back to -Z ^ using a linear transformation (Eq. A5.6.4). Average w ater column tem perature ( T ), required to solve equation A5.6.3, is determined with equation A5.6.5. 229 5.6.1 a 300 1 0 0 150 200 250 o 5.6.1b o < 0 1 0 0 150 200 250 300 Julian Day 1987 Figure A5.6.1. Sea surface heat fluxes measured during Biowatt. Figure a shows the summation of latent, sensible and long wave heat flux. Figure b shows short wave insolation. These data are used as surface conditions in the heat budget analysis. Units are in W/m2. This type of calculation was previously performed on data obtained during MILE. This experiment took place in 1977 near Ocean W eather Station P in the North Pacific and consisted of two surface moorings which gathered high resolution current and tem perature time series data while surface heat and momentum flux data were obtained from research vessels occupying station P [Davis et al., 1981]. This site is well-suited for studies of locally generated variations in w ater column heat content due to the lack of advective activity. This is emphasized by the fact th at this site's heat budget closed within 20 W/m2 using the equation set just presented with the additional assumption th a t there was no horizontal adv&tive component to the h eat budget. This contrasts the Biowatt case for which the heat budget is calculated with the goal of estimating the magnitude of the horizontal heat flux. The heat budget analysis is summarized (Fig. A5.6.2) by time series of temporal change in surface layer heat content (Term A in Eq. A5.6.2), summation of surface and bottom boundary conditions (Terms B-D in Eq. A5.6.2), vertically advected heat flux (Term E in Eq. A5.6.2) and horizontally advected heat flux (LHS of Eq. A5.6.2). The estimated upwelling velocity (Eq. A5.6.4) is also shown along with current speed from 14m. Taken together, these flux and velocity term s are useful in defining periods of strong advection by intercomparing periods of elevated horizontal transport as revealed in the current time series and as indicated by the flux budget. The change in heat content of the surface layer (Fig. A5.6.2a) ranges from ± 16 W/m2 over the course of the entire experiment w ith the largest positive values occurring during the first deployment and during the advective 231 130 300 A5.6.2 a s i I e !. 8 7 3 0 0 A5.6.2 a 8 r !. e 100 300 A5.6.2 a 232 Spmi Upm M tm t C U m O * i — e- r f flat (9 /m ? ) 8 I 7 g N A5.6.2 d N « o ( 0 0 300 300 350 A5.6.2 e s I . I e A5.6.2 f Figure A5.6.2. Time series of terms in heat flux budget (W/m2). Figure a shows the change in water column heat content (term A in eq. A5.6.2). Figure b shows the summation of the surface fluxes. Figure c shows the estimated vertical heat flux (eq. A5.6.3). Figure d shows the estimated horizontal heat flux (lhs of eq. A5.6.2). Figure e shows upwelling velocity (m/day). Figure f shows current speed at 20m (cm/s). 233 event of the second deployment (JD 190-210). Prior to this event and following the onset of perm anent stratification (Fig. 1.17), values are consistently positive and characterized by relatively low amplitudinal shifts. Following this event, values oscillate around zero until JD 260. The experiment's largest negative values and amplitudinal changes of increasing frequency occur during the end of the third deployment and are associated with the breakdown of seasonal stratification and increasing advective activity. This seasonal cycle of changes in w ater column heat content determined from in-situ tem perature tim e series strongly corresponds to the independently determined boundary condition summation (Fig. A5.6.2b) where net positive heat flux into the surface layer shows strong correspondence to the period of positive changes in h eat content. The tim e series of vertical heat flux (Fig. A5.6.2c), based on the estim ated upwelling velocity (Fig. A5.6.2e) and a tem perature gradient (Eqs. A5.6.3-5), shows maximum values (*> 170 W/m2) around JD 165-175. The coinciding upwelling velocity is - 0.7 m/day and although this is not a maximum for the experiment, it is significantly higher than typical values for the experiment's duration. These characteristics are associated w ith the lowest current speeds of the experiment and elevated mesoscale band variance in the bio-optical fields associated with large amplitude variations in phytoplankton biomass (Fig. 1.8) and may be indicative of vertical transport of inorganic nutrients. Two other periods of elevated vertical heat flux are associated with elevated currents during the first and second deployments (Fig. A5.6.2f, JD 100-120, JD 200-210) and the concomitant mesoscale events. The upwelling velocities associated with these mesoscale periods attain values of around 2 234 m/day. Maximal values of upwelling velocity (~ 7 m/day) are achieved toward the end of the period of deep convective mixing (JD 80*90) and are likely an artifact of the developing stratification and the need to continually redefine the reference isotherm. The time series of estimated advective heat flux shows the