Microsoft Word - GTF_Dissertation_Full 082911_FINAL.docx

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content UNDERWATER HYPERSPECTRAL OPTICAL MEASUREMENTS AS A TOOL FOR CHARACTERIZING THE SPATIAL-TEMPORAL DISTRIBUTION OF WATER COLUMN CONSTITUENTS CONTRIBUTING TO OCEAN COLOR by Gerardo A. Toro-Farmer A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL OF THE UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirement for the Degree DOCTOR OF PHILOSOPHY (OCEAN SCIENCES) December 2011 Copyright 2011 Gerardo A. Toro-‐Farmer ii ACKNOWLEDGMENTS I thank Burt Jones for his immense support during all these years, guiding me to fulfill all academic and personal requirements of a PhD. Dale Kiefer, for his input and support since the first day in Graduate school. I thank Doug Hammond and Doug Capone for their valuable input during the early stages of this dissertation. Also, I want to thank John Heidelberg for his commitment during the culmination of my studies. I am very thankful to all the members of the Jones’ and Kiefer’s Labs: Matthew, Ivona, Bridget, Vardis and Tim. I would like to acknowledge all the institutions supporting me at different stages of my doctorate: NASA Headquarters for the Earth System Science Fellowship to work in my Bermuda project; NASA and the University of Maine for the fellowship to participate in the course “Application of Remote Sensing and In-situ Ocean Optical Measurements to Ocean Biogeochemistry”; the Bermuda Biodiversity Project (BREAM) - Bermuda Zoological Society, and the Bermuda Government - Department of Conservation Services, for their support during the Bermuda project; the Wrigley Institute for Environmental Studies for the Rose Hill summer fellowships; the Naval Research Laboratory at Stennis Space Center and their Bio-Optical Physical Processes and Remote Sensing Section for providing invaluable support during the Philippine cruises and for the training and support with the satellite images at Stennis; the Marine Science Institute (University of the Philippines) provided great logistics and scientific iii support; ONR for supporting the Philippine Sea Experiment (PhilEx); the University of Southern California and the Office of Graduate Programs for the “Diversity Enhancement Placement Assistance Award”. I also thank all the collaborators and additional support that facilitated my research, including optical equipment loaned as support for field work: Emmanuel Boss, Rick Gould, Sherwin Ladner, David Ruble, Heidi Dierssen, Greg Mitchell, Dariusz Stramski, Michael Twardowski and Richard Zimmerman. Infinite thanks to Thaddeus Murdoch for all his support and friendship, the seagull race and the adventures during my trips to Bermuda. Thanks to Linda Bazilian, Don Bingham, Adolfo de la Rosa and Cyndi Waite for years and years of continuous support. To my Parents, my sister, and my nephews, thanks for being always there for me even if I am hundreds of miles away. They have always supported my most crazy ideas, including the one of becoming a Marine Biologist / Oceanographer. Of course there are no enough words to thank my two girls Isa and Susana, I wouldn’t be writing these lines without their support and company, without their love, care and smiles (or cries), without them being them. iv TABLE OF CONTENTS Acknowledgments ii List of Tables v List of Figures vi Abstract x Chapter 1: General Introduction 1 Global significance of the Coastal Ocean 2 Marine Optics 3 Inherent optical properties (IOPs) 4 Apparent optical properties (AOPs): 5 Derived properties from optical measurements 7 Overall Objective 9 Chapter 2: Characterization of the in situ particle spectra in an archipelago strait using hyperspectral IOP measurements from a towed vehicle 11 Chapter 3: Spatial gradients of water column optical properties across a coral reef lagoon in Bermuda Islands 47 Chapter 4: In situ quasi-synoptic optical measurements as tools for validating MODIS 250m ocean color products in a highly dynamic coastal area 74 Summary 103 References 107 v LIST OF TABLES Table 2.1. Mean values (and standard deviation SD) for chlorophyll concentration (C a(acs) ), beam attenuation (c 650 ), gamma (γ) and the ratio of B:R as indicative of detritus prescence, for each one of the groups of a p and b p . N indicates the number of values included in the calculation. 37 Table 2.2. Range of values for chlorophyll concentration (C a(acs) ) and beam attenuation (c 650 ) for the different sub-areas, by spectral group during all the tidal stages. 38 Table 3.1. Geographic position, depth, substrate and distance to the navigation channel for the sampled stations. Depth corresponds to the estimated station depth for a Tidal level = 0m. Currents directions and speed corresponds to the mean value at mid depth during the period each station was surveyed. 57 Table 4.1. Correlations between in situ observations and satellite estimated values for absorption at 412 nm (a 412 ), and between in situ CDOM fluorescence and satellite a 412 . Numbers in parenthesis are the sample number N. 89 Table 4.2. Correlations between in situ observations of total attenuation at 650 nm (c 650 ) (from the AC-s and transmissometer) vs. the satellite estimated c 645 . Numbers in parenthesis are the sample number N. 90 Table 4.3. Correlations between in situ observations of chlorophyll concentration (C a ) and chlorophyll fluorescence vs. the estimated absorption by phytoplankton (a ph(443) ) from the different ocean color models. Numbers in parenthesis are the sample number N. 91 vi LIST OF FIGURES Figure 2.1. A) The Philippine Island archipelago showing the location of the study area. The image is a 5-day mosaic of chlorophyll concentration from the MODIS-Aqua sensor. B) Sampling transect (dashed line) across the San Bernardino strait. 17 Figure 2.2. TRIAXUS on deck ready to be deployed showing the AC-s and CTD locations. The rest of the optical instruments were mounted inside the vehicle’s body. 18 Figure 2.3. Quasi-synoptic water column profiles in San Bernardino strait during different tidal stages: mid ebb state (ME), late ebb (LE), early flood (EF), mid flood (MF1 and MF2), and late flood (LF). T: Temperature; Sal: salinity; σ θ : density; DO: dissolved oxygen; Chl-f: chlorophyll fluorescence; Atten: beam 23 Figure 2.4. Total absorption a T (minus water absorption) as a function of wavelength for the different areas (A1-5) during the six tidal stages (ME, LE, EF, MF1, MF2, LF). 28 Figure 2.5. Total scattering b T as a function of wavelength for the different areas (A1-5) during the six tidal stages (ME, LE, EF, MF1, MF2, LF). 29 Figure 2.6. Cluster analyses results for particulate absorption (a p ). (A) dendrogram showing the three groups and their subgroups. (B) particulate absorption spectra for each group separated by the cluster analysis. Grey lines represent individual spectra curves, and the black line represents the mean spectra values. (C) First derivative for the particulate absorption spectra as a function of wavelength. (D) Second derivative for the particulate absorption spectra as a function of wavelength. 31 vii Figure 2.7. Cluster analyses results for particulate scattering (b p ). (A) dendrogram showing the four groups. (B) particulate absorption spectra for each group separated by the cluster analysis. Grey lines represent individual spectra curves, and the black line represents the mean spectra values. (C) First derivative for the particulate scattering spectra as a function of wavelength. (D) Second derivative for the particulate scattering spectra as a function of wavelength. 32 Figure 2.8. Water column distribution of the derived products during the different tidal stages. C a(acs) : chlorophyll concentration from the AC-s. B:R: blue to red ratio as a proxy for detritus. γ: total attenuation spectral slope. a p _goups and b p _groups: absorption and scattering groups respectively separated by the cluster analyses. 35 Figure 2.9. (A) Mean absorption spectra by tide at different depths on the water column for the border and center of the San Bernardino channel. (B) the same mean absorption spectra normalized to 676 nm. 40 Figure 2.10. Temperature-salinity plot for the San Bernardino survey color-coded by the particulate absorption spectral groups. 44 Figure 3.1. The Bermuda Islands showing sampling stations (black dots, 1 to 11). Double dashed lines indicate the main navigation channels. 52 Figure 3.2. Total attenuation (A), dissolve absorption (B), particulate absorption (C), normalized absorption to 676 nm (D), particulate scattering (E) and particulate backscattering (F) as a function of wavelength for all sampling stations. 59 Figure 3.3. Particulate absorption (A to C), first derivative (D to F), and second derivative (G to I) for the spectral groups separated by the cluster analyses as a function of wavelength. Grey lines represent individual spectra curves, and the black line represents the mean spectra values. Dotted lines are ± two times de standard deviation. 60 viii Figure 3.4. Particulate scattering (A to C), first derivative (D to F), and second derivative (G to I) for the spectral groups separated by the cluster analyses as a function of wavelength. Grey lines represent individual spectra curves, and the black line represents the mean spectra values. Dotted lines are ± two times de standard deviation. 62 Figure 3.5. Mean values by station for A) dissolve absorption at 400 nm, B) the slope of the dissolved absorption spectra, C) particulate absorption, D) slope of the particulate absorption, E) chlorophyll concentration derived from absorption, F) blue to red ratio, and G) backscattering to scattering ratio. Vertical bars represent ± standard deviation. 64 Figure 3.6. Mean values by station for A) volume concentration of particles by station, and B) volume concentration as a function of Particle size distribution (PSD). Vertical bars represent ± standard deviation. 67 Figure 3.7. Linear regressions between water level and A) dissolved absorption a t 400 nm, B) chlorophyll concentration, C) backscattering to scattering ratio, D) blue to red ratio, and D) volume concentration of suspended particles. 68 Figure 3.8. Quickbird satellite image showing a boat leaving a plume of resuspended sediments along a navigation channel in the Great Sound. The plume (indicated by the arrow) remains in the water column for several minutes, as suggested by this image. 72 Figure 4.1. A) Philippine Islands. B) Map showing the location of the transect (dashed line) used for the in situ deployment of the TRIAXUS in San Bernardino Strait. 81 Figure 4.2. Spatial distribution of in situ water column measurements for: A) absorption at 412 nm (a 412 ), B) CDOM fluorescence (CDOM), C) total attenuation at 650 nm from the AC-s (c 650 ), D) total attenuation from the transmissometer (c 650(Trans) ), E) chlorophyll concentration from absorption (C a_acs ), and F) chlorophyll fluorescence (chl_f). Transects are oriented from west (left side) to east (right side). 87 ix Figure 4.3. A) Correlation between in situ vs. remotely sensed estimated dissolve absorption at 412 nm (a 412 ). B) Correlation between in situ chlorophyll concentration (C a_acs ) and the chlorophyll concentration estimated with the standard ocean color OC3 algorithm for MODIS (C a(sat) ). 92 Figure 4.4. Horizontal distribution along the channel of in situ and estimated products from MODIS: A) absorption at 412 (a 412 ), B) beam attenuation c 650 and c 645 , C) in situ chlorophyll concentration (C a_acs ) and satellite phytoplankton absorption at 443 (a ph443 ), and D) in situ chlorophyll concentration (C a_acs ) and the chlorophyll concentration estimated with the standard ocean color OC3 algorithm for MODIS (C a(sat) ). 93 Figure 4.5. Total absorption at 412 nm (a 412 ) estimated with the GSM algorithm for: A) February 16, B) February 18, and C) February 19. The yellow line represents the in situ transect. 95 Figure 4.6. Absorption by phytoplankton at 443 nm (aph(443) estimated with the Carder mode for: A) February 16, B) February 18, and C) February 19. 96 Figure 4.7. Chlorophyll concentration (C a ) estimated with the OC3 MODIS algorithm for: A) February 16, B) February 18, and C) February 19. 97 x ABSTRACT In coastal areas the spectral composition and intensity of light can change significantly in response to the optical properties of the water across horizontal distances and over different periods of time. Resuspension of particulate matter, transformations of the chemical and physical characteristics of the suspended matter, release of organic carbon from suspended particles, changes in phytoplankton biomass and species composition, and alteration of natural environments by human activities, all contribute to the optical complexity of the water in coastal areas. Due to the high productivity of this area and its importance in the global ocean there is a need to better characterize water optical properties in coastal areas and, hence improve our understanding of how the coastal oceans respond to short-term events, a variety of natural processes and human activities, over different spatial ranges. This will lead to better monitoring of long-term changes due to climate and sea level change. In order to understand the distribution of hyperspectral optical properties in coastal areas in space and time, a better understanding of their relationship to processes occurring in the water column is needed but the bio-optical complexity and physical dynamics of coastal areas impose a limitation. In this study new tools and approaches were tested taking into account the complex distribution of optical properties associated with the physical and biological environment. In particular, the spatial and temporal variations of in situ hyperspectral optical properties and hydrographic observations from two contrasting island environments were studied: a dynamic strait in the Philippine Islands and a shallow coral reef area from Bermuda Islands. In the Philippines remote xi sensing validation was also studied to provide a tool for monitoring straits at a larger spatial and temporal scale. Specifically this study focused on optical proxies for biogeochemical constituents including the concentration of total suspended particulates, concentration of colored dissolved organic matter (CDOM), chlorophyll a concentration, and particle size distribution. In the Philippines, we tested the use of quasi-synoptic observations achieved with an undulating towed vehicle to characterize the water constituents and distribution of water masses in response to strong tidal forcing in a highly dynamic strait. The hyperspectral optical properties and their derived products were used to successfully characterize the transport and exchange of dissolve and particulate matter throughout semidiurnal and diurnal tidal fluctuations. Statistical analyses of the hyperspectral data set provided discrimination of water masses within the strait. These differences allowed delineating particular water masses coming in and out the strait with the tidal cycle, carrying particulate materials with characteristic spectral properties. The general pattern observed was a high oxygen level, small suspended particles, and low chlorophyll concentration in the oceanic waters entering the strait during flood tide. Also in the Philippines, we evaluate