largest magnitude (100-275 W/m2) through JD ilO corresponding to the elevated current speeds and mesoscale activity of the period. Outside of this time period values for advected heat flux generally range between -100 and 100 W/m2, even during the period characterized by minimal mesoscale activity and low current speeds. The fact th at the second deployment mesoscale event does not have a stronger representation within this flux estim ate is somewhat surprising. There is a period during this event when the experiment's maximum positive advective flux (~ 170 W/m2) is achieved but this is not distinctively higher than the periods surrounding this event Obviously, this site is highly advective and closure (i.e., the advective term ) within 20 W/m2 is not observed as in previous studies [Davis et al., 1981]. But the values of estimated heat flux on the order of ± 80-100 W/m2 during the quiescent period (JD 130-180) is surprising and certainly disturbing with respect to recreating the physical environment observed in the time series in the hopes of emulating the measured ecosystem param eters w ith the interdisciplinary model. 235 B ibliography Ackelson, S. G., J. J. Cullen, J. Brown, and M. P. Lesser, Some changes in the optical properties of marine phytoplankton in response to high light intensity, Ocean Optics, 10,238-249,1990. Ackelson, S. G., J. J. Cullen, J. Brown, and M. Lesser, Irradiance-induced variability in light scatter from marine phytoplankton in culture, J. Plankton Res., 1 5 ,737-759,1993. Altabet, M. A., Particulate new nitrogen fluxes in the Sargasso Sea, J. Geophys. Res., 9 4 ,12771-12779,1989. Apel, J. R., Principles o f Ocean Physics, 634 pp., Academic Press, New York, 1987. Atkinson, L., OceanAtlas and Atlast, Oceanography, 7 ,63-64,1994. Baker, E. T., and J. W. Lavelle, The effect of particle size on the light attenuation coefficient of natural suspensions, J. Geophys. Res., 89, 8197-8203, 1984. Baker, K. S., and R. Frouin, Relation between photosynthetically available radiation an total insolation a t the ocean surface under d ear skies, Limnol. and Ocean., 3 2 ,1370-1377,1987. Bartz, R., R. Spinrad, and J. C. Kitchen, A low power, high resolution, in situ fluorometer for profiling and moored applications in water, Ocean Optics, 9 , 157-170,1988. 236 Bartz, R., J. R. V. Zaneveld, and H. Pak, A transmissometer for profiling and moored observations in water, Ocean Optics, 5 ,102-108,1978. Bendat, J. S., and A. G. Piersol, Random Data: Analysis and measurement procedure, 566 pp., W iley-Intersdence, New York, 1986. Betzer, P. R., W. J. Showers, E. A. Laws, C. D. Winn, G. R. DiTullio, and P. M. Kroopnick, Prim ary productivity and particle fluxes on a transect of the equator at 153W in the Pacific Ocean, Deep-Sea Res., 3 1 ,1-11, 1984. Bidigare, R. R., Nitrogen excretion by marine zooplankton, in Nitrogen in the Marine Environment, edited by E. J. Carpenter and D. G. Capone, pp. 385-409, Academic Press, New York, 1983. Bidigare, R. R., J. M arra, T. D. Didrey, R. Iturriaga, K S. Baker, R. C. Smith, and H. Pak, Evidence for phytoplankton succession and chromatic adaptation in the Sargasso Sea during springtime 1985, Mar. Ecol. Prog. Ser., 6 0 ,113-122., 1990. Bidigare, R. R., B. B. Prezelin, and R. C. Smith, Bio-optical models and problems of scaling, in Primary Productivity and Biogeochemical Cycles in the Sea, edited by P. G. Falkowski, pp. 175-212, Plenum Press, New York, 1992. Bidigare, R. R., R. C. Smith, K. S. Baker, and J. Marra, Oceanic primary production estim ates from measurements of spectral irradiance and pigment concentrations, Global Biogeochem. Cycles, 1 ,171-186,1987. 237 Bishop. J . K. B., The correction and suspended m atter calibration of Sea Tech transmissometer d ata, Deep-Sea Res., 3 3 ,121-134,1986. Bishop, J . K B., M. H. Conte, P. H. Wiebe, M. R. Roman, and C. Langdon, Particulate m atter production and consumption in deep mixed layers: Observations in a warm-core ring, Deep-Sea Res., 3 3 ,1813-1841,1986. Bloomfield, P., Fourier Analysis o f Time Series: An Introduction, 268 pp., Wiley and Sons, New York, 1976. Blumberg, A. F., and G. L. Mellor, A coastal ocean numerical model, in Proceedings o f the Symposium on Mathematical Modeling of Estuarine Physics, edited by pp. Springer-Verlag, New York, 1980. Booth, C. R., The design and evaluation of a m easurement system for photosynthetically active quantum scalar irradiance, Limnol. Oceanogr., 19, 326-335,1976. Briscoe, M. G., and R. A. Weller, Preliminary results from the long-term upper-ocean study (LOTUS), Dyn. Atm . and Oceans, 8 ,243-265,1984. Broecker, W. S., and T. Peng, Tracers in the Sea, 690 pp., Eldigio Press, Palisades, N. Y„ 1982. Bunker, A. F., Computation of surface energy flux and annual air-sea interaction cycles of the N orth Atlantic ocean, Mon. Weather Rev., 104, 1122-1140, 1976. Bunker, A. F„ and L. V. Worthington, Energy exchange charts of the North Atlantic ocean, Bull. Am. Met. Soc., 5 7 ,670-678,1976. 