the quality of remotely sensed ocean color using the hyperspectral in situ inherent optical properties. The use of MODIS images at 250 m resolution was validated for this area using the Carder, QAA, GSM, Gould and OC3 standard ocean color models previously developed mainly for open ocean areas. We evaluated the validity of using MODIS imagery to study a region at high spatial and temporal resolution to measure the distributions of dissolved and suspended particulate xii matter, and chlorophyll-a concentration. Ocean color products derived from MODIS data positively correlated with the tridimensional in situ observations gathered in the San Bernardino strait. Satellite products were able to resolve the intrusion of oceanic water into the strait during flood tide, and the seaward transport of interior water during the ebbing tide. This study demonstrated for the first time the utility of using 250 m resolution MODIS images to resolve inherent optical properties and biogeochemical variables in a strongly forced coastal archipelago strait. Giving the limited in situ observations for the Philippines this validation study provides promises for monitoring at large spatial and temporal scales hyperspectral optical properties of straits. In the Bermuda Islands, we characterize spatial gradients of hyperspectral optical properties to estimate suspended materials, dissolved matter, particle size distribution and phytoplanktonic pigments in the water column. This enables the identification of the processes that primarily cause spatial variation in the optical characteristics from inland to the open ocean in this poorly characterized coral reef lagoon. A well defined gradient from inshore to the open ocean was observed, with the highest chlorophyll concentrations, dissolved absorption and high concentration of small particles close to land. Hyperspectral absorption and scattering magnitudes and shapes also discriminated between inside and outside areas. Although other processes may influence the observed optical gradient and characteristics, phytoplankton and suspended sediments are the dominant components of the spectral signatures in the water column. The hyperspectral optical measurements were able to distinguish sources with different biogeochemical signatures in this large island lagoon, indicating that ocean color observations can aid in xiii better understanding how natural and anthropogenic disturbances may disrupt shallow coastal ecosystems, and help develop and tune remote sensing algorithms for investigating spatial-temporal variations of suspended materials in shallow coastal waters. 1 CHAPTER 1 GENERAL INTRODUCTION The coastal ocean is an optically complex area where a variety of dissolved and particulate materials contribute to the visible spectral characteristics of the water (Morel and Prieur, 1977; Morel, 1988; Coble et al., 2004; Schofield et al., 2004; Mobley et al., 2004). Coastal regions are often highly dynamic due to physical processes that include tides, wind driven circulation, remotely driven processes, riverine influence and anthropogenic inputs, which through the effects of mixing and stirring generate the characteristic complexity of coastal areas. Resuspension of particulate matter, input from anthropogenic sources, transformations of the chemical and physical characteristics of suspended matter, and changes in phytoplankton biomass and species composition all contribute to the optical complexity of coastal water. The study of this spatial-temporal variability of optical properties in coastal areas and the physical factors that determine their distribution will help understand how natural and anthropogenic disturbances affect coastal systems. Optical measurements have been used to study the behavior of suspended materials, their particle size distribution, and their biogeochemical origins (e.g. Babin et al, 2003; Boss et al., 2009; McKee et al., 2009; Para et al., 2010; Reynolds et al., 2010; Jones et al., 2011). Bulk optical properties can elucidate processes in the water column of coastal environments. In the research described in this dissertation a combination of in situ and 2 remote optical measurements were used to examine the spatial and temporal dynamics of water column constituents in two distinct island regions where previous in situ studies are scarce - straits within the Philippine archipelago and the Bermuda Island lagoon. Global significance of the Coastal Ocean Approximately 26% of the total primary production in our planet and about 85% of the fisheries production occurs in the coastal ocean (Rao et al., 2008). Very dynamic and complex ecosystems can develop in coastal areas (e.g. straits, embayments, headlands, coral reefs, seagrasses), depending on local topography and geographic features, physical processes, biota, and general environmental conditions. Anthropogenic stressors affect many coastal ecosystems around the world. In the Western Atlantic, the Caribbean, the Indo-Pacific and Southern California, coastal development and dredging associated with construction of hotels, beach replenishment, deepening of navigation channels and harbors increase suspended materials. Modification of the underwater light field by these processes may compromise entire ecosystems, and eventually resulting in the death of many organisms (Rogers, 1990; Edinger et al., 1998; Barnes and Lough, 1999; Szmant, 2002; Fortes, 2004). Straits are constrictions usually between two adjacent land masses that channelize the flow of local currents and tides. Due to their topographic characteristics, straits can create complex water circulation patterns resulting in spatial and temporal exchanges of suspended materials critical for biological processes. San Bernardino in the Philippine Islands is an example of a highly dynamic strait, where tidal currents have been reported 3 to be ~3 m•s -1 and possibly higher (Jones et al., 2011); these strong tidal curents, in combination with underwater topographic features generate transport and mixing of nutrients into the upper water column contributing to higher phytoplanktonic biomass and productivity in the area (Jones et al., 2011). Coral reefs cover over 600,000 km 2 globally, approximately a third of the tropical coastlines of the planet (Birkeland, 1997). Around 800 species of scleractinian corals, 4,000 species of fish and thousands of other species have been described to date, demonstrating why coral reefs are considered among the most diverse ecosystems on earth (Paulay, 1997). Seagrasses also cover extensive coastal areas and are more widely distributed, from the temperate northern to the temperate southern oceans including all the tropical coasts in the world (Short et al., 2007). Although there are few species of seagrasses (about 60 species), the geographic distribution of a single species can cover thousands of kilometers (Short et al., 2007). Coral reefs and seagrass beds may moderate global climate and biogeochemical cycles and yet are simultaneously affected by them. These ecosystems provides valuable human services including seafood production, pharmaceutical products, recreation and coastal protection, estimated to have an annual value of about $375 billion and $3801 billion for coral reefs and seagrass beds respectively (Costanza et al. 1997). Marine Optics Light from the sun is absorbed or scattered at specific wavelengths by materials in the atmosphere before reaching the earth’s surface (Horvath, 1993; Mobley, 1994). Two 4 primary factors reducing sunlight transmittance through the atmosphere are water vapor and carbon dioxide (Horvath, 1993; Schowengerdt, 2007). Atmospheric aerosols are the most important absorber in the visible range of the spectrum (Horvath, 1993). Once these photon fluxes enter the water column they undergo additional spectrally dependent absorption and scattering, resulting in the optical properties of the water column. These optical properties of water are divided in two classes: inherent and apparent optical properties (IOPs and AOPs respectively) (Mobley, 1994). Inherent optical properties (IOPs): IOPs depend on the amount and biogeochemical characteristics of the dissolved and particulated materials in the water, as well as the water itself. Two fundamental IOPs in marine optics are the absorption (a) and the scattering (b) coefficient functions (Mobley, 1994; Zaneveld et al., 2005), which are expressed as: ! (!) ≡ !"# ∆!→! ! ! ! ! ! ! ∆! (1) and ! (!) ≡ !"# ∆!→! ! ! ! ! ! ! ∆! (2) where ∆r is the thickness of a volume of water through which a light beam will pass, Φi is the radiant flux or collimated beam of light, Φa and Φs are the spectrally dependent absorbed and scattered parts of the collimated beam in the volume of water, 5 and λ is the wavelength of the light (Kirk, 1994; Mobley, 1994). Total absorption in the water column is the sum of the individual absorption coefficients of water constituents: a T(λ) = a w(λ) + a s(λ) + a dom(λ) + a ph(λ) + a NAP(λ) + a r(λ) (3) were w is water, s is sea salts, dom is the dissolved organic matter, ph is live phytoplankton, NAP is nonalgal particles, and r are the remaining parts such as air bubbles (Mobley, 1994; Wozniak & Dera, 2007). Total scattering coefficient is also the sum of the mentioned water constituents, except for the dissolved fraction (Mobley, 1994). In order to identify the concentration and physical characteristics of materials in the water column by optics, a (λ) and b (λ) must be determined. Once these parameters have been measured or calculated, the total attenuation of light (c T ) can be determined as: c T(λ) = a T(λ) + b T(λ) (4) Apparent optical properties (AOPs): Like IOPs, AOPs depend on the materials in the water, but in addition they depend on the geometrical structure of the light field (directional beam of light). AOPs take into account radiance (L) and irradiance (E), which are radiometric measurements of the light field in the water (Mobley, 2004; Zaneveld et al., 2005): ! !,∅ = ! ! ! !"!"#!"# (5) 6 where θ and ∅ indicate the direction of the beam in terms of the zenith and azimuth angles respectively, Φ is the radiant flux or collimated beam of light, dS is the area of a small element of surface, and dω is the solid angle (Kirk, 1994). Unlike radiance (L), which considers a beam with a specific direction (specific elevation and azimuth), irradiance (E) can be defined as the integration of L for all solid angles; for example in the upper hemisphere, the downwelling irradiance (E d ) is defined as: ! ! = ! !,∅ !"#$%& !! (6) When both hemispheres are considered (upwelling and downwelling), the Scalar Irradiance (E 0 ) is calculated as (Kirk, 1994): ! ! = ! !,∅ !" !! (7) Two of the most common derived AOP variables with practical applications are the diffuse coefficient of attenuation (K d ) and the remote sensing reflectance (R rs ) just above the water (z~a) (Mobley, 1994): ! !(!,!) = − !!"! !(!,!) !" (8) and ! !"(!,∅,!) = ! (!~!,!,∅,!) ! !(!~!,!) (9) 7 For remote sensing studies, AOPs and IOPs can be related as: ! !"(!) =! ! !(!) ! (!) ! ! !(!) (10) where G is a constant dependent on the volume scattering function and the light field, and b b is the scattering in the backward direction (Schofield, 2004 and references within). Derived properties from optical measurements: Important biogeochemical characteristics of the water column constituents can be studied with IOPs and AOPs measurements. One of the most common indexes derived from optical measurements is the presence and concentration of photosynthetic and photoprotective pigments from phytoplankton cells in the water column. Chlorophyll-a (Chl-a) is the most abundant pigment in the ocean, and can be identified by its characteristic covariant absorption peaks around 446 and 676 nm (Jeffrey et al., 1997). Other pigments can be identified by their characteristic absorption peaks with multi- or hyper- spectral measurements, facilitating the identification of different taxonomical groups in mixed communities (e.g. Bidigare et al. 1989; Cleveland and Perry 1994; Eisner et al., 2003; Chazottes et al. 2007). Other properties of phytoplanktonic groups than can be studied with optics include size distribution of planktonic cells using beam c attenuation with a particle size analyzer (Karp-Boss et al., 2007), and life cycles and community dynamics of phytoplanktonic species based on their scattering properties (Nencioli et al., 2010). 8 Marine optics has been fundamental in studying global primary productivity and the carbon cycle (e.g. Behrenfeld and Falkowski 1997; Gardner et al., 2006), helping in the creation of global productivity maps and the better understanding of global biogeochemical cycles. Primary productivity in the ocean is often limited by nutrient availability, which can be linked to availability of dissolved organic matter. The distribution and chemical properties of dissolved organic matter in the ocean can be traced with optics (Del Castillo, 2005; Wozniak & Dera, 2007). Colored dissolved organic matter (CDOM) absorbs most strongly in the UV portion of the light, but CDOM absorption extends into the blue portion of the visible spectrum (~400 nm), overlapping with the phytoplankton pigment absorption (Coble et al., 2004; Twardowski et al., 2004). The combination of in situ optical measurements and remote sensing analysis has facilitated the identification and classification of water masses in the open ocean and coastal systems (Arnone et al., 2004; Schofield et al., 2004; Jones et al., 2011). The combined use of in situ and remotely sensed optics has advanced the understanding of local coastal processes associated with shallow ecosystems and also the interactions of these coastal processes with the open ocean and adjacent coastal regions. Empirical algorithms for remote sensing analyses are based on in situ optical sets for specific geographic areas. Although these algorithms have been tested for several regions showing good correlation between in situ and remotely sensed measurements (IOCCG, 2006; Dupouy et al., 2010; Hu et al., 2010; Volpe et al., 2011), additional calibrations of remotely sensing algorithms with in situ measurements are needed for improved accuracy of remotely sensed ocean properties. 9 Overall Objective For the research described in this thesis, I investigated the spatial and temporal variations of in situ optical variables and satellite remote sensed data from two contrasting island environments: a tidally forced strait in the Philippine Islands (Southeast Asia) where tidal velocities up to 4.5 m•s -1 have been observed (Peña and Mariño, 2009), and the comparatively calm, shallow coral reef area around the Bermuda Islands (Western North Atlantic). The study of the variability of inherent optical properties of these two contrasting marine systems provides insights into the ecological dynamics and physical processes in these coastal areas. The improved understanding of the biogeochemical processes in these areas provides additional validation of remote sensing algorithms, fundamental for future management and conservation efforts in coastal shallow ecosystems. Chapter 2 of this dissertation focuses on characterizing the shape, amplitude and peaks of the hyperspectral signatures of absorption and scattering measurements in relation to biogeochemical components and water column processes in a very dynamic strait system from the Philippines Islands. The highly resolved spatial distribution of hyperspectral IOPs enables the discrimination of water masses on the basis of the carried suspended materials with specific optical characteristics. Chapter 3 deals with the spatial distribution of water column IOPs measured at different distances from land in the Bermuda Islands crossing the main benthic 10 ecosystems in the area. The role of the particulate and dissolved components in light attenuation was analyzed and the relation of the estimated IOPs with tidal stages was established. A well defined gradient was found for CDOM and chlorophyll decreasing from land to the open ocean. These observations not only add to the scarce knowledge of optical properties and physical dynamics around the Bermuda reefs, but also set a baseline for monitoring the impacts of future development in the area. Chapter 4 correlates in situ optical measurements from the Philippines with remotely sensed data at different space and time scales. In the Philippines, a towed vehicle equipped with optical sensors provided a quasi-synoptic view of the optical properties in the highly dynamic San Bernardino Strait. These measurements correlated well with 250 m resolution satellite data processed using different ocean color algorithms. 11 CHAPTER 2 CHARACTERIZATION OF THE IN SITU PARTICLE SPECTRA IN AN ARCHIPELAGO STRAIT USING HYPERSPECTRAL IOP MEASUREMENTS FROM A TOWED VEHICLE 2.1 ABSTRACT Optical properties in coastal areas and especially in straits can be highly variable and difficult to study, due to strong currents, tides and underwater or coastal topographic features. In order to characterize the spatial-temporal distribution of particulate spectral signatures and the main biogeochemical components in a highly dynamic strait, a towed profiling vehicle was deployed following a cross-channel transect during a full tidal cycle. The observed spectra for absorption and scattering were grouped based on their similarities and collocated with the distribution of other environmental variables, showing good correlations between specific spectral groups and chlorophyll concentration, beam attenuation, and particle size. Differences in spectral characteristics and biogeochemical variables were identified for the different tidal stages and for areas within the channel. These differences allowed delineating particular water masses coming in and out the strait with the tidal cycle, carrying particulate materials with characteristic spectral properties. 12 2.2 INTRODUCTION Inherent optical properties (IOP’s) depend on the amount and composition of dissolve and suspended materials in the water, as well as on the optical characteristics of the water itself. In coastal areas, IOP’s can vary over short distances (vertical and horizontally) due to tidal cycles, regional circulation patterns, environmental processes and physical barriers or bathymetric features in the area. Thus, coastal areas can be highly dynamic and optical properties can be used as tracers for physical and biogeochemical processes and water column components (Babin et al., 2003; Boss et al., 2001; Cunningham et al., 2003; Oliver et al., 2004). Open ocean water is considered optically clear, where the main component contributing to the bulk of spectral characteristics is phytoplankton, while other suspended materials (dissolve or particulate) are found in low concentrations with minimum optical signatures covarying with the phytoplankton (Bricaud et al., 2004; Dall'Olmo et al., 2009; Stramski et al., 2001). Conversely in coastal regions, spectral signals are a mix of suspended and resuspended materials, organic and inorganic in dissolve or particulate states, either from the bottom, from surrounding land masses or from biological processes in the water column. Local phytoplankton species with different scattering (from cell shells) and absorbing properties (from associated pigments) can have overlapping spectral signatures (Bricaud et al., 2004; Chazottes et al., 2006; Ciotti et al., 2002; Dall'Olmo et al., 2009; Eisner and Cowles 2005) making it difficult to discriminate the optical contributions from individual components. 13 Several approaches have been used to rigorously model the contributions of the different water components to total absorption or scattering in open ocean and coastal areas (e.g. Bricaud and Stramski 1990; Cleveland and Perry 1994; Stramski et al., 2001), all reporting moderate to good success in their calculations. For absorption spectral analysis of natural waters, both the amplitude and shape of the spectra must be resolved to better differentiate the contributions of different pigments, detrital materials or CDOM molecules (Roesler et al., 1989). Derivative analysis of spectral measurements has been used to locate key peaks resulting from the presence of specific pigments (Bidigare et al., 1989; Chazottes et al., 2007; Faust and Norris 1985), helping find small variations in the wavelengths where absorption maxima occur for specific pigments in nature. These variations have been associated with intrinsic differences between species, their physiological state, and environmental conditions. Based on these studies, individual absorption peaks and ratios between key wavelengths have been implemented as standard proxies for the presence and concentration of biogeochemical compounds. In order to study optical properties in coastal areas, standard field methodologies impose extensive and delicate on-board sample preparations and measurements, and frequently shipping of fragile samples to land laboratories for further analysis (e.g. HPLC analysis for pigments). The sampling method and the processing of samples can limit the ability to adequately resolve temporal and spatial variability in the coastal region. In vivo approaches have resulted in the development of several in situ instruments and deploying platforms, where traditional profile systems carrying full suites of optical instrumentation are the most common. These packages provide well resolved vertical profiles of optical 14 properties including hyperspectral absorption, attenuation and scattering measurements for discrete stations (Twardowski et al., 2005), but their spatial and temporal coverage is usually limited to a few spatial points during field campaigns. Laboratory optical measurements of discrete water samples resolve the individual water constituents as well as their full spectral characteristics, including origin of the particle or dissolved materials, or the different phytoplankton pigments concentration. Unfortunately, these techniques are limited by the number of samples that can be processed, and their main utility is for cross calibration for other more autonomous sampling techniques. Newer sampling technologies such as autonomous gliders and floats provide greater spatial and temporal coverage (Cunningham et al., 2003; Johnson et al., 2009), but their instrument load and time of deployment are limited by physical space within the platform itself and the need to conserve power in order to provide for a longer deployment duration. These limitations constrain spectral measurements to just a few wavelengths used as proxies for chlorophyll or CDOM fluorescence, total beam attenuation, or optical scattering by particles. Hyperspectral sensors, although considered to be able to provide better insights of the water constituents based on more detailed spectral characterization, are not routinely deployed on autonomous platforms due to their high power consumption, high data rate, and their size and mass. Multi and hyperspectral absorption and attenuation sensors have been successfully deployed on towed vehicles with continuous power supply by the towing vessel and real time telemetry (e.g. Ackleson, 2006; Barth et al., 2005; Barth and Bogucki, 2000; Peterson and Peterson, 2008; Petrenko et al., 1997). These vehicles can cover extensive areas during long periods of time carrying not only 15 the optical payload, but also other fundamental oceanographic instrumentation; as a result, more comprehensive datasets from the study region are gathered maximizing the tridimentional sampling capabilities of the vehicle. The goal of this study is to characterize the spectral signatures (shape, amplitude, peaks) of absorption and scattering measurements as “tracers” for biogeochemical components and water column processes in a very dynamic strait system within the Philippines Islands. Our approach elucidates the utility of using hyperspectral sensors onboard towed vehicles as an efficient tool for quickly assessing real time data at different spatial-temporal scales under very complex environmental conditions where the use of other sampling platforms including autonomous vehicles, single profilers, and even traditional ship-based profiling can be limited. 16 2.3 STUDY AREA The Philippine archipelago is comprised of more than 7000 islands with a complex underwater topography, which in association with monsoonal winds create highly dynamic circulation patterns (Gordon et al., 2011). As a result, vertical and horizontal mixing of nutrients occur in the inward straits, promoting high phytoplankton productivity (Jones et al., 2011), as can be observed with satellite imagery (Fig. 2.1A). The San Bernardino strait (12°36' N, 124°12' E) is one of the two connections between the Pacific Ocean and the interior of the archipelago. Surface water from the ocean is transported over shallow sills (<100 m) through the strait mainly by the strong tides (Meñez et al., 2006; Jones et al., 2011) that can generate current speeds up to 4.5 m•s -1 , although the net flux of water from outside to the interior seems to be minimal (Gordon et al., 2011; Jones et al., 2011). These circulation patterns can generate some of the highest recorded vorticity and strain rates for the Philippines area (Ohlmann, 2011). The main channel of the San Bernardino Strait was surveyed repeatedly during February 17-18 of 2009, using a pre-defined survey pattern (Fig. 2.1B) that would provide 6 passes within 24 hours, one every 4 hours, in order to resolve tidal variations. The length of the transect was 12 km (+/- ~380 m), requiring for each pass about one hour to be completed. This sampling strategy provided the opportunity to observe different phases of the tide inside the strait where tidal velocities of up to 2.5 m•s -1 were observed (Jones et al, 2011): the first pass corresponds to a mid ebb state (ME); second pass occurred during the late ebb tide (LE); third line corresponds to the early flood (EF) 17 phase; the fourth and fifth lines are during a mid flood tide (MF1 and MF2 respectively); and finally the sixth line was done during the late flood tide (LF). Figure 2.1. A) The Philippine Island archipelago showing the location of the study area. The image is a 1 km resolution, 7-day composite (February 15 to 21, 2009) of chlorophyll concentration from the MODIS-Aqua sensor. B) Sampling transect (dashed line) across San Bernardino strait. 18 2.4 METHODS Towed Vehicle Water column measurements were obtained along the transect in a sawtooth pattern using an undulating towed vehicle (TRIAXUS - MacArtney) equipped with hydrographic and optical sensors (Fig. 2.2). The TRIAXUS was towed from the R/V Melville using a steel armored cable with power conductors and a fiber optic telemetry system allowing two-way communication and real time data display on the Figure 2.2. TRIAXUS on deck ready to be deployed showing the AC-s and CTD locations. The rest of the optical instruments were mounted inside the vehicle’s body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±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¶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`#/ $ -O ;# !4.#/*+2#64*22.,#"3#%4.#B*2#@.02*0'+2"#B%0*+%#:*$#$&01.(.'# 0.).*%.',(#3"0#/"0.#%4*2#\P#40$;#a*%.0#6",&/2#)0"3+,.$#:.0.# "5%*+2.'#:+%4#*#E*6M0%2.(#!0+*L&$#&2'&,*%+27#%":#1.4+6,.# .X&+)).'#:+%4#B.*5+0'#F!b#+2$%0&/.2%#H)0.$$&0.9#6"2'&6%+1+%(# *2'#%./).0*%&0.J9#*#'+$$",1.'#"L(7.2#$.2$"09#*,%+/.%.0$#*2'# MbF8$;#A)%+6*,#/.*$&0./.2%$#+26,&'.'W#4().0$).6%0*,#5.*/# *%%.2&*%+"2#*2'#*5$"0)%+"2# /.*$&0.'#:+%4#*#aK!#?*5$#MF- $#$.2$"0#HRO#'+33.0.2%# :*1.,.27%4$#30"/#PNN#%"#[`N# 2/9#0.$",&%+"2#cP2/Jd#5.*/# *%%.2&*%+"2#/.*$&0.'#*%#Q`N# 2/#:+%4#*#aK!#?*5$#F-B%*0# %0*2$/+$$"/.%.0d#*2'# 64,"0")4(,,#*2'#F","0-b+$$",1.'# A07*2+6#E*%%.0#HFbAEJ# 3,&"0.$6.26.#/.*$&0.'#:+%4# aK!#?*5$#aK!B%*0# 3,&"0"/.%.0$;# V+7;#\W#!0+*L&$#"2#'.6<#0.*'(#%"# 5.#'.),"(.'#$4":+27#%4.#MF-$# ,"6*%+"2;#!4.#0.$%#"3#%4.#")%+6*,# +2$%0&/.2%$#:.0.#/"&2%.'# +2$+'.# WKHY HKLFOH ¶ V ERG \ # V+7;#]W#b+$%*26.#1$;# '.)%4#3"0#.*64#"2.# "3#%4.# .21+0"2/.2%*,# )*0*/.%.0$# /.*$&0.'#+2#%4.# :*%.0#6",&/2# '&0+27#*#3&,,#%+'*,# F\ FOH 7KU HH µ $U HDV¶ :.0.#$.,.6%.'#%"# 6"1.0#%4.#/*+2# '..)#64*22.,# HM0.*#OJ#*2'#5"%4# $+'.$#6,"$.#%"# $4"0.#HM0.*$#\#*2'# ]J#HM0.*$#/*0<.'# +2#%4.#,":.0#0+74%# )*2.,#*0.#%4.#$*/.# 3"0#*,,#%4.#"%4.0# %0*2$.6%$#/*'.J;# !4.#$*/),+27#$%0*%.7(#)0"1+'.'#X&*$+-$(2")%+6#$.6%+"2$#"3#5+")4($+6*,# )*0*/.%.0$#+2#%4.#:*%.0#6",&/2#HV+7;#]J9#*$#:.,,#*$#%4.#*5$"0)%+"2# $).6%0*#%40"&74"&%#%4.#%+'*,#6(6,.#HV+7;#P#*2'#`J; F,&$%.0#*2*,($+$#:.0.#*)),+.'#%"#%4.#'*%*#$.%#+2#"0'.0#%"#3+2'#$+/+,*0+%+.$# */"27#$).6%0*,#$4*).$#30"/#'+33.0.2%#'.)%4$##+2#.*64#*0.*#HV+7;#PJ;# ACS CTD V+7;#PW#F4*27.$#+2#%4.#!"%*,#*5$"0)%+"2#$).6%0*#'&0+27#*#%+'.#6(6,.#3"0#*,,# *0.*$;#b.)%4#0*27.$#*0.#0.)0.$.2%.'#5(#'+33.0.2%#6","0$#3"0#.*64#),"%;# V+7;#`W#a*%.0#6",&/2#64*27.$#+2#%4.#$).6%0*,#*5$"0)%+"2#30"/#`PN# %"#QUN#2/;#!4.#%:"#"1.0,*+'#1.0%+6*,#,+2.$#+2#.*64#),"%#0.)0.$.2%# %4.#/.'+*2#64,"0")4(,,#3,&"0.$6.26.#H,.3%#4*2'#,+2.J#*2'#5.*/#F# *%%.2&*%+"2#)0"3+,.#1*,&.$#H0+74%#4*2'#,+2.J;#E&,%+1*0+*%.#*2*,($+$# H8FM#C#FFMJ#6"23+0/.'#4":#%4.$.#.21+0"2/.2%*,#)*0*/.%.0$# .L),*+2#%4.#1*0+*5+,+%(#"3#%4.#*5$"0)%+"2#$).6%0*#%40"&74#%4.#:*%.0# 6",&/2#+2#%4+$#:*1.,.27%4#0*27.;#!4.#$6*,.#3"0#%4.$.#,+2.$#*0.#*%# %4.#%")#"3#%4.#3+7&0.;#!4.+0#,"6*%+"2#"1.0,*(+27#%4.#6","0),"%$#'".$# 2"%#0.)0.$.2%#*#,+2<#:+%4#*2(#$).6+3+6#:*1.,.27%4#30"/#%4.#5"%%"/# $6*,.;#H="%.#%4*%#%4.#'.)%4#$6*,.#1*0+.$#5.%:..2#*0.*$J;# K55+27# ?":#!+'.# V,""'+27# Y+74#!+'.# # N;NR# # N;NQ# # N;NP# # N;N\# # N# # N;NR# # N;NQ# # N;NP# # N;N\# # N# # N;NR# # N;NQ# # N;NP# # N;N\# # N# *#H/ -O J# PNN##P`N##`NN##``N##QNN##Q`N##[NN# PNN##P`N##`NN##``N##QNN##Q`N##[NN# PNN##P`N##`NN##``N##QNN##Q`N##[NN# PNN##P`N##`NN##``N##QNN##Q`N##[NN# a*1.,.27%4#H2/J# M0.*#O# M0.*#\# M0.*#]# ___ 0-15 m ___ 15-63 m ___ 0-15 m ___ 15-62 m ___ 0-15 m ___ 15-35 m ___ 35-60 m ___ 0-30 m ___ 30-55 m ___ 0-30 m ___ 30-116 m ___ 0-30 m ___ 30-95 m ___ 95-130 m ___ 0-35 m ___ 35-50 m ___ 50-125 m ___ 0-20 m ___ 20-50 m ___ 50-85 m ___ 85-120 m ___ 0-6 m ___ 6-30 m ___ 30-60 m ___ 60-65 m ___ 0-60 m ___ 0-8 m ___ 8-70 m ___ 70-105 m ___ 0-32 m ___ 32-60 m Ȗ # K55+27# ?":#!+'.# V,""'+27# Y+74#!+'.# b.)%4#H/J# N# # PN# # RN# # O\N# # OQN# N# # PN# # RN# # O\N# # OQN# N# # PN# # RN# # O\N# # OQN# N# # PN# # RN# # O\N# # OQN# \U;N# # \[;`# # \Q;N# # \P;`# # \];N# ]`;N# # ]P;`# # ]P;N# # ]];`# # ]];N# U;N# # R;`# # R;N# # [;`# # [;N# N;N`# # N;NP# # N;N]# # N;N\# # N;NO# !#H " FJ# B*,#H)$&J# A \ #H/,C,J# F4,e3,&"#H1J# F Q`N #H/ -O J# N;\`# # N;\N# # N;O`# # N;ON# # N;N`# N;]N# # N;\R# # N;\`# # N;\]# # N;\N# FbAE#H1J# b+$%*26.