238 Cairns, J . L., Internal wave measurements from a midwater float, J. Geophys. Res., 8 0 ,299-306,1975. Cho, B. C., and F. Azam, Mqjor role of bacteria in biogeochemical fluxes in the ocean's interior, Nature, 3 3 2 ,441-443,1988. Cleveland, J. S., M. J. Perry, D. A. Kiefer, and M. C. Talbot, Maximal quantum yield of photosynthesis in the northwestern Sargasso Sea, J. Mar. Res., 4 7 ,869-886,1989. Comillon, P., D. Evans, and W. Large, Warm outbreaks of the Gulf Stream into the Sargasso Sea, J. Geophys. Res., 9 1 ,6583-6596,1986. Cullen, J. J., On models of growth and photosynthesis in phytoplankton, Deep-Sea Res., 3 7 ,667-683,1990. Cullen, J. J., Hypotheses to explain high-nutrient conditions in the open sea, Limnol. Oceanogr., 3 6 ,1578-1599,1991. Cullen, J. J., M. R. Lewis, C. O. Davis, and R. T. Barber, Photosynthetic characteristics and estimated growth rates indicate grazing is the proximate control of primary production in the Equatorial Pacific, J. Geophys. Res., 9 7 ,639-654,1992a. Cullen, J. J., E. Stewart, E. Renger, R. W. Eppley, and C. D. W inant, Vertical motion of the thermocline, nitracline and chlorophyll maximum layers in-relation to currents on the Southern California Shelf, J. Mar. Res., 4 1 ,239-262,1983. 239 Cullen, J. J., X. Yang, and H. MacIntyre, N utrient limitation of marine photosynthesis, in Primary Productivity and Biogeochemical Cycles in the Sea, edited by P. G. Falkowski, pp. 69-88, Plenum Press, New York, 1992b. Cullen, J . J., C. M. Yentsch, T. L. Cued, and H. L. MacIntyre, Autofluorescence and other optical properties as tools in biological oceanography, Ocean Optics, 9 ,149-156,1988. Davis, R E., R DeSzoeke, D. Halpem, and P. Niiler, Variability in the upper ocean during MILE. P a rti: The heat and momentum balances, Deep- Sea Res., 28A, 1427-1451,1981. de Angelis, M. A , and C. Lee, Methane production during zooplankton grazing on marine phytoplankton, Limnol. Oceanogr., 3 9 ,1298-1308, 1994. Denman, K., Covariability of chlorophyll and tem perature in the sea, Deep- Sea Res., 2 3 ,539-550,1976. Denman, K. L., and M. R Abbott, Time evolution of surface chlorophyll patterns from cross-spectrum analysis of satellite color images, J. Geophys. Res., 9 3 ,6789-6798,1988. Denman, K L., and A. E. Gargett, Time and space scales of vertical mixing and advection of phytoplankton in the upper ocean, Limnol. Oceanogr., 28, 801-815, 1983. 240 Denman, K L., and T. Platt, Coherences in the horizontal distributions of phytoplankton and tem perature in the upper ocean, Mem. Soc. Roy. des Sciences de liege, 6, 19-30,1975. Denman, K L., and T. M. Powell, Effects of physical processes on planktonic ecosystems in the coastal ocean, Oceanogr. Mar. Biol. Ann. Rev., 2 2 ,125-168,1984. Deuser, W. G., Seasonal and interannual variations in deep-water particle fluxes in the Sargasso Sea and their relation to surface hydrography, Deep-Sea Res., 3 3 ,225-246,1986. Dickey, T. D., The emergence of concurrent high resolution physical and bio- optical measurements in the upper ocean and their applications, Rev. Geophys., 2 9 ,383-413,1991. Dickey, T. D., T. Granata, M. Hamilton, J. Wiggert, D. Manov, D. Siegel, Z. Chai, M. Stramska, and J. Scott, Final report on the multi-variable moored system (MVMS) and meteorological data sets collected during the Biowatt II experiment in the Sargasso Sea in 1987, OPG-90-01, University o f Southern California, 1990. Dickey, T. D., T. Granata, J. M arra, C. Langdon, J. Wiggert, Z. Chai, M. Hamilton, J. Vasquez, M. Stramska, R. Bidigare, and D. Siegel, Seasonal variability of bio-optical and physical properties in the Sargasso Sea, J. Geophys. Res., 9 8 ,865-898,1993. 241 Dickey, T. D., J. M arra, T. Granata, C. Langdon, M. Hamilton, J. Wiggert, D. Siegel, and A. Bratkovich, Concurrent high resolution bio-optical and physical time series observations in the Sargasso Sea during the spring of 1987, J. Geophys. Res., 9 6 ,8643-8664,1991. Dickey, T. D., D. A. Siegel, A. Bratkovich, and L. Washburn, Optical features associated with thermohaline structures, Ocean Optics, 8, 308-313, 1986. Dickey, T. D., and J. J. Simpson, The influence of optical water type on the diurnal response of the upper ocean, Tellus, 3 5 ,142-154,1983. Doney, S. C., D. M. Glover, and R. G. Nqjjar, A new coupled, one-dimensional biological-physical model for the upper ocean: Applications to the JGOFS Bermuda Atlantic Time Series (BATS) site, Deep-Sea Res., In press, 1995. Dortch, Q., The interaction between ammonium and nitrate uptake in phytoplankton, Mar. Ecol. Prog. Ser., 6 1 ,183-201,1990. Dugdale, R. C., and J. J. Goering, Uptake of new and regenerated forms of nitrogen in primary productivity, lim nol. Oceanogr., 12,196-206, 1967. Eppley, R. W., Temperature and phytoplankton growth in the sea, Fish. B ull., 7 0 ,1063-1085, 1972. 242 Eppley, E. W., S. G. Horrigan, J. A. Puhrman, E. R. Brooks, C. C. Price, and K. Sellner, Origins of dissolved organic m atter in southern California coastal waters: Experiments on the role of zooplankton, Mar, Ecol. Prog. Ser., 6 ,149-159,1981. Eppley, R. W., and B. J. Peterson, Particulate organic m atter flux and planktonic new production in the deep ocean, Nature, 282,677-680, 1979. Eriksen, C. C., Variability in the upper-ocean internal wave field a t a Sargasso Sea site, J. Phys. Oceanogr., 1 8 ,1496-1513,1988. Evans, G. T., A framework for discussing seasonal succession and coexistance of phytoplankton species, Limnol. Oceanogr., 3 3 ,1027- 1036, 1988. Evans, G. T., and J. S. Parslow, A model of annual plankton cycles, Biol. Ocean, 3, 327-347,1985. Falkowski, P. G., Physiological responses of phytoplankton to natural light regimes, J. Plankton Res., 6, 295-307,1984. Falkowski, P. G., C. N. Flagg, G. T. Rowe, S. L. Smith, T. E. Whitledge, and C. D. Wirrick, The fate of the spring phytoplankton bloom: Export or oxidation, Cont. Shelf Res., 8 ,457-484,1988. Falkowski, P. G., R. M. Greene, and R. M. Geider, Physiological limitations on phytoplankton productivity in the ocean, Oceanography, 5 ,84-91, 1992. 243 Falkowski, P. G., and D. A. Kiefer, Chlorophyll a fluorescence in phytoplankton: Relationship to photosynthesis and biomass, J. Plankton Res., 7, 715-731,1985. Fasham, M., H. Ducklow, and S. McKelvie, A nitrogen-based model of plankton dynamics in the oceanic mixed layer, J. Mar. Res., 4 8 ,591- 639,1990. Fasham, M. J . R., The statistical and mathematical analysis of plankton patchiness, Oceanogr. Mar. Biol. Ann. Rev., 16,43-79,1978. Fasham, M. J . R., and P. R. Pugh, Observations of horizontal coherence of chlorophyll a and tem perature, Deep-Sea Res., 2 3 ,527-538,1976. Fasham, M. J . R., J. L. Sarmiento, R. D. Slater, H. W. Ducklow, and R. Williams, Ecosystem behavior a t Bermuda Station 'S' and ocean weather station 'India': A general circulation model and observational analysis, Global Biogeochem. Cycles, 7 ,379-415,1993. Frost, B. W., A threshold behavior in Calanus pacificus, Limnol. Oceanogr., 20, 263-266,1975. Frost, B. W., The role of grazing in nutrient-rich areas of the open sea, Limnol. Oceanogr., 3 6 ,1616-1630,1991. Fuhrman, J ., Bacterioplankton roles in cycling of organic m atter: The microbial food web, in Primary Productivity and Biogeochemical Cycles in the Sea, edited by P. G. Falkowski, pp. 361-383, Plenum Press, New York, 1992. 244 Gargett, A. E., Ocean turbulence, Annual Rev. Fluid Mechanics, 2 1 ,419- 451, 1989. Goldman, J. C., D. A. Caron, and M. R. Dennett, Regulation of gross growth efficiency and ammonium regeneration in bacteria by substrate C:N ratio, Limnol. Oceanogr., 3 2 ,1239-1252,1987. Gonella, J., A rotary-component method for analyzing meteorological and oceanographic vector time series, Deep-Sea Res., 1 9 ,833-846,1972. Granata, T., J. Wiggert, T. Dickey, and M. Hamilton, Synoptic and seasonal scale control of vertical chlorophyll distributions by physical processes, Deep-Sea Res., manuscript in preparation, 1994. Gregg, M. C., Diapycnal mixing in the thermodine: A review, J. Geophys. Res., 92, 5249-5286,1987. Gregg, M. C., and M. G. Briscoe, Internal waves, finestructure, microstructure and mixing in the ocean, Rev. Geophys. Space Phys., 17, 1524-1548, 1979. Halpem, D., Y. Chao, C.-C. Ma, and C. Mechoso, Comparison of tropical Pacific tem perature and current simulations with two vertical mixing schemes embedded in an ocean general circulation model and reference to observations, J. Geophys. Res., 100, 2515-2522,1995. Hamilton, M., T. Granata, T. Dickey, J. Wiggert, D. Siegel, J. Marra, and C. Langdon, Diurnal variations of bio-optical properties in the Sargasso Sea, Ocean Optics, 1 0 ,214-224,1990. 245 Harris, G. P., Phytoplankton Ecology: Structure, Function and Fluctuation, 384 pp., Chapman and Hall, New York, 1986. Harrison, W. G., Regeneration of nutrients, in Primary Productivity and Biogeochemical Cycles in the Sea, edited by P. G. Falkowski, pp. 385- 407, Plenum Press, New York, 1992. Hebert, D., Closing a heat budget: Effect of internal waves, Deep-Sea Res., 4 1 ,1-8,1994. Hitchcock, G. L., C. Langdon, and T. J. Smayda, Seasonal variations in the phytoplankton biomass and productivity of a warm-core Gulf Stream ring, Deep-Sea Res., 3 2 ,1287-1300,1985. Hobson, L. A., The seasonal and vertical distribution of suspended particulate m atter in an area of the N ortheast Pacific ocean, Limnol. Oceanogr., 12,642-649,1967. Hofinann, E. E., and J. W. Ambler, Plankton dynamics on the outer southeastern US continental shelf. Part II: A time-dependent biological model, J. Mar. Res., 4 6 ,883-917,1988. Hollibaugh, J. T., A. B. Carruthers, J . A. Fuhrman, and F. Azam, Cycling of organic nitrogen in marine plankton communities studied in enclosed water columns, Mar. Biol., 5 9 ,15-21,1980. Holloway, G., Effects of velocity fluctuations on vertical distributions of phytoplankton, J. Mar. Res., 4 2 ,559-571,1984. 