#H</J# N###\###P###Q###R###ON###O\# N###\###P###Q###R###ON###O\# N###\###P###Q###R###ON###O\# M0.*#O# M0.*#\# M0.*#]# N###\###P###Q###R###ON###O\# O;P# # O;N`# # N;[N# # N;]`# # N# N# # PN# # RN# # O\N# # OQN# N# # PN# # RN# # O\N# # OQN# N# # PN# # RN# # O\N# # OQN# K55+27# ?":#!+'.# V,""'+27# Y+74#!+'.# a*1.,.27%4#H2/J# `PN###`[N###QNN###Q]N###QQN###QUN# `PN###`[N###QNN###Q]N###QQN###QUN# `PN###`[N###QNN###Q]N###QQN###QUN# `PN###`[N###QNN###Q]N###QQN###QUN# N# # PN# # RN# # O\N# # OQN# N# # PN# # RN# # O\N# # OQN# N# # \N# # PN# # QN# # RN# b.)%4#H/J# *#H/ -O J# N;N\N# # N;NO`# # N;NON# # N;NN`# # N# N;N\N# # N;NO`# # N;NON# # N;NN`# # N# N;N\N# # N;NO`# # N;NON# # N;NN`# # N# M0.*#O# M0.*#\# M0.*#]# F4,e3,&"# H1J# F Q`N # H/ -O J# F4,e3,&"# H1J# F Q`N # H/ -O J# F4,e3,&"# H1J# F Q`N # H/ -O J# F4,e3,&"# H1J# F Q`N # H/ -O J# 19 boat, as well as pre-processing and storage of the data. The boat speed while towing was maintained between 5 to 9 knots with a median speed of 6.9 knots, with a resulting TRIAXUS' vertical speed of approximately 1 m•s -1 . The vehicle was equipped with Seabird CTD sensors (pressure, conductivity and temperature), a SBE 43 dissolved oxygen sensor, altimeters and ADCPs. Optical measurements included: hyperspectral total beam attenuation and total absorption measured with a WET Labs AC-s sensor (81 different wavelengths from 400 to 750 nm, resolution ~4 nm); beam attenuation measured at 650 nm with a WET Labs C-Star transmissometer; and chlorophyll and Colored Dissolved Organic Matter (CDOM) fluorescence (hereafter chl_f and CDOM_f respectively) measured with WET Labs WETStar fluorometers. Over the 24-hour period a total of 274 profiles were done in the area, with an average of 46 profiles per pass (+/- ~5 profiles). The resulting horizontal resolution of sampling (distance between profiles) was approximately 264 m. Data Processing and Analysis CTD - AC-s time offset correction: since the CTD sensors were located outside the body in the forward position of the towed vehicle, they sampled the water at the leading edge of the vehicle. The optical instruments were located inside the middle part of the body’s vehicle and water was pumped from the front of the vehicle to the sensors through opaque black Tygon ® tubes (Fig. 2.2). In order to synchronize the AC-s data with the CTD for later temperature and salinity corrections (see AC-s data processing below) and align with the measurements from other sensors, the time offset between the CTD and the 20 other sensors was calculated comparing co-variations in temperature data from the CTD and the uncorrected absorption of light in the red part of the spectrum, which is most affected by temperature (Boss et al., 2001); we used the measured total absorption at 740 nm from the AC-s and found a time lag of 1.7 seconds relative to the CTD temperature. This time offset was applied to the data set when merging the CTD data with the other data sets. AC-s data processing: The AC-s instrument was calibrated with Milli-Q treated water at the beginning of the cruise and each time that the towed vehicle was pulled out of the water. All the measurements made during this study correspond to total absorption (a T ) and attenuation (c T ), where a T is the sum of the absorption by the dissolved (a g ), particulated (a p ) and water (a w ) fractions (a T = a g + a p + a w ). Absorption of water (a w ) was subtracted from the measurements during data processing using the calibration data, therefore the reported total absorption (a t ) for this study corresponds to: a t = a T - a w ; the discrimination between the dissolve (a g or c g ) and the particulate fractions (a p and c p ) was later calculated as described in the following section. Data from the AC-s instrument was collected at 4Hz and a median value was obtained for each one-second interval. Temperature and salinity corrections were performed according to Sullivan et al. (2006) and the coefficients derived in there for AC-s corrections; the proportional method (Zaneveld et al., 1994) was used for scattering corrections of the absorption values with 700 nm as the reference wavelength for absorption and scattering. Total scattering (b t(λ) ) was computed by subtracting a t(λ) from the corresponding c t(λ) . 21 Discrimination of dissolved and particulate components: Dissolved and particulate attenuation and absorption are routinely calculated using one or two AC-s instruments, subtracting from the total signal the filtered measurements taken when the intake water is filtered through a 0.2 µm filter; in the case of using only one AC-s, the filter can be designed to be on and off during the deployment. Due to the long deployments of the TRIAXUS, a filter was not used with the AC-s, since it can clog during extended sampling. Because filtered measurements were not obtained on the TRIAXUS vehicle, dissolved measurements taken with a second AC-s onboard the ship were correlated with the measurements of the CDOM fluorometer onboard the TRIAXUS. This second AC-s was part of an underway flowthrough system pumping water from approximately 5 m deep, with an electronic switch to turn filtered sampling on and off, resulting in several a g(λ) measurements per hour. Only concurrent measurements made during the time when the TRIAXUS depth was between 4 and 6 m were used for the correlation. The flowthrough AC-s data was smoothed with a moving average and span of 5 nm, and the spectral dissolved absorption was calculated following the exponential function: a g(λ)(uw) = a g(ref) • e -Se(λ-λref) (11) where a g(λ)(uw) is the dissolved absorption as a function of wavelength measured with the underway AC-s; a g(ref) is the absorption at a reference wavelength; and Se is the slope of the spectrum. Using a 400 as reference, we calculated a slope of 0.0124. 22 The measured TRIAXUS’ CDOM fluorescence (CDOM_f (tri) ) was then correlated with the underway a 400(uw) with a linear fit, obtaining an R-squared of 0.71 (RMSE = 0.0141): CDOM_f (tri) = 5.626 • a 400(uw) – 0.1527 (12) Based on the relation from eq. (12), the dissolved absorption for the TRIAXUS was calculated and used as the absorption reference value (a g(400)(tri) ) when calculating the TRIAXUS spectral dissolve absorption as a function of wavelength a g(λ)(tri) : a g(400)(tri) = (CDOM_f (tri) – 0.1527) / 5.626 (13) a g(λ)(tri) = a g(400)(tri) • e -0.0124(λ-λ400) (14) Beam C attenuation and fluorescence processing: Dark counts were measured for the transmissometer and fluorometers before the deployment of the vehicle and every time the vehicle was retrieved from the water in order to determine deviations from the calibration. The data from these fluorometers is reported as raw volts. Cluster Analysis: In order to group spectral data with similar shapes and amplitude, a cluster analysis (Euclidean distance) was performed to the entire dataset. For this analysis, each wavelength was treated as an independent variable. The Cophenetic correlation coefficient (Ccc) was calculated to assess the accuracy of the 23 Figure 2.3. Quasi-synoptic water column profiles in San Bernardino strait during different tidal stages: mid ebb state (ME), late ebb (LE), early flood (EF), mid flood (MF1 and MF2), and late flood (LF). T: Temperature; Sal: salinity; σ θ : density; DO: dissolved oxygen; Chl-f: chlorophyll fluorescence; Atten: beam attenuation. 24 linking dendrogram to represent the Euclidean distances between pairs of data points. The closer the cophenetic value is to 1, the stronger the correlation. Derivative Analysis: Derivative analyses were used to enhance irregularities or subtle changes in the absorption spectral shape; these irregularities have been used to help in the identification of absorption peaks due generally to pigments (see for example: Bidigare et al., 1989; Hunter et al., 2008; Kirkpatrick et al., 2000; Millie et al., 1997, 2002). The first derivative is as definition the slope between consecutive wavelengths with a predefined window span. Minimum values in the second derivative of the spectral curve indicate the presence of specific pigments or other materials in the sample. The 4th derivative enhances even more the differences between overlapping peaks, showing maximum values in those areas of the spectrum with higher absorption or scattering values. Chlorophyll estimation from absorption: Chlorophyll concentration (C a ) was estimated from the difference between the total absorption at 676 nm and 650 nm measured with the AC-s instrument; this difference was then divided by a chlorophyll specific absorption coefficient assumed to be 0.014 m 2 •mg -1 at 676 nm by other studies on cultured phytoplankton and field measurements (Boss et al., 2004; Sullivan et al., 2005; Dierssen et al., 2009): C a(acs) (mg•m -3 ) = (a (676) – a (650) )/0.014 (15) 25 This chlorophyll concentration calculated from absorption correlated well with the measured chlorophyll fluorescence (R-squared = 0.73, N = 73,456, RMSE = 0.0039). The variability in the correlation between these two parameters may result from photoadaptation, chlorophyll fluorescence quenching, and diurnal fluctuations, or from the presence of other pigmented and fluorescence particles. Another possible factor is that the excitation wavelength for the chlorophyll fluorometer is offset from the chlorophyll-a absorption peak into a region where absorption by accessory pigments can occur (Perry et al., 2008). Blue to Red as indicator for Detritus: Direct measurements of detrital absorption were not made during the present study. Instead, the ratio of absorption at 440 to the absorption at 676 was used (Boss et al., 2001), where values around 2 indicate phytoplankton absorption, while values around 10 indicate the presence of detrital materials. The absorption at 650 was substracted from these wavelengths in order to minimize the effect of phytoplanktonic pigments absorption when calculating the B:R values: B:R = (a (440) – a (650) ) / (a (676) – a (650) ) (16) Areas within the channel: To facilitate data analysis the transect was divided into 5 areas in order to account for spatial variations across the channel (Fig. 2.3, lower right panel): two shallower areas (<90 m) at either end of the transect, two areas covering the 26 slope on each side of the main channel, and the deepest area in the channel center. For each area, consecutive profiles (3 downcast and 3 upcast) were binned at 1 m intervals. 27 2.5 RESULTS Environmental Variables The sampling strategy provided quasi-synoptic sections of the measured physical, chemical and bio-optical variables in the water column, revealing complex spatial and temporal distribution of hydrographic and optical variables (Fig. 2.3). The middle ebb tide (ME) and the late flood (LF) showed a similar distribution for all the measured parameters in the water column; dissolved oxygen concentration (DO) was higher through the water column (> 196 µmol•kg -1 ), indicating the presence of rich oxygen waters from the Pacific Ocean (Fig. 2.3d). Conversely, chlorophyll fluorescence (chl_f) and attenuation (c 650 ) showed higher values towards the end of the ebb tide (LE) and beginning of flooding (EF and MF1) (Fig. 2.3e and f), also with a similar spatial distribution except for the deeper regions of areas 3 and 4 (> 80 m) during EF and MF1. The CDOM fluorometer had problems during the first tide surveyed (ME) in areas 2 to 5, as can be seen in Fig. 2.3g first panel; this entire CDOM_f lane was eliminated from further analysis consequently limiting the discrimination between dissolved and particulated absorption for this tidal period. During the other tide stages, CDOM signals were higher close to the bottom following the density (σ θ ) distribution in the water column, with the highest values recorded also during EF for regions deeper than 80 m (areas 3 and 4, Fig. 2.3g third panel), similar to the total attenuation values. 28 Total Absorption and Scattering Spectra Areas 1, 2 and 3 showed higher variability of total absorption spectra at all depths during the 6 sampling periods, with lower values during ME, MF2 and LF across the whole spectral range (Fig. 2.4a, b and c). The lowest amplitudes during the study period were observed towards the end of the flood tide (MF2 and LF) in the first area (A1); low values for chlorophyll fluorescence and bean attenuation were also measured for this area during the same tide periods, suggesting a clear water Figure 2.4. Total absorption a T (minus water absorption) as a function of wavelength for the different areas (A1-5) during the six tidal stages (ME, LE, EF, MF1, MF2, LF). a T (m -1 ) 29 mass coming from the open ocean (Fig.3e and f, last two panels). This pattern was consistent with observed low amplitudes for scattering spectra with the lowest values and lowest mean amplitude in area 1 at the end of the flood tide (MF2 and LF; Fig. 2.5a). Conversely, areas 4 and 5 showed less variability between tide stages for total absorption and scattering spectra shape and amplitude, except at the end of the flood tide (LF) where both absorption and scattering showed higher variability in amplitude and shape (Fig. 2.4d and e, and Fig. 2.5d and e). Intermediate to high chlorophyll Figure 2.5. Total scattering b T as a function of wavelength for the different areas (A1-5) during the six tidal stages (ME, LE, EF, MF1, MF2, LF). 30 fluorescence and beam attenuation values were also recorded for these areas during this tide stage (Fig. 2.3e and f, LF). During the tide transition (LE, EF and MF1) the average total absorption as well as its variability increased for all wavelengths. This change between LE and EF generated the higher total absorption values, especially for the short wavelengths in areas 1, 2 and 3 (Fig. 2.4a, b and c). The highest values of total absorption correspond to the deepest regions in these areas where higher values for beam attenuation, CDOM fluorescence and scattering were also measured (Fig. 2.3f and g; Fig. 2.5a, b and c; LE and EF tides). Particulate Spectral Groups Two independent cluster analyses, one for particulate absorption (a p ) and another for particulate scattering (b p ), were performed in order to group similar spectral shapes and amplitudes. Three main groups were separated for a p spectra, each one with two defined subgroups (Fig. 2.6a); for b p , the cluster analysis separated 4 main groups (Fig. 2.7a). In general, Group 1 for a p showed higher absorption values for all the wavelengths, with higher variation around the 400 and 673 nm peaks (Fig. 2.6b, first panel). Group 2 showed intermediate absorption through their entire spectrum and less variation around the main peaks (Fig. 2.6b, second panel). Group 3 comprised observations with lower amplitudes for all wavelengths with values close to zero from ~600 to 650 nm, and a maximum absorption peak at ~400 and 673 nm (last panel Fig. 2.6b). 