246 Holloway, G., and K. Denman, Influence of internal waves on primary production, J. Plankton Res., 11,409>413,1989. Hutson, V., Predator mediated coezistance with a switching predator, Math Biosci., 68, 233-246,1984. Isemer, H. J., and L. Hasse, The Bunker climate atlas o f the North Atlantic Ocean, Vol. 1: Observations, pp., Springer-Verlag, New York, 1987. Jassby, A. D., and T. P latt, Mathematical formulation of the relationship between photosynthesis and light for phytoplankton, lim nol. Oceanogr., 2 1 ,540-547,1976. Jenkins, W. J., and J. C. Goldman, Seasonal oxygen cycling and primary production in the Sargasso Sea, J. Mar. Res., 4 3 ,465-491,1985. Jerlov, N. G., Marine Optics, 231 pp., Elsevier, Amsterdam, 1976. Jones, B., L. Washburn, and R. Smith, Spatial variability, turbulent mixing, and phytoplankton productivity in the North Atlantic ocean south of Iceland in August 1991, Deep-Sea Res., in prep., 1995. Kahru, M., Phytoplankton patchiness generated by long internal waves: A model, Mar. Ecol. Prog. Ser., 1 0 ,111-117,1983. Kamykowski, D., Possible interactions between phytoplankton and semidiurnal internal tides, J. Mar. Res., 3 2 ,67-89,1974. 247 Kamykowski, D., Possible interactions between plankton and semidiurnal internal tides. II. Deep thermodines and trophic effects, J. Mar. Rea., 34, 499-509,1976. Kamykowski, D., The simulation of a southern California red tide using characteristics of a simultaneously-measured internal wave field, Ecological Modeling, 12 ,253-265,1981. Kamykowski, D„ S. A. McCollum, and G. J. Kirkpatrick, Observations and a model concerning the translational velocity of a photosynthetic marine dinoflagellate under variable environmental conditions, Limnol. Oceanogr., 3 3 ,66-78,1988. Kantha, L. H., and C. A. Clayson, An improved mixed layer model for geophysical applications, J. Geophys. Res., 9 9 ,25,235-25,266,1994. Kaplan, W. A., Nitrification, in Nitrogen in the Marine Environment, edited by E. J. Carpenter and D. G. Capone, pp. 139-190, Academic Press, New York, 1983. Kiefer, D. A., Chlorophyll a fluorescence in m arine centric diatoms: Responses of chloroplasts to light and nutrient stress, Mar. Biol., 23, 39-46, 1973. Kiefer, D. A., Growth and light absorption in the marine diatom Skeletonema Costatum, in Towards a Model o f Ocean Biogeochemical Processes, edited by G. Evans and M. Fasham, pp. 93-122, Springer-Veriag, New York, 1993. 248 Kiefer, D. A., and J. N. Kremer, Origins of vertical patterns of phytoplankton and nutrients in the tem perate, open ocean: A stratigraphic hypothesis, Deep-Sea Res., 28a, 1087-1105,1981. Kiefer, D. A., and B. G. Mitchell, A simple, steady state description of phytoplankton growth based on absorption cross section and quantum efficiency, Limnol. Oceanogr., 2 8 ,770-776,1983. Kirk, J . T. 0., Light and Photosynthesis in Aquatic Ecosystems, 401 pp., Cambridge Univ. Press, Cambridge, 1983. Klein, P., and B. Coste, Effects of wind-stress variability on nutrient transport into the mixed layer, Deep-Sea Res., 3 1 ,21-37,1984. Knap, A. H., A. F. Michaels, R. L. Dow, R. J. Johnson, K Gundersen, J. C. Sorensen, A. R. Close, M. Hammer, G. A Knauer, S. A. Lohrenz, V. A Asper, M. Tuel, H. Ducklow, H. Quinby, P. Brewer, and R. Bidigare, BATS Data Report, B-2, BBSR , 1992. Knap, A H., A F. Michaels, R. L. Dow, R. J. Johnson, K. Gundersen, J. C. Sorensen, A. R. Close, F. Howse, M. Hammer, N. Bates, G. A Knauer, S. A Lohrenz, V. A. Asper, M. Tuel, H. Ducklow, and H. Quinby, BATS D ata Report, B-3, BBSR, 1993. Kolber, Z., and P. G. Falkowski, Use of active fluorescence to estimate phytoplankton photosynthesis in situ, Limnol. Oceanogr., 3 8 ,1646- 1665,1993. 249 Komar, P. D., A. P. Morse, L. F. Small, and S. W. Fowler, An analysis of sinking rates of natural copepod and euphasid fecal pellets, Limnol. Oceanogr., 2 6 ,172-180,1981. Hondo, J., Air-sea bulk transfer coefficients in diabatic conditions, Boundary Layer Met., 9 ,91-112,1975. Kraus, E. B., and J. S. Turner, A one-dimensional model of the seasonal thermocline n . The general theory and its consequences, Tellus, 19, 98-105, 1967. Lande, R., and A. M. Wood, Suspension times of particles in the upper ocean, Deep-Sea Res., 3 4 ,61-72,1987. Langdon, C., Dissolved oxygen monitoring system using a pulsed electrode: Design, performance and evaluation, Deep-Sea Res., 3 1 ,1357-1367, 1984. Langdon, C., On the causes of interspecific differences in the growth- irradiance relationship for phytoplankton. II. A general review, J. Plankton Res., 1 0 ,1291-1312,1988. Large, W. G., J. C. McWilliams, and S. C. Doney, Oceanic vertical mixing: A review and a model with a nonlocal boundary layer parameterization, Reviews o f GeophysicSi 3 2 ,363-403,1994. Larsson, U., and A. Hagstrttm, Phytoplankton exudate release as an energy source for the growth of pelagic bacteria, Mar. Biol., 5 2 ,199-206, 1979. 250 Lars son, U., and A. Hagstrtfm, Fractionated phytoplankton primary production, exudate release and bacterial production in a Baltic eutrophication gradient, Mar. Biol., 67,57*70,1982. Laws, E. A., and T. T. Bannister, Nutrient- and light-limited growth of Thalassiosira fluviatilis in continuous culture, with implications for phytoplankton growth in the ocean, lim nol. Oceanogr., 2 6 ,457*473, 1980. Liss, P. S., and L. Merlivat, Air-sea gas exchange rates: Introduction and synthesis, in The Role of Air-sea exchange in Geochemical Cycling, edited by P. Buat-Menard, pp. 113-129, D. Reidel, Hingham, M assachusetts, 1986. Liu, W. T., T. V. Blanc, T. Liu, K. Katsaros, and J. Businger, Bulk atmospheric flux computational iteration program in FORTRAN and BASIC, Memor. Rep. 5291, Naval Research Lab, Washington, D. C., 1984. Lohrenz, S. E„ G. A. Knauer, V. L. Asper, M. Tuel, A F. Michaels, and A. H. Knap, Seasonal variability in primary production and particle flux in the northwestern Sargasso Sea: U. S. JGOFS Bermuda Atlantic Time-series Study, Deep-Sea Res., 3 9 ,1373-1391,1992. Longhurst, A. R., Role of the marine biosphere in the global carbon cycle, lim n o l. Oceanogr., 3 6 ,1507-1526,1991. 251 Longhurst, A. R., A. Bedo, W. G. Harrison, E. J. H. Head, E. P. Home, B. Irwin, and C. Morales, NFLUX: A test of vertical nitrogen flux by diel m igrant biota, Deep-Sea Res., 3 6 ,1705-1719,1989. Longhurst, A. R., and H. W. G., Vertical nitrogen flux from the oceanic photic zone by diel m igrant zooplankton and nekton, Deep-Sea Res., 36, 881-889, 1988. M arra, J., Diurnal variability in chlorophyll fluorescence: Observations and modeling, Ocean Optics, 11,1992. M arra, J., R. R. Bidigare, and T. D. Dickey, Nutrients and mixing, chlorophyll and phytoplankton growth, Deep-Sea Res., 3 7 ,127-143, 1990. M arra, J., T. Dickey, W. S. Chamberlin, C. Ho, T. Granata, D. A. Kiefer, C. Langdon, R. Smith, K. Baker, R. Bidigare, and M. Hamilton, The estimation of seasonal primary production from moored optical sensors in the Sargasso Sea, J. Geophys. Res., 9 7 ,7399-7412,1992. M arra, J., and K. Heinemann, Photosynthesis response by phytoplankton to sunlight variability, lim nol. Oceanogr., 2 7 ,1141-1153,1982. M artin, J. H., and S. E. Fitzwater, Iron deficiency lim its phytoplankton growth in the north-east Pacific Subarctic, Nature, 3 31,341-343, 1988. 252 M artin, J. H., R. M. Gordon, S. Fitzwater, and W. Broenkow, VERTEX: PhytoplanktonAron studies in the Gulf of Alaska, Deep-Sea Res., 36, 649-680, 1989. M artin, J. H., G. A. Knauer, D. M. Karl, and W. W. Broenkow, VERTEX: Carbon cycling in the northeast Pacific, Deep-Sea Res., 3 4 ,267-285, 1987. M artin, P. J., Simulation of the mixed layer a t OWS November and Papa with several models, J. Geophys. Res., 9 0 ,903-916,1985. McCave, I. N., Vertical flux of particles in the ocean, Deep-Sea Res., 2 2 ,491- 502, 1975. McClain, C. R., J. Ishizaka, and E. E. Hofmann, Estim ation of the processes controlling variability in phytoplankton pigment distributions on the southeastern U.S. continental shelf, J. Geophys. Res., 95, 20,213-20235,1990. Mellor, G. L., Analytic prediction of the properties of stratified planetary surface layers, J. Atmos. Sciences, 3 0 ,1061-1069,1973. Mellor, G. L., and P. A. Durbin, The structure and dynamics of the ocean surface mixed layer, J. Phys. Oceanogr., 5 ,718-728,1975. Mellor, G. L., and H. J. Herring, A survey of the mean turbulent field closure models, AIAA Journal, 1 1 ,590-599,1972. Mellor, G. L., and T. Yamada, A hierarchy of turbulence closure models for planetary boundary layers, J. Atmos. Sciences, 3 1 ,1791-1806,1974. Mellor, G. L., and T. Yamada, Development of a turbulence closure model for geophysical fluid problems, Rev. o f Geophys. and Space Physics, 20, 851-875, 1982. Menzel, D. W., and J. H. Ryther, The annual cyde of primary production in the Sargasso Sea off Bermuda, Deep-Sea Res., 6 ,351-367,1960. Menzel, D. W., and J. H. Ryther, Annual variations in primary production of the Sargasso Sea off Bermuda, Deep-Sea Res., 7 ,282-288,1961. Michaels, A. F., N. R Bates, K. 0 . Buesseler, C. A. Carlson, and A. H. Knap, Carbon-cycle imbalances in the Sargasso Sea, Nature, 3 7 2 ,537-540, 1994a. Michaels, A. F., A. H. Knap, R. L. Dow, K Gundersen, R. J. Johnson, J. Sorensen, A. Close, G. A. Knauer, S. E. Lohrenz, V. A. Asper, M. Tuel, and R. Bidigare, Seasonal patterns of ocean biogeochemistry a t the US JGOFS Bermuda Atlantic time-series study site, Deep-Sea Res., 41, 1013-1038, 1994b. Michaels, A. F., and M. W. Silver, Primary production, sinking fluxes and the microbial food web, Deep-Sea Res., 3 5,473-490,1988. Millero, F. J., and A. Poisson, International one-atmosphere equation of state of seawater, Deep-Sea Res., 28A, 625-629,1981. Mobley, C. D., Light and Water: Radiative Transfer in Natural Waters, 592 pp., Academic Press, New York, 1994. 254 Morel, A., Optical modeling of the upper ocean in relation to its biogenous m atter Content (Case I waters), J. Geophys. Res., 9 3 ,10,749-10,768, 1988. Morel, A., Light and marine photosynthesis: A spectral model with geochemical and dimatological implications, Prog. Oceanogr., 2 6 ,263- 306,1991. Morel, A., and D. Antoine, Heating rate within the upper ocean in relation to its bio-optical state, J. Phys. Oceanogr., 2 4 ,1652-1665,1994. Morel, A., and A. Bricaud, Inherent optical properties of algal cells including picoplankton: Theoretical and experimental results, Can. Bull. Fish, Aquat. Sci., 2 1 4 ,521-559,1986. Nqjjar, R., Marine Biogeochemistry, in Climate System Modeling, edited by K. E. Trenberth, pp. 241-283, Cambridge University Press, Cambridge, 1992. Najjar, R., J. L. Sarmiento, and J. R. Toggweiler, Downward transport and fate of organic m atter in the ocean: Simulations with a general circulation model, Global Biogeochem. Cycles, 6 ,45-76,1992. Nelson, D. M., H. W. Ducklow, G. L. Hitchcock, M. A. Brzezinski, T. J. Cowles, C. Garside, R. W. Gould Jr., T. M. Joyce, C. Langdon, J. J. McCarthy, and C. S. Yentsch, Distribution and composition of biogenic particulate m atter in a Gulf Stream warm-core ring, Deep-Sea Res., 32, 1347-1369,1985. 