31 Figure 2.6. Cluster analyses results for particulate absorption (a p ). (A) dendrogram showing the three groups and their subgroups; Ccc = cophenetic correlation coefficient. (B) particulate absorption spectra for each group separated by the cluster analysis. Grey lines represent individual spectra curves, and the black line represents the mean spectra values. (C) First derivative for the particulate absorption spectra as a function of wavelength. (D) Second derivative for the particulate absorption spectra as a function of wavelength. a p (m -1 ) Ccc = 0.66 32 Figure 2.7. Cluster analyses results for particulate scattering (b p ). (A) dendrogram showing the four groups; Ccc = cophenetic correlation coefficient. (B) particulate absorption spectra for each group separated by the cluster analysis. Grey lines represent individual spectra curves, and the black line represents the mean spectra values. (C) First derivative for the particulate scattering spectra as a function of wavelength. (D) Second derivative for the particulate scattering spectra as a function of wavelength. Ccc = 0.70 33 The first derivative for the particulate absorption spectra (which represents the slope of the spectrum), showed a maximum value at 665 nm for all groups (Fig. 2.6c), indicating a distinctive change of slope or peak of absorption at this wavelength, corresponding to a chlorophyll absorption peak. Group 1 showed the highest variation, with strong negative and positive peaks, while Group 3 showed the lowest variation with a flat curve compared with the other groups. The region of the spectra with the lowest values for the second derivative was from 669 to 681 nm (671 nm always the lowest) in Groups 1 and 2, followed by the range from 415 to 427 nm (Fig. 2.6d). Group 3 showed an opposite pattern, with the lowest values for the second derivative in the range between 413 to 427 nm, followed by the range 669-681 nm. The 4 groups identified by the cluster analysis for particulate scattering differed mainly in their amplitude (Fig. 2.7b). Groups 1 and 2 presented the lower scattering values and no major peaks. Group 3 showed higher scattering values especially between 500 and 600 nm, and lower values around 669 nm. Group 4 included spectra with the highest amplitudes and more variation in shape, with distinctive higher scattering values around 500 to 600 nm and a low peak also at 669 nm. The first derivative for b p showed 2 distinct positive peaks in all groups, one around 487 nm and the second at 683 nm. The variation for these peaks and in general for the whole mean spectrum was higher for Group 4 (Fig. 2.7c). The lowest value for the second derivative was always present at 411 nm, followed by 421 nm in the lower portion of the spectra, and 647 nm in the higher side of the spectra (Fig. 2.7d). For groups 3 and 4, an additional lower value was also found in the area between 693 and 697 nm. 34 Calculated variables Calculated chlorophyll concentration from the AC-s measurements (C a(acs) ) correlated well with the measured chlorophyll fluorescence throughout the water column for all tidal stages (Fig. 2.3e and 2.8a). During the late ebb tide (LE) high C a(acs) was observed on both sides of the channel (areas 1 and 5) and in the middle of the channel in area 2. With the beginning of the flood (EF), the chlorophyll concentration in the middle of the channel decreased, but remains high in both sides (areas 1 and 5). During the middle stages of the flood tide (MF1 and MF2) chlorophyll concentration decreases with time but remains present subsurface in the middle of the channel. At the end of the flood (LF), high chlorophyll signal was observed in the eastern half of the channel (areas 3 to 5) with highest concentrations in the upper 40 meters toward the eastern boundary of the strait. The calculated blue to green ratio for absorption (B:R as a proxy for detritus materials) presented an opposite distribution to the calculated chlorophyll concentration for all depths during all tide stages covered (Fig. 2.8b). In general detritus contributes strongly to the particulate absorption in deeper zones of the middle of the channel (below 80 m). This deep central region also showed significant variation in particle size between tidal stages (indicated by the c p slope γ), from big particles (small γ) during the changing tide period (LE to EF) to smaller particles (large γ) at the end of the flood tide (LF; Fig. 2.8c). Small particles (large γ) were also abundant from 0 to ~50 m during MF1 and MF2 in areas 1 and 2. In general, high beam attenuation correlates better with bigger particles 35 (lower γ) during all tidal periods regardless of depth or area in the channel (Fig. 2.3f and 2.8c). Figure 2.8. Water column distribution of the derived products during the different tidal stages. C a(acs) : chlorophyll concentration from the AC-s. B:R: blue to red ratio as a proxy for detritus. γ: total attenuation spectral slope. a p _goups and b p _groups: absorption and scattering groups respectively separated by the cluster analyses. C a(acs) (mg•m -3 ) 36 Group 1 of a p from the cluster analysis showed a good spatial correlation with the distribution of higher beam attenuation (c 650 ), bigger particles, intermediate to high values of chlorophyll concentration C a(acs) , and low B:R ratio values (Table 2.1); these relationships were observed for all tide stages at different depths, especially during LE and EF in areas 1 to 3, and surface water in area 5 during LF (Fig. 2.8d). Group 2 was better correlated with intermediate values of C a(acs) , c 650 and intermediate to small particle size (intermediate to large γ) (Table 2.1), especially during the mid and late flood periods. Group 3 was corresponded with low values for C a(acs) , c 650 , B:R ratio, and small particle size (high γ) through the water column during MF2 and LF. For b p , the distribution of Group 1 corresponds to mid-range chlorophyll and beam attenuation values (Table 2.1), as can be seen for surface waters during MF1 and half way the water column during the MF2 and LF (Fig. 2.8e). This group is also spatially correlated with smaller particles during the end of the ebb tide (LE) and the middle of the flood (MF2). Group 2 has a similar distribution with high B:R ratio (~higher detritus) and smaller particles (large γ, especially during MF1 and 2, and LF), and is present where chlorophyll concentration and c 650 reach their lower values (e.g, surface and deeper waters during MF2 & LF) (Fig. 2.8e). Group 3 coincides with regions of higher chlorophyll concentration, especially during EF and MF1. Beam attenuation is also high following Group 3, but seems to be mainly due to phytoplanktonic cells since it correlates with C a(acs) . Finally, Group 4 is best correlated with bigger particles (smaller γ, e.g. deep in area 3 during the EF). This last group is present in areas where c 650 and C a(acs) values are higher, but the variation in both datasets is high too (Table 2.1), 37 indicating possible contribution of other materials in addition to live phytoplankton to the total beam attenuation (e.g. deep in area 1 during the LE, and surface water in area 5 during the LF). Table 2.1. Mean values (and standard deviation SD) for chlorophyll concentration (C a(acs) ), beam attenuation (c 650 ), gamma (γ) and the ratio of B:R as indicative of detritus prescence, for each one of the groups of a p and b p . N indicates the number of values included in the calculation. ! Mean SD Mean SD Mean SD Mean SD a p 1a 0.51 0.10 0.19 0.02 0.71 0.06 6.47 1.81 109 1b 0.67 0.16 0.22 0.02 0.67 0.07 5.42 1.29 22 2a 0.33 0.07 0.14 0.02 0.87 0.07 7.07 1.87 90 2b 0.45 0.06 0.16 0.01 0.81 0.06 6.05 1.45 154 3a 0.28 0.05 0.12 0.02 0.93 0.10 7.07 1.99 59 3b 0.20 0.04 0.12 0.01 0.93 0.08 7.41 0.78 40 b p 1 0.39 0.07 0.14 0.01 0.84 0.07 6.26 1.23 197 2 0.23 0.05 0.12 0.01 0.94 0.08 8.06 1.79 89 3 0.49 0.09 0.18 0.01 0.75 0.06 6.05 1.70 136 4 0.58 0.15 0.21 0.02 0.67 0.06 6.38 1.94 52 C a(acs) (mg •m -3 ) c 650 (m -1 ) B:R N 38 Overall Spectral and biogeochemical Spatial Patterns The spectral characteristics and the biogeochemical variables showed higher variability in both sides of the channel during the studied tidal cycle (areas 1 & 5). At surface (0-5 m deep), the highest particulate spectral changes were observed in area 5, with a representative of each spectral group for each tidal period studied; in addition, mean chlorophyll concentration increased by a factor of three and mean beam attenuation almost doubled from the mid flood periods to the late flood (Table 2.2), increasing also the amplitude of the absorption for all wavelengths of the spectra at LF (Fig. 2.9a first panel). Normalized values of particulate absorption (a p(norm) = a p(λ) /a p(675) ) for this region showed a small difference between the LE and MF2 tide stages; for the later, normalized absorption was in general higher towards the shorter wavelengths and between ~550 - 650 nm (Fig. 9b, first panel). Table 2.2. Range of values for chlorophyll concentration (C a(acs) ) and beam attenuation (c 650 ) for the different sub-areas, by spectral group during all the tidal stages. Sub- Area a p b p Mean SD Min Max Mean SD Min Max 3 surf1 0-5 3a, 3b 1, 2 0.25 0.04 0.21 0.30 0.12 0.01 0.11 0.15 5 surf2 0-5 1a, 1b, 2a, 1, 2, 3, 4 0.44 0.22 0.22 0.75 0.16 0.04 0.12 0.22 3a, 3b 4 midw1 30 1a, 2a, 2b 1, 3 0.45 0.06 0.40 0.54 0.16 0.02 0.14 0.19 1 midw2 20-25 1b, 2a, 2b, 1, 2, 3, 4 0.43 0.28 0.17 0.84 0.15 0.04 0.11 0.22 3b 3 deep 110-115 1a, 2a, 3a 1, 2, 3, 4 0.28 0.08 0.22 0.37 0.18 0.06 0.11 0.23 Spectral Groups Depth (m) Area C a(acs) (mg •m -3 ) c 650 (m -1 ) 39 At mid-water depth, area 1 showed the largest changes in chlorophyll concentration between tides as well as in the other environmental variables studied (~20-25 m deep); marked spectral differences in amplitude for all the tides periods were present, except for the MF2 and LF, which showed similar patterns (Fig. 2.9a second panel). This variation corresponds well with higher changes in C a(acs) (almost 5 times from the LF to the LE) and c 650 values (Table 2.2). The normalized spectra showed for the late flood higher values from 400 to ~ 640 nm (Fig. 2.9b second panel). Specific peaks were more pronounced also during this tide, including peaks at 413, 425, 471 and 529 nm. The late ebb showed a rather smooth spectrum, with higher values towards the short wavelengths and relatively flat shape from ~560 to 640. The center of the channel presented the lowest spectral and biogeochemical variability for the entire water column, except in deeper areas close to the bottom. Low values across the whole spectrum during all the tidal stages were observed in area 3 at 0-5 m deep (Fig. 2.9a third panel). Only spectral groups 1a, 2a and 2b for absorption, and groups 1 and 3 for scattering (from the cluster analysis, Fig. 2.6 and 2.7) were present during all tides (Table 2.2) while chlorophyll concentration and total attenuation were low, below 0.30 mg•m -3 and 0.15 m -1 respectively. This region showed no major changes in the normalized spectral shapes between the end of the ebb and flood tides (Fig. 2.9b, third panel); a peak is present ~443 nm for the mean LE spectrum, as well as higher absorption values ~413 nm. At mid water depth (~30 m deep) in area 4 minimum changes in chlorophyll concentration or other environmental variables between tides were observed (Table 2.2), 40 as well as more similar spectral shapes and amplitudes for all tides (Fig. 2.9a fourth panel). There were no major differences in the normalized spectral shapes between the end of the ebb and flood tides (Fig. 2.9 fourth panel). Figure 2.9. (A) Mean absorption spectra by tide at different depths on the water column for the border and center of the San Bernardino channel. (B) the same mean absorption spectra normalized to 676 nm. 41 The deepest portion in the channel (area 3) showed lower absorption values towards the end of the flood tide (MF2 and LF; Fig. 2.9a last panel), and higher values for the period of changing tides, especially for the EF tide in the deeper area. Chlorophyll concentration and beam attenuation showed similar patterns, from lower values at the end of the flood tide to more than double towards the changing tide (LE to ~MF1; Table 2.2). Higher normalized absorption values were observed during the early and late flood (EF & LF) all along the spectrum, especially between 400 and ~460 nm (Fig. 2.9b, last panel). 42 2.6 DISCUSSION In this study we gathered a comprehensive dataset of IOP’s and analyzed their relationship with physical-chemical and biological dynamics of a very complex coastal area to characterize in situ particle spectra. We demonstrated the utility of towed vehicles to provide quasi-synoptic images of the cross section in a dynamic system where measurements were carried out without major interruptions for ~ 1hr each pass. This sampling approach enabled the delineation of tidal stages not only based on currents, but also on hydrographic and optical characterization of the water masses coming from within the strait and from oceanic sources. During the ebb tide, the outflowing water was characterized by higher chlorophyll concentrations originating in the more productive areas inside the strait (Fig. 2.8a). We observed that the oceanic waters entering into the strait during the flood phase of the tides were characterized by higher temperatures, lower salinities, lower densities, higher oxygen concentration, lower chlorophyll concentration, lower total water attenuation and smaller sized suspended particles. This observed inflow of oceanic water reaching interior areas of San Bernardino in combination with the topographic structure of the strait has been observed to generate complex vertical mixing in the water column including aspiration of deep water containing elevated levels of non- phytoplankton suspended particles from the inland side of a sill (Jones et al., 2011). Based on the results from the cluster analysis, we were able to identify the spectral characteristics for different water masses in a very complex coastal area, and relate them with environmental variables. We found that Group 1 for particulate absorption showed a 43 particular spatial distribution well related with characteristic water from inside, with higher C a(acs) and bigger particles; at the same time Group 3 was more related with oceanic water. This distribution agrees with the results presented in Fig. 4 by Jones et al. (2011) for the same strait, as shown in a T/S diagram for the present study (Fig. 2.10); Group 1 corresponds to the area called by Jones et al. as “Interior Surface Water” (temperatures around 26ºC, salinities ~24 and depth ~<30 m); another end point corresponds to the area called “Exterior Surface Water”, where our Group 3b overlays well and temperatures are higher, salinity lower and depth from 0 to ~20 m; Group 3a well corresponds to the “Exterior intermediate” and “Exterior deep water” endpoints, with depths from ~60 to 120 m. Group 2 is a mix of water from inside and outside at different depths. Exterior or oceanic waters are characterized by lower chlorophyll concentrations and distinct phytoplankton groups when compared to coastal areas. For the Philippine Sea (next to San Bernardino strait), the main oceanic groups are Prochlorococcus and Synechococcus (O. Cabrera, pers.comm.; Zhao et al., 2010). During the present study the characteristic absorption peaks for these phytoplanktonic groups (Divinyl chl-a and chl- b) were not observed. Since the concentration of these pigments from open waters is expected to be low, it is plausible that their spectral properties were overlayed by stronger absorption peaks from other pigments present 44 Figure 2.10. Temperature-salinity plot for the San Bernardino survey color-coded by the particulate absorption spectral groups. inside the strait. For the interior waters of San Bernardino, diatoms and chlorophytes are the most abundant groups accounting for more than 50% of the total population (O. Cabrera, pers.comm.). Vertical mixing due to topographic features brings nutrients to interior surface waters of San Bernardino (Jones et al., 2011); these deep nutrients can help sustaining high chlorophyll concentrations inside the strait (>2 mg•m -3 ), which are advected to exterior waters during flood tides (Jones et al., 2011). After chl-a, the main pigment in diatoms is Fucoxanthin, which has a known peak of absorption around the ~ 446 / 468 nm. The first derivative for the particulate a p sub-groups 45 absorption spectra of Groups 1 and 2 showed a marked change in slope at this wavelength, indicating the presence of an absorbing peak around 446 (Fig. 2.6c), in addition to the standard chl-a peaks; these peaks corroborate the association of the spectral groups 1 and 2 with the “Interior surface water” endpoint classification. The analysis of the 2 nd derivative results within the spectral groups showed to be useful for partitioning the particulate absorption between phytoplankton and detritus. Groups 1 and 2 showed lower 2 nd derivative values at ~ 671, indicating more influence of phytoplanktonic pigments in that range. For Group 3 the more negative values were located at ~413-429 followed by ~669-681 nm, which can indicate a stronger influence of detritus absorption at short wavelengths. In general for deeper areas in the middle of the channel, and surface water in area 1, the blue to red absorption ratio (435:675 nm) was high (2.82-6.50), indicating the presence of detrital particles highly absorbing on the blue region. Deep in the channel in area 3 C a(acs) was low, while c 650 was high, suggesting a strong signal from non-algal particles consistent with resuspension from the bottom of the channel. In the case of surface water, high detritus occur during the late flood, where the intrusion of oceanic water is evident, carrying lower chlorophyll (~lower 675 nm peak), and a different combination of pigments from oceanic phytoplanktonic groups with stronger absorptions towards shorter wavelengths. Ocean color products derived from satellite imagery can provide additional insights of water masses distribution around archipelago systems. Inward highly productive waters can be differentiated from less productive oceanic waters as observed in a multi- 46 day composite for the San Bernardino area (Fig. 2.1A). Group 1 for particulate absorption from the cluster analysis corresponds to the interior surface waters observed by Jones et al. (2011) and to highly productive areas (C a ~ ≥ 0.7 mg•m -3 ) from the satellite image (Fig. 2.1A). Conversely, Group 3b for the particulate absorption corresponded to oceanic waters that can also be observed outside San Bernardino with lower chlorophyll concentrations (C a ~ 0.05 mg•m -3 ; Fig. 2.1A). These good correlations between the general distribution of in situ spectral data with coarse satellite imagery shows the utility of remote sensors for monitoring coastal water constituents. Additional validation studies of satellite imagery are necessary to test existing ocean color algorithms in coastal areas, and to asset the benefits of increased spatial temporal in situ and satellite observations to better estimate coastal productivity. Statistical analysis of particulate absorption spectra provided a characterization of water mass sources and processes that is consistent with the physical characteristics and tidal transport that were observed within the San Bernardino Strait. Spectral absorption and scattering not only differ between tides, but also between shallow and deeper areas across the channel. 47 CHAPTER 3 SPATIAL GRADIENTS OF WATER COLUMN OPTICAL PROPERTIES ACROSS A CORAL REEF LAGOON IN BERMUDA ISLANDS 3.1 ABSTRACT Coral reefs are one of the most diverse coastal ecosystems and yet, water masses sources and spatial gradients of biogeochemical signatures are not well known. We measured hyperspectral water optical properties and concentration of suspended materials at different locations across a reef lagoon in Bermuda Islands to characterize the distribution of the main water components. A well defined gradient from inshore to the open ocean was observed, with high chlorophyll concentration, high dissolved absorption and high concentration of small particles close to land. Hyperspectral absorption and scattering amplitudes and shapes also discriminated between inside and outside areas. Cluster analyses separated these absorption and scattering spectra in 3 groups each, also distributed spatially from inland to out shore. Stations close to shore showed higher values for dissolved absorption at 400 nm, higher beam attenuation at 650 nm, and chlorophyll concentration (> 0.08 m -1 , 0.5 m -1 and 0.3 mg•m -3 respectively). The effects of tides (water level) on the distribution of water constituents was observed with a significant inverse correlation with dissolved absorption at 400 nm and chlorophyll concentration (R 2 > 0.51, p<0.0001), and a positive correlation with volume 48 concentration and origin of suspended particles (R 2 > 0.30, p≤0.0001). The use of hyperspectral optical measurements in large shallow coastal systems was showed to distinct water sources with different biogeochemical signatures, relevant for the application of remote sensing algorithms and implementation of monitoring programs in shallow coastal waters. 49 3.2 INTRODUCTION Studies of the variability of inherent optical properties and their relation to particulate and dissolved materials in the water column of coral reefs are few (Boss and Zaneveld, 2003; Otis et al., 2004; Blondeau-Patissier et al., 2009) despite their potential impact on the benthic and phytoplanktonic communities inhabiting this ecosystem (McCulloch et al., 2003). Although nutrients are near the limit of detection in coral reefs (Capone, 1996), tides and circulation dynamics can resuspend sediments and dissolved compounds changing the chemical composition of the water column when organic and inorganic materials trapped in the sediments are released (Larcombe et al., 1995; Boss et al., 2001; Boss and Zaneveld, 2003; Ogston et al, 2004; Otis et al., 2004; Piniak and Storlazzi, 2008; Ouillon et al., 2010). In addition, mixing between reef waters and the open ocean forms complex dynamic systems influencing local geochemical conditions (Hoitink, 2004; Fishez et al., 2010). More suspended organic and inorganic materials can be present in the water column of coral reefs than the adjacent open ocean (Coble et al., 2004; Jouon et al., 2008) producing distinct optical properties. In coastal areas, water optical properties have been used as tracers for physical, chemical and biogeochemical processes (Babin et al., 2003; Boss et al., 2001; Cunningham et al., 2003; Oliver et al., 2004; Twardowski et al., 2001). Inherent optical properties (IOPs) are determined by the nature of particulated and dissolved matter in the water column (Mobley, 1994; IOCCG, 2006) and are used to estimate turbidity, pigment concentration, suspended sediments, particulate size distribution, and colored dissolved 50 organic matter (CDOM) in the water column (Schofield et al., 2004; Zaneveld et al., 2005). IOPs have been also used to characterize biogeochemical processes associated with benthic environments (corals, seagrasses, sand) over large spatial and temporal scales (Boss et al., 2001; Boss and Zaneveld, 2003). Studying the spatial variability of optical properties and the physical factors driving their distribution is essential for understanding the effects of natural and anthropogenic disturbances including hurricanes, and the deepening of navigation channels on coastal systems (Rogers, 1990; Edinger et al., 1998; Barnes and Lough, 1999; Szmant, 2002; Fortes, 2004). Studies in Bermuda and other coastal areas have indicated the importance of coral reefs for coastal protection, fisheries production, etc. (Costanza et al., 1997). However, characterization of the water components, water masses sources and spatial gradients of IOPs in shallow areas surrounding the northern coast of the Bermuda Islands and its implications to biogeochemical processes are not well known. The objectives of this paper are to: (A) characterize the hyperspectral properties of absorption and scattering to identify absorption peaks associated with water column components; (B) estimate the spatial distribution of dissolved and particulate materials, and particle size distribution in relation to distance to land; and (C) identify the temporal variations in the optical properties in relation to tidal stage. 51 3.3 STUDY AREA The Bermuda Islands (32˚ 18’ N, 64˚ 47’ W) are located in the Western North Atlantic where the Gulf Stream warms the ocean (Smith, 1998), promoting the development of a variety of subtropical marine flora and fauna, including seagrass beds, mangroves and the northernmost coral reefs in the world (Fig. 3.1). Deep reefs followed by a well covered rim reef surround the island and the North Lagoon located north-west to the islands. High diversity of corals distributed in patch reefs and reefs flats, and seagrass beds extend from nearshore across the extensive North Lagoon to the outer reefs (Smith, 1998; Murdoch et al., 2007). An interior bay (Great Sound) surrounded by different heavily populated islands is located south of the lagoon, and experiences navigation traffic. Two navigation channels crossing the North lagoon allow large commercial and cruise ships access from outside the reefs into Great Sound (Fig. 3.1). Tides in Bermuda are semidiurnal with a period of ~12.42 hrs (Principal Lunar M2 tide) and a mean annual range of 0.75 m (Morris et al., 1977), generating most of the exchange with offshore waters of the Atlantic Ocean. Due to low freshwater inflow, salinity values remains close to open ocean water of 36.5. Residence time of water inside the lagoon and bays is being estimated as ~4 d (Morris et al, 1977). Direction and speed of currents measured in the present study at different locations inside the lagoon and bay showed the open ocean inflow predominately from the north crossing over the outer reefs (Table 3.1). The recorded average speed was 7 cm•s -1 , reaching speeds of up to 15 cm•s -1 . 52 ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 1 2 3 4 5 6 7 8 9 10 11 Bermuda Islands Navigation Channel Outer Reefs Atlantic Ocean Atlantic Ocean Ireland Island Great Sound North Lagoon 64°40'0"W 64°40'0"W 64°45'0"W 64°45'0"W 64°50'0"W 64°50'0"W 64°55'0"W 64°55'0"W 32°25'0"N 32°25'0"N 32°20'0"N 32°20'0"N 32°15'0"N 32°15'0"N 0 3 6 1.5 Km Figure 3.1. The Bermuda Islands showing sampling stations (black dots, 1 to 11). Double dashed lines indicate the main navigation channels. 53 3.4 METHODOLOGY Field work Spatial and temporal variability of water optical properties were studied at eleven sampling stations located from inside Great Sound close to the islands, to the outer reefs. These stations covered areas with different substrates and depth previously identified by SCUBA dives (Fig. 3.1, Table 3.1) with an average depth of 4.5 m. The first three stations (1 to 3) presented a mix of sandy bottom and seagrass beds, with sporadic coral heads forming small patch reefs. Stations 4 to 9 were located over patch reefs in pairs of stations per location, one station over sandy bottom and the other over a coralline covered area. Stations 10 and 11 were located in the outer reefs over areas with a sandy and a coralline bottom (Fig. 3.1). Station 2 was further divided in 3 subareas perpendicular to the main navigation channel and revisited on four different days covering different tidal phases in order to correlate the observed variables with water level (Table 3.1). Optical profiles were obtained at each station measuring: a) filtered and unfiltered hyperspectral attenuation (c (λ) ) and absorption (a (λ) ) with an AC-s underwater spectrophotometer (81 wavelengths from 400 to 750 nm, spectral resolution 4 nm; WET Labs Inc.), b) particle size distribution and volume concentration with a LISST-100X (Type B, 32 size classes from 1.5 to 250 µm; Sequoia Scientific Inc), and c) optical backscatter (b b(λ) ) at 6 wavelengths with a HydroScat-6 (Hobi Labs.). All measurements were obtained within 15 minutes of each other at each station. AC-s filtered absorption spectra were obtained immediately after the unfiltered spectra profile by attaching a 0.2 54 µm filter cartridge to the seawater intake of the instrument. In addition, direction and velocity of the water column currents were measured with a bottom mounted, upward looking ADCP (Argonaut XR 1.5 MHz, SonTek/YSI Inc.). Data Processing and Analysis Water calibrations of the AC-s instrument were conducted with Milli-Q treated water at the beginning, middle and the end of the sampling effort. Temperature and salinity corrections were applied to the spectral measurements according to Sullivan et al. (2006) and the coefficients derived in there for AC-s corrections. The proportional method (Zaneveld et al., 1994) was used to correct absorption for optical scattering using 700 nm as the reference wavelength for absorption and scattering. The effect of seawater absorption (a w(λ) ) was removed during these procedures, and the particulate and dissolve fractions were calculated using data from the filtered and unfiltered measurements as: c t(λ) = a g(λ) + a p(λ) + a w(λ) + b p(λ) (17) c p(λ) = a p(λ) + b p(λ) (18) where c t(λ) and c p(λ) are the total and particulate attenuation for wavelength λ respectively, a g(λ) is the dissolved absorption, a p(λ) is particulate absorption, and b p(λ) is the particulated scattering. 55 The slope for the spectral dissolved absorption was calculated from: a g(λ) = a g(ref) • e -S(λ-λref) (19) where a g(λ) is the dissolved absorption as a function of wavelength measured with the filtered AC-s; a g(ref) is the dissolve absorption at a reference wavelength, which was 400 nm for the present study; and S is the slope of the spectrum. Measured a g(400) is a proxy for colored dissolve organic matter (CDOM) concentration, while the slope S is an indicator of the origin and chemical properties of that organic matter (Coble et al., 2004; Twardowski et al., 2004). Cluster analyses (Euclidean distance) were performed in order to group spectral data with similar shapes and amplitudes were each wavelength was treated as an independent variable. Derivative analyses were used in order to enhance irregularities in the spectral shapes of absorption and scattering (Bidigare et al., 1989; Hunter et al., 2008; Kirkpatrick et al., 2000; Millie et al., 1997, 2002). The first derivative is as definition the slope between consecutive wavelengths with a predefined window span. Minimum values in the second derivative of the spectral curve indicate the presence of specific pigments or other materials in the sample. Chlorophyll concentration C a(acs) was calculated by differencing total absorption at 676 nm and 650 nm measured with the AC-s instrument, then dividing by the chlorophyll specific absorption coefficient assumed to be 0.014 m 2 •mg -1 (Boss et al., 2004; Sullivan et al., 2005; Dierssen et al., 2009). 