255 Nelson, D. M., J. J. McCarthy, T. M. Joyce, and H. W. Ducklow, Enhanced near-surface nutrient variability and new production resulting from the frictional decay of a Gulf Stream warm-core ring, Deep-Sea Res., 36, 705-714, 1989. Peters, H., M. C. Gregg, and J. M. Toole, On the param eterization of equatorial turbulence, J. Geophys. Res., 9 3 ,1199-1218,1988. Platt, T„ Local phytoplankton abundance and turbulence, Deep-Sea Res., 1 9 ,183-187, 1972. P latt, T., Primary productivity in the central North Pacific: Comparison of oxygen and carbon fluxes, Deep-Sea Res., 3 1 ,1311-1319,1984. Platt, T., and D. V. Subba Rao, Prim ary production of marine microphytes, in Photosynthesis and Productivity in Different Environments, edited by pp. 249-280, Cambridge University Press, Cambridge, 1975. Powell, T. M„ P. J. Richerson, T. M. Dillon, B. A. Agee, B. J. Dozier, D. A. Godden, and L. O. Myrup, Spatial scales of current speed and phytoplankton biomass fluctuation in Lake Tahoe, Science, 189,1088- 1089, 1975. Prezelin, B. B., Diel periodicity in phytoplankton productivity, Hydrobiologia, 2 3 8 ,1-35,1992. 256 Prezelin, B. B., M. M. Tilzer, 0 . Schonfield, and C. Haese, The control of the production process of phytoplankton by the physical structure of the aquatic environment with special reference to its optical properties, Aquat. Sci., 6 3 ,136-186,1991. Price, J . F., C. N. K Mooers, and J. C. Van Leer, Observation and simulation of storm-induced mixed-layer deepening, J. Phys. Oceanogr., 8, 582-599, 1978. Price, J. F., R. Weller, and-R. Pinkel, Diurnal cycling: observations and models of the upper ocean response to diurnal heating, cooling and wind mixing, J. Geophys. Res., 9 1 ,8,411-8,427,1986. Richardson, K., J. Beardall, and J. A. Raven, Adaptation of unicellular algae to irradiance: An analysis of strategies, New Phytologist, 9 3 ,157-191, 1983. Roman, M. R., Nitrogenous nutrition of marine invertebrates, in Nitrogen in the Marine Environment, edited by E. J. Carpenter and D. G. Capone, pp. 347-384, Academic Press, New York, 1983. Roman, M. R., H. G. Dam, A. L. Gauzens, and J. M. Napp, Zooplankton biomass and grazing a t the JGOFS Sargasso Sea time series station, Deep-Sea Res., 4 0 ,883-901,1993. Sakshaug, E., D. A. Kiefer, and K. Andersen, A steady state description of growth and light absorption in the marine planktonic diatom Skeletonema costatum, Limnol. Oceanogr., 3 4 ,198-205,1989. 257 Sands trom, H., and J . A. Elliott, Internal tide and soli tons on the Scotian Shelf: A nutrient pump a t work, J. Geophys. Res., 8 9 ,6415-6426, 1984. Sarmiento, J. L., M. J. R Fasham, U. Siegenthaler, R Najjar, and J . R Toggweiler, Models of Chemical Cycling in the Oceans: Progress Report n , Ocean Tracers Lab. Tech. Report 6, Princeton University, 1989. Sarmiento, J. L., and U. Siegenthaler, New production and the global carbon cycle, in Primary Productivity and Biogeochemical Cycles in the Sea, edited by P. G. F. a. A. H. Woodhead, pp. 317-332, Plenum Press, New York,-1992. Sarmiento, J. L., R D. Slater, M. J. R Fasham, H. W. Ducklow, J. R. Toggweiler, and G. T. Evans, A seasonal three-dimensional ecosystem model of nitrogen cycling in the North Atlantic euphotic zone, Global Biogeochem. Cycles, 7,417-450,1993. Sathyendranath, S., L. Lazzara, and L. Prieur, Variations in the spectral values of specific absorption of phytoplankton, Limnol. Oceanogr., 32, 403-415, 1987. Siegel, D. A., T. D. Dickey, L. Washburn, M. K. Hamilton, and B. G. Mitchell, Optical determination of particulate abundance and production variations in the oligotrophic ocean, Deep-Sea Res., 3 6 ,211-222,1989. Siegel, D. A., T. C. Granata, A. F. Michaels, and T. D. Dickey, Mesoscale eddy diffusion, particle sinking, and the interpretation of sediment trap data, J. Geophys. Res., 9 5 ,5305-5311,1990a. 258 Siegel, D. A., R. Iturriaga, R. R. Bidigare, R. C. Smith, H. Pak, T. D. Dickey, J. M arra, and K S. Baker, Meridional variations of the springtime phytoplankton community in the Sargasso Sea, J. Mar. Res., 4 8 ,379- 412, 1990b. Slater, R. D., J. L. Sarmiento, and M. J. R. Fasham, Some param etric and structural simulations with a three-dimensional ecosystem model of nitrogen cycling in the North Atlantic euphotic zone, in Towards a Model o f Ocean Biogeochemical Processes, edited by G. T. Evans and M. J. R. Fasham, pp. 261-294, Springer-Verlag, New York, 1993. Smith, R. C., 0 . B. Brown, F. E. Hoge, K. S. Baker, R. H. Evans, R. N. Swift, and W. E. Esaias, Multiplatform sampling (ship, aircraft and satellite) of a Gulf Stream warm core ring, Applied Optics, 2 6 ,2068-2081,1987. Smith, R. C., K J. W aters, and K. S. Baker, Optical variability and pigment biomass in the Sargasso Sea as determined using deep-sea optical mooring data, J. Geophys. Res., 9 6 ,8665-8686,1991. Spinrad, R. W., A calibration diagram of specific bean attenuation, J. Geophys. Res., 9 1 ,7761-7764,1986. Spinrad, R. W., Testing of optical properties and development of application procedure for OMP-8 antifoul an to n submersible optical surfaces, Ref. Tech. Rep. 8701,44 pp., Sea Tech Inc., Corvalis, OR, 1987. Star, J. L., and J. J. Cullen, Spectral analysis: A caveat, Deep-Sea Res., 28a, 93-97,1981. 259 Staresinic, N., G. T. Rowe, D. Shaughnessey, and A. J. Williams II, Measurement of the vertical flux of particulate organic m atter w ith a free-drifting sediment trap, Limnol. Oceanogr., 2 3 ,559-563,1978. Steele, J. H., Environmental control of photosynthesis in the sea, lim nol. Oceanogr., 7 ,137-150,1962. Steele, J. H., and E. W. Henderson, The role of predation in plankton models, J. Plankton Res., 1 4 ,157-172,1992. Steele, J. J., and E. H. Henderson, The significance of interannual variability, in Towards a Model o f Ocean Biogeochemical Processes, edited by G. T. Evans and M. J. R. Fasham, pp. 237-260, Springer- Verlag, New York, 1993. Steemann Nielsen, E., and V. K Hansen, Measurements with the carbon- 14 technique of the respiration rates in natural phytoplankton populations, Deep-Sea Res., 5 ,222-230,1959. Stramska, M., and T. D. Dickey, Short-term variations of the bio-optical properties of the ocean in response to cloud-induced irradiance, J. Geophys. Res., 9 7 ,5713-5721,1992. Stramska, M., and T. D. Dickey, Modeling phytoplankton dynamics in the northeast Atlantic during the initiation of the spring bloom, J. Geophys. Res., 9 9 ,10,241-10,253, 1994. Stramski, D., and R. A. Reynolds, Diel variations in the optical properties of a marine diatom, Limnol. Oceanogr., 3 8 ,1347-1364,1993. 260 Sugimura, Y., and Y. Suzuki, A high-temperature catalytic oxidation method for the determination of non-volatile dissolved oiganic carbon in seawater by direct injection of a liquid sample, Mar. Chem., 2 4 ,105- 131,1988. Suzuki, Y., Y. Sugimura, and T. Itoh, A catalytic oxidation method for the determination of total nitrogen dissolved in seawater, Mar. Chem., 16, 83-97, 1985. Sverdrup, H. U., On condition of vernal blooming of phytoplankton, J. Com. Exp. Mer., 18, 287-295,1953. Totterdell, I. J., R. A. Armstrong, H. Drange, J. S. Parslow, T. M. Powell, and A. H. Taylor, Trophic resolution, in Towards a Model o f Ocean Biogeochemical Processes, edited by G. T. Evans and M. J. R. Fasham, pp. 71-92, Springer-Verlag, New York, 1993. Trask, R. P., M. G. Briscoe, and N. J. Pennington, Long term upper ocean study (LOTUS): A summary of the historical data and engineering test data, Tech. Report WHOI-82-53, Woods Hole Oceanographic Im titution, 1982. Vandevelde, T., L. Legendre, J. Therriault, S. Demers, and A. Bah, Subsurface chlorophyll maximum and hydrodynamics of the water column, J. Mar. Res., 4 6 ,377-396,1987. Varela, R. A., A. Cruzado, J. Tintor6, and E. G. Ladona, Modeling the deep- chlorophyll maximum: A coupled physical-biological approach, J. Mar. Res., 6 0 ,441-463,1992. Verity, P. G., Ammonia excretion rates of oceanic copepods and implications for estim ates of prim ary production in the Sargasso Sea, BO, 3, 249-283,1985. Wannikhof, R , Relationship between wind speed and gas exchange over the ocean, J. Geophys. Res., 9 3 ,10749-10768,1992. W ashburn, L., D. A. Siegel, T. D. Dickey, and M. K Hamilton, Isopycnal mixing and the distribution of particles across the North Pacific Subtropical Front, Deep-Sea Res., 3 6 ,1607-1620,1989. Waters, K. J., R C. Smith, and J. M arra, Phytoplankton production in the Sargasso Sea as determined using optical mooring data, J. Geophys. Res., 9 9 ,18,385-18,402,1994. Weiss, R. F., The solubility of nitrogen, oxygen and argon in seawater, Deep- Sea Res., 17,721-735,1970. Weller, R. A., and R E. Davis, A vector measuring current meter, Deep-Sea Res., 2 7 ,565-582,1980. Weller, R. A., D. L. Rudnick, R. E. Payne, J. P. Dean, N. J. Pennington, and R P. Trask, Measuring near-surface meteorology over the ocean from an array of surface moorings in the subtropical convergence zone, J. Atm. and Ocean Tech., 7 ,85-103,1990. Whitledge, T. E., and C. D. Wirrick, Observations of chlorophyll concentrations off Long Island from a moored in situ fluorometer, Deep-Sea Res., 3 0 ,297-309,1983. 262 Wiggert, J., T. Dickey, and T. Granata, The effect of temporal undersampling on primary production estimates, J. Geophys. Res., 99, 3361-3371, 1994. Winn, C., R. Lukas, D. Karl, and E. Firing, Hawaii Ocean Time-series, 3, University o f Hawaii, 1993. Worthington, L. V., The 18e w ater in the Sargasso Sea, Deep-Sea Res., 5, 297-305, 1959. 263
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
Asset Metadata
Core Title
00001.tif
Tag
OAI-PMH Harvest
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC11257195
Unique identifier
UC11257195
Legacy Identifier
9621646