56 C a(acs) (mg•m -3 ) = (a (676) – a (650) ) / 0.014 (20) Relative detrital absorption is inferred from the ratio of absorption at 440 nm (blue region of the spectrum) to the absorption at 676 nm (red region), hereafter referred to as B:R (Boss et al., 2001), where lower values indicate low detrital absorption, while higher values indicate the presence of more absorbing detrital materials. Calibration of the LISST was tested at different times during the sampling period using background measurements with Milli-Q treated water. Management and processing of the data was done following the recommendations of the manufacturer. Total volume concentration was calculated integrating the values measured over all the size classes; measurements of the particle size distribution (PSD) and concentration are reported in terms of particle diameter measured in microns (µm) and volume concentration (µL•L -1 ) respectively. Backscattering data from the HydroScat instrument (b b ) was obtained after processing the measured volume scattering function using the estimated a (λ) and b (λ) from the AC-s for backscattering corrections as suggested by the manufacturer. Water level information was retrieved from the NOAA Tides & Currents database (http://tidesandcurrents.noaa.gov) for the Ireland Island Tide Station (Fig. 3.1), and correlated with IOP variables and suspended materials for station 2. 57 Table 1: Geographic position, depth, substrate and distance to the navigation channel for the sampled stations. Depth corresponds to the estimated station depth for a Tide Level = 0m. Currents direction and speed correspond to the mean value at mid depth during the period each station was surveyed. Station Latitude Longitude Depth Tide Substrate Distance to Currents Currents # (Deg.) (Deg.) (m) Level (m) channel (m) Direction (Deg.) Speed (cm•sec -1 ) 1 64.837107 32.293286 5.5 (+) 0.75 Sand/Seagrass 337 312.8 6.8 2.1 64.827747 32.313471 4.1 (+) 0.78 Sand/Seagrass 101 NA NA 2.2 64.825566 32.312877 4.0 (+) 1.00 Sand/Seagrass 319 266.9 11.8 2.3 64.824615 32.311607 3.9 (+) 0.89 Sand/Seagrass 460 NA NA 3.1 64.830167 32.314661 1.1 (+) 0.63 Coral 158 NA NA 3.2 64.832046 32.316618 2.5 (+) 0.87 Coral 415 NA NA 4 64.803949 32.354926 1.2 (+) 0.58 Coral 1855 82.9 6.1 5 64.800082 32.365881 6.9 (+) 0.43 Sand 2232 106.7 4.2 6 64.788483 32.392731 2.9 (+) 0.56 Coral 2200 218.7 7.9 7 64.801156 32.398315 3.6 (+) 0.40 Sand 1695 162.2 4.9 8 64.828006 32.417432 7.2 (+) 0.23 Coral 1608 NA NA 9 64.820703 32.420654 4.2 (+) 0.60 Sand 1336 NA NA 10 64.822206 32.453518 9.5 (+) 0.48 Sand 4627 NA NA 11 64.814688 32.456310 4.2 (+) 0.77 Coral 4772 NA NA Table 3.1. Geographic position, depth, substrate and distance to the navigation channel for the sampled stations. Depth corresponds to the estimated station depth for a Tidal level = 0m. Currents directions and speed corresponds to the mean value at mid depth during the period each station was surveyed. 58 3.5 RESULTS Spectral variations of IOPs Hyperspectral attenuation and absorption measurements (Fig. 3.2A, B, C and E) and multispectral measurements of backscattering (Fig. 3.2F) were obtained in all stations. A defined spectral gradient was found for these optical parameters, from high mean spectral amplitudes for all wavelengths inside Great Sound (Stations 1, 2 and 3) to lower values towards the outer reefs (Stations 10 and 12). Spectral particulate absorption was normalized to the chlorophyll absorption peak at 676 nm (a p(676) ) for each station. When normalized, the absorption spectra sort into two major groups with distinctive differences in the short wavelengths. The first group included stations inside Great Sound (1, 2 and 3) showing high absorption in the blue region (~400 nm). The second group covered all the other stations from the lagoon and the outer reefs where absorption at 400 nm was lower (Fig. 3.2D). In addition, the spectral shape for the stations in the second group showed two absorption peaks around 446 and 500 nm indicating the presence of different photosynthetic and photoprotective pigments respectively. Particulate Spectral Groups Three groups were separated by cluster analysis for the a p spectra (Fig. 3.3). Group 1 included all spectra for stations 1 and 3, and spectra from station 2 during low tide; in general this group showed high absorption values and less variation 59 Figure 3.2. Total attenuation (A), dissolve absorption (B), particulate absorption (C), normalized absorption to 676 nm (D), particulate scattering (E) and particulate backscattering (F) as a function of wavelength for all sampling stations. 400 450 500 550 600 650 700 0.00 0.02 0.04 0.06 0.08 0.10 0.12 ag (m -1 ) B 400 450 500 550 600 650 700 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 Sta1 Sta2.2 Sta3 Sta4 Sta5 Sta6 Sta7 Sta8 Sta9 Sta10 Sta11 ap (m -1 ) C Wavelength (nm) 0.0 0.2 0.4 0.6 0.8 1.0 Ct (m -1 ) A 400 450 500 550 600 650 700 0.0 0.2 0.4 0.6 0.8 1.0 bp (m -1 ) E 400 450 500 550 600 650 700 0.6 1.0 1.4 1.8 2.2 2.6 ap(norm) (m -1 ) 400 450 500 550 600 650 700 D 400 500 600 700 800 900 0.000 0.004 0.008 0.012 0.016 0.020 bbp (m -1 ) F Wavelength (nm) 60 Figure 3.3. Particulate absorption (A to C), first derivative (D to F), and second derivative (G to I) for the spectral groups separated by the cluster analyses as a function of wavelength. Grey lines represent individual spectra curves, and the black line represents the mean spectra values. Dotted lines are ± two times de standard deviation. 61 through their entire spectrum (Fig. 3.3A). Group 2 (Fig.3B) grouped stations 4 and 5, and also included spectra from station 2 during high tide and spectra from station 8 between 4 and 7 m deep during ebb tide. This group presented higher variation around the short wavelengths (~400 to 500 nm), and low variation and amplitude around the second peak at ~676 nm compared to the first group (Fig. 3.3A). Group 3 included spectra with the lowest amplitudes for all wavelengths, and maximum absorption peaks at ~440 and 673 nm (Fig. 3.3C). This last group contained stations 6, 7, 9, 10 and 11, and the shallow (1-3 m) spectra for station 8. Maximum values at ~410, 429 and 667 nm were found with the first derivative for the particulate absorption spectra for all groups (Fig. 3.3D, E and F). Groups 1 and 2 showed higher variation in their first derivative, compared to Group 3, which showed lower variation and less number of peaks. Groups 1 and 2 showed strong peaks in their second derivative at different wavelengths from 400 to 500 nm and 676 to 684 nm (Fig 3.3G and H). Lower amplitude peaks were observed for the second derivative of Group 3 (Fig. 3.3I). Three groups were also separated by the cluster analyses for the b p spectra (Fig. 3.4). As the first group for a p , Group 1 for b p included station 3, the deepest spectra for station 1 (5 to 6 m deep) and spectra from station 2 during low tide; this group showed high scattering values and intermediate variation through their entire spectrum compared to the other groups (Fig. 3.4A). Group 2 (Fig. 3.4B) included shallow spectra of stations 1 (1 to 4 m deep), station 2 during high tide, and the deepest spectra of station 8 (7 m deep). This group showed intermediate scattering values and 62 Figure 3.4. Particulate scattering (A to C), first derivative (D to F), and second derivative (G to I) for the spectral groups separated by the cluster analyses as a function of wavelength. Grey lines represent individual spectra curves, and the black line represents the mean spectra values. Dotted lines are ± two times de standard deviation. 63 low variability. Group 3 grouped stations 4 to 11, and station 8 from 1 to 6 m deep (Fig. 3.4C). This group had the lowest amplitudes of b p for all wavelengths, and the highest variability. Values for the first and second derivatives varied between groups primarily in the 400 to 550 nm region (Fig 3.4D - I), where Group 3 showed lower values and variability. Spatial variations of IOPs The highest values and variation of dissolved absorption at 400 nm occurred at the stations inside Great Sound (Stations 1, 2 and 3) (Fig. 3.5A), with values of a g(400) > 0.08 m -1 . The stations located next to the Great Sound in the lagoon (Stations 4 and 5) and the exterior stations 11 showed intermediate a g(400) values (~0.05 – 0.06 m -1 ), while the lowest dissolved absorption values occurred in the middle of the lagoon (stations 6 to 9) and in station 10 (0.03 – 0.05 m -1 ) (Fig. 3.5A). The spectral slope of the dissolved fraction, S, showed small variation among stations with a range from 0.012 to 0.014 m -1 (Fig. 3.5B), except for the most interior station where S was >0.014 m -1 . Particulate attenuation at 650 nm, C p(650) , was more than 0.5 m -1 at stations 1, 2 and 3 (Fig. 3.5C), indicative of higher suspended particle concentrations; for the other stations, a gradient of decreasing attenuation was observed from ~0.36 m -1 inside of the lagoon to ~0.15 m-1 at the outside station (Fig. 3.5C). The slope for the spectral particulate attenuation (γ), indicative of the particle size distribution, was smaller 64 Figure 3.5. Mean values by station for A) dissolve absorption at 400 nm, B) the slope of the dissolved absorption spectra, C) particulate absorption, D) slope of the particulate absorption, E) chlorophyll concentration derived from absorption, F) blue to red ratio, and G) backscattering to scattering ratio. Vertical bars represent ± standard deviation. C a(acs) (mg . m -3 ) 0.1 0.2 0.3 0.4 0.5 0.6 ag(400) (m -1 ) 0.02 0.04 0.06 0.08 0.10 0.12 S 0.011 0.012 0.013 0.014 0.015 0.016 Cp (650) (m -1 ) 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Ȗ 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.2 B:R 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 0.01 0.02 0.03 0.04 0.05 0.06 b b(550) /b (550) 1 2.2 3 4 5 6 7 8 9 10 11 Station 1 2.2 3 4 5 6 7 8 9 10 11 Station A C E G B D F 65 (larger particle size) for all the inside stations, while the outside stations 10 and 11 showed considerably higher values (smaller particle size; Fig. 3.5D). The indicators of two major components of the particulate fraction, chlorophyll concentration (C a(acs) ) and detritus (B:R), were higher within Great Sound, decreasing towards the outside reef (Fig. 3.5E and F). Station 10 on the outer reef differed slightly from the surrounding stations having both higher C a(acs) and B:R values than the surrounding stations (Fig. 3.5E).The ratio b bp(550) /b p(550) is indicative of the relative proportion of organic and inorganic material in the suspended particles, higher ratios indicative of more inorganic particles, and lower ratios indicative of more organic material with less hard surfaces. The values of b bp(550) /b p(550) were highest at stations 7, 8 and 9 (Fig. 3.5G). This ratio also revealed the presence of more organic particles with higher scattering in relation to backscattering for station 10, which is typical of phytoplankton cells. This finding corresponds with higher C a(acs) in station 10 compared to the other stations towards the outer reefs. Spatial variations of suspended materials Mean volume concentrations of suspended particles were highest at stations 1 to 4 (>2.0 µL•L -1 ), decreasing steadily from high values inside Great Sound to the lowest mean concentrations at stations 10 and 11 (< 0.5 µL•L -1 ) (Fig. 6A). Stations 3 and 4 were an exception in this gradient, having station 3 lower values comparable to stations from the lagoon (~1.4 µL•L -1 ), and station 4 the highest mean volume concentration of all 66 stations (3.38 µL•L -1 ). The higher concentration at this station results primarily from the concentrations of particles larger than 100 µm (Fig. 3.6B). Station 1 also showed higher concentration of larger particles, followed by stations 2 and 5. The lowest volume concentrations of suspended particles across all measured sizes occurred at stations 10 and 11 on the outer reef. All the other stations from the lagoon had similar concentrations of mid-size particles (~52 to 166 µm, Figure 3.6B). Temporal variations of IOPs and suspended materials The dissolve fraction a g(400) and chlorophyll concentration C a(acs) showed an inverse linear correlation with water level (R 2 > 0.51) (Fig. 3.7A and B). a g(400) values ranged from ~0.11-0.12 m -1 during low water level (~ebb tide) to ~0.08 m -1 during periods of high water level (~flood tide), while for C a(acs) values ranged from ~0.4 mg•m -3 during ebb tide to ~0.25 mg•m -3 during flood tides. The blue to red ratio showed a low coefficient of determination when correlated to water level and higherd dispersion of the data (Fig. 3.7D). In contrast, b b /b, detritus and suspended materials showed a lower positive correlation with water level (Fig. 3.7C and E). Volume concentration showed this positive trend, although high variation was observed during all the water levels considered. 67 Figure 3.6. Mean values by station for A) volume concentration of particles by station, and B) volume concentration as a function of Particle size distribution (PSD). Vertical bars represent ± standard deviation. 0.0 1.0 2.0 3.0 4.0 1 2.2 3 4 5 6 7 8 9 10 11 Stations Sta1 Sta2.2 Sta3 Sta4 Sta5 Sta6 Sta7 Sta8 Sta9 Sta10 Sta11 0.0 0.2 0.4 0.6 0.8 231 196 166 130 101 73 52 23 4 9 ROXPHFRQFHQWUDWLRQȝ/ . / -1 ) 36'ȝP 9 ROXPHFRQFHQWUDWLRQȝ/ . / -1 ) A B 68 Figure 3.7. Linear regressions between water level and A) dissolved absorption a t 400 nm, B) chlorophyll concentration, C) backscattering to scattering ratio, D) blue to red ratio, and D) volume concentration of suspended particles. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Water level (m) 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Water level (m) 1.0 2.0 3.0 4.0 5.0 6.0 0.020 0.025 0.030 0.035 0.06 0.08 0.10 0.12 0.14 0.2 0.3 0.4 0.5 2.10 2.15 2.20 2.25 2.30 2.35 ag(400) (m -1 ) b b550 /b 550 C a(acs) (mg . m -3 ) B:R 9 ROXPHFRQFHQWUDWLRQȝ/ . / -1 ) A C E B D r 2 =0.51 p<0.0001 N=89 r 2 =0.30 p=0.0001 N=44 r 2 =0.32 p<0.0001 N=2597 r 2 =0.53 p<0.0001 N=89 r 2 =0.19 p<0.0001 N=89 69 3.6 DISCUSSION The spatial temporal distribution of hyperspectral IOPs and the concentration of suspended particles were studied across different areas in the reef lagoon of Bermuda Islands. Spectral characteristics clearly showed the separation of different water masses with distinctive optical signatures, supporting the utility of hyperspectral measurements for future studies in coral reefs. A distinct gradient was observed for all parameters from nearshore (interior stations) at Great Sound across North Lagoon to the outside reefs. The interior stations showed higher absorption of dissolve materials, higher chlorophyll concentrations and higher detritus signals, as commonly observed in coastal areas (Coble et al., 2004; Zanardi_Lamardo et al., 2004; Para et al., 2010), but scarcely reported for Bermuda Islands (von Bodungen et al., 1982; Jensen et al., 1998). Derivative analysis showed the common appearance of peaks at ~430 and 662 nm (expected for chlorophyll-a absorption) for all the spectral groups. Differences in chlorophyll concentration and spectral absorption between stations across the reef platform can be attributed to variations in species composition with different absorption properties. The higher chlorophyll concentration inside Great Sound can be attributed to nutrients inputs from terrestrial sources and linked to higher detritus concentration (Babin et al., 2003; Coble et al., 2004; Blanco et al., 2008). Higher concentrations of nutrients have been reported for Great Sound due primarily to human development in the islands (von Bodungen et al., 1982). This nutrient availability may promote the establishment of 70 local phytoplankton communities resulting in higher water column primary productivity compared to the adjacent lagoon. Nutrients inputs can also benefit local seagrass beds in the area, as observed over the last few decades (see Murdoch et al., 2007). Seagrass beds can also be a source of nutrients in the form of detritus or dissolved materials to local phytoplankton communities. Higher B:R values inside Great Sound reported in the present study, in combination with high particulated absorption in the blue region (~400 nm) for the area indicate higher detritus concentrations in the interior stations, attributed to phytoplankton and seagrass natural processes. Clusters analyses separated 3 reef areas based on their spectral characteristics. In general, particulate absorption and scattering spectra separated the stations inside the Great Sound from the ones near the outer reefs. An intermediate group was identified which included stations from the lagoon but can also extend to the interior or exterior areas. Station 8 located close to a navigation channel was one of the deepest stations surveyed (7.2 m, Table 3.1) for the North Lagoon area and presented spectral characteristics of group 1 at depths > ~4 m, while its shallow water resembles group 3. This station was surveyed at the end of the ebb tide, and this vertical distribution of spectral characteristic indicate two water fronts: a deeper water mass coming from the interior and leaving to the outer reefs, and a shallower water mass reflecting the intrusion of oceanic water into the lagoon. The other stations in the lagoon are shallower, which limits the clear differentiation of water masses in their water column. IOPs and suspended materials were statistically correlated with water level (tidal stages) at Station 2, indicating an active exchange of water and materials between Great 71 Sound and the North Lagoon. During ebb tides Great Sound is a source of chlorophyll and dissolved materials to the lagoon, with higher concentrations of chlorophyll and a g(400) during low water levels, and lower C a and a g(400) values when water level is high. Lower concentrations in Great Sound at maximum flood tide are consistent with water coming into Great Sound from the lagoon and open ocean. Conversely, the incoming tide seems to resuspend or bring suspended inorganic matter with higher b bp(550) /b p(550) ratios (Hoitink, 2004; Jouon et al., 2008). Particulate attenuation at 650 nm measured with the AC-s (c 650 ) and the volume of suspended materials from the LISST showed a consistent distribution between stations. The difference between volume concentration and c p(650) at station 4 may be attributed to change in wind forcing during the period of sampling at the site. Between the time of the AC-s profiles and the LISST profiles, the wind speed increased from ~4 to ~7 m/s. This increased wind speed can resuspend sediments and organic detritus from the bottom (Larcombe et al., 1995). In addition to tides and wind, navigation activities can also cause recurrent resuspension of sediments in Bermuda. These resuspension events change drastically the optical properties of the water column and can be detected in situ or with remote sensing imagery (Fig. 3.8). The fact that no plumes were detected during the sampling dates around the navigation channels in Bermuda can indicate that the optical effects of these plumes occur at spatial-temporal resolutions different than the observations made in this study. 72 Figure 3.8. Quickbird satellite image showing a boat leaving a plume of resuspended sediments along a navigation channel in the Great Sound. The plume (indicated by the arrow) remains in the water column for several minutes, as suggested by this image. The use of satellite imagery can help with long term monitoring of suspended sediments in coral reefs, assuming the existing spatial and spectral resolution of remote sensors are suited for this purpose. Absorption and scattering spectral analysis, in combination with the concentration of suspended sediments provided a characterization of different water masses in 73 Bermuda. Tidal dynamics were identified as an important mechanism for the transport and exchange of materials between the areas inside the lagoon and the open ocean. 74 CHAPTER 4 IN SITU QUASI-SYNOPTIC OPTICAL MEASUREMENTS AS TOOLS FOR VALIDATING MODIS 250m OCEAN COLOR PRODUCTS IN A HIGHLY DYNAMIC COASTAL AREA 4.1 ABSTRACT Validation of satellite imagery with quasi-synoptic in situ measurements is a fundamental process for testing and tuning standard ocean color models in coastal areas. To test existing ocean color algorithms in a dynamic strait from the Philippines archipelago, we follow a 12-km cross-channel transect collecting quasi-synoptic in situ optical measurements using an undulating underwater vehicle. Satellite imagery was processed to produce standard ocean color products with a spatial resolution of 250 m for the MODIS Semi Analytic Algorithm (Carder), Quasi-Analytical Algorithm (QAA), GSM Semi-Analytical Bio-optical model (GSM), Gould model (Gould), and the OC3 for MODIS. Data from pixels matching the spatial locations of the in situ data were extracted for ground-truth comparisons. Tested satellite algorithms for absorption at 412 nm (a 412(sat) ) showed good correlations with in situ a 412(in_situ) and colored dissolved organic matter (CDOM) fluorescence, with R 2 > 0.7. Estimated suspended sediments (TSS) and chlorophyll concentration (C a(sat) ) from satellite data showed good correlations with in situ beam attenuation (c 650 ) and C a(in_situ) respectively (R 2 > 0.8). These correlations 75 persisted along the surveyed transect and from the surface to ~20 m depth. This good agreement between the satellite products at 250 m resolution and the in situ dissolve and particulated measurements in the water column shows the possibility of using satellite imagery with higher spatial resolution for future modeling and monitoring of optical properties, mixing of water masses, and algal blooms in coastal systems. 76 4.2 INTRODUCTION Remote sensed information from satellites has become fundamental for global and local studies of natural and anthropogenic processes in the ocean, yet satellite data still requires spatial temporal validation in order to generate more robust predicting data and biogeochemical models. The Moderate Resolution Imaging Spectroradiometer sensor (MODIS) on board the satellites Terra and Aqua (part of the NASA’s Earth Observing System - EOS) is one of the main satellite sensors for studying ocean global dynamics. MODIS has 36 spectral bands, ranging from the visible (7 bands) to the near infrared (9 bands) light and with bands having a spatial resolution of 250, 500 m and 1 km (Xion et al., 2009). These spectral characteristics of MODIS allow the estimation of suspended materials in the ocean (Salomonson et al., 1995; Esaias et al., 1998), although traditionally the highest spatial resolution used is 1 km. Recent studies have proposed the use of the 250 m resolution bands (Miller & McKee, 2004; Franz et al., 2006) and a combination of atmospheric corrections (Ladner et al., 2007) in order to produce standard ocean color products at a higher spatial resolution. This improved spatial resolution is ideal for coastal studies, where land generally masks close to shore waters, and there can be large horizontal gradients in water column constituents (Bissett et al., 2004; Coble et al., 2004; Ladner et al., 2007). Different optical models and calibration methods have been used to compare in situ properties with satellite data to develop satellite ocean color algorithms (Twardowski et 77 al., 2005; IOCCG, 2006). The resulting algorithms are continuously validated with in situ data collected at specific monitoring stations in the ocean. For MODIS, the validation/calibration station is the Marine Optical BuoY (MOBY), located approximately 18 km west of the Hawaiian island of Lanai (Clark et al., 2002). This single point in the ocean has served as the sole source of in situ data for the latest NASA’s validation of ocean color algorithms (Bailey et al., 2008). Regional efforts have been made in order to test the validity of the standard ocean color algorithms for specific coastal areas (e.g. Werdell et al., 2007, Dupouy et al., 2010), but at present there are no validation efforts carried out in the interior seas of island archipelagos. There are large differences in the water column optical properties between coastal areas and the open ocean (Morel & Prieur, 1977; Mobley et al., 2004) requiring regional tuning of ocean color algorithms. Water column optical properties observed by satellites are determined by the concentration and composition of the suspended materials (organic and inorganic particulate matter and dissolved organic compounds) and the optical properties of the water itself (Schofield et al., 2004; IOCCG, 2006,). The satellite derived constituents are used to characterize water clarity, quality, and biogeochemical components and processes (Coble et al., 2004; Schofield et al., 2004, Muller-Karger et al., 2005). The open ocean is characterized by clear waters having low concentrations of particulate and dissolved materials, where phytoplankton dominates the optical variability (Bricaud & Stramski, 1990; Mobley et al., 2004). The clearest oceanic water is found in the South Eastern Pacific where chlorophyll concentrations are as low as 0.02 mg•m -3 , and total absorption and backscattering coefficients are lower than the previously known 78 values for pure water, indicating extremely low concentrations of CDOM (color dissolve organic matter) and particulate materials (Morel et al., 2007; Twardowski et al., 2007). In contrast, coastal waters are often loaded with particulate and dissolved organic matter, inorganic suspended matter and high phytoplankton biomasssupported by large nutrient inputs from land or from coastal upwelling. Thus, chlorophyll concentrations of ~100 mg•m -3 around enclosed areas or productive bays are not uncommon (Wozniak & Dera, 2007). CDOM and suspended materials concentrations are usually high near rivers discharges (dissolve absorption (a g(412) ) > 0.4 m -1 and total attenuation (c t(650) ) > 1 m -1 ; Coble et al., 2004). In this study, we assess the feasibility of using quasi-synoptic in situ towed optical data to validate MODIS 250 m resolution satellite ocean color products in a dynamic archipelago strait in the Philippines. In situ sampling platforms equipped with optical instrumentation and water samplers are typically used for validation of remotely sensed ocean color. Although these platforms provide detailed vertical resolution of the water column, they are limited to a single point and/or a few casts per day (Twardowski et al., 2005). For coastal studies where a more comprehensive spatial and temporal coverage (horizontal and vertical) are required, towed undulating vehicles can provide detailed coverage (Twardowski et al., 2005). These vehicles have been used primarily to study physical and biological processes in coastal and open ocean areas (Petrenko et al., 1997; Barth et al., 2001; Barth et al., 2005; Ackleson, 2006; Peterson & Peterson, 2008; Horner-Devine, 2009; Jay et al., 2009). Because the data from these vehicles can be easily synchronized with overpassing satellites, we used vehicles data in this study to 79 compare with ocean color data processed with five algorithms: MODIS Semi Analytic Algorithm (Carder), Quasi-Analytical Algorithm (QAA), GSM Semi-Analytical Bio- optical model (GSM), Gould model (Gould), and the OC3 for MODIS (Garver and Siegel 1997; Carder et al., 1999 & 2004; Lee et al., 2002; Maritorena et al., 2002). The algorithm evaluation between in situ towed optical measurements and high spatial resolution images (250 m) from this dynamic Philippine strait provides valuable information for coastal remote sensing monitoring programs and coastal satellite missions. 80 4.3 STUDY AREA The Philippine Islands located in Southeastern Asia between 116° 40' and 126° 34' E, and 4° 40' and 21° 10' N comprise an archipelago of over 7000 thousand islands with numerous straits (Fig. 4.1A) having very complex, dynamic circulation patterns (Han et al., 2009). Flow through these straits can rapidly mix water masses from either side and cause significant vertical exchange of dissolved and particulate constituents as the flow interacts with underwater topographic features such as sills (Jones et al., 2011). The San Bernardino strait (12°36' N, 124°12' E) is one of the two connections between the Pacific Ocean and the archipelago interior (Fig. 4.1B). Surface water from the Pacific is transported over shallow sills (~100 m) through the strait by the strong tides in the area with currents of up to 2.5 m•s -1 , and by subtidal currents (Meñez et al., 2006; Jones et al., 2011). The resulting circulation patterns in San Bernardino generate horizontal and vertical mixing of water and high vorticity at spatial scales ≤ 1 km (Ohlmann, 2011; Jones et al., 2011). The high current velocities through this strait represents a challenge for retrieving valid information from standard ocean color products with a standard maximum resolution of 1 km, making the San Bernardino Strait an ideal test field for higher resolution satellite products. 81 Figure 4.1. A) Philippine Islands. B) Map showing the location of the transect (dashed line) used for the in situ deployment of the TRIAXUS in San Bernardino Strait. 125°0'E 120°0'E 115°0'E 20°0'N 15°0'N 10°0'N 5°0'N 124°30'0"E 124°0'0"E 13°0'0"N 12°30'0"N Pacific Ocean Pacific Ocean San Bernardino Strait 0 10 20 5 Km Philippines Islands A B 82 4.4 METHODOLOGY In situ data collection and analysis During the PhilEx Intensive Observational Period in 2009, water column measurements were obtained along a 12-km long transect in a sawtooth pattern using an undulating towed vehicle (TRIAXUS - MacArtney) (Figure 4.1B), as described in Chapter 2 of this dissertation. Briefly, the vehicle, towed at ~7 knots (13 km/h), carried hydrographic (SeaBird pressure, conductivity, temperature and dissolved oxygen probes), and optical sensors. A Wet Labs AC-s was used to measure optical total beam attenuation and total absorption at 81 wavelengths from 400 to 750 nm. A Wet Labs C-Star transmissometer measured beam attenuation at 650 nm, and WetLabs WETStar chlorophyll and Color-Dissolved Organic Matter (CDOM) fluorometers measured chlorophyll and CDOM fluorescence. Additional sampling was conducted during the field campaign covering different areas in the strait. In particular, the transect chosen from the validation experiment was sampled at a time coinciding with an overpass of the NASA satellite Aqua. A total of 47 full water column profiles were done with a horizontal resolution of approximately 264 m. For the purpose of this study, the in situ data set was vertically integrated in four depth ranges (0-15 m, 0-10 m, 0-15 m, and 0-20 m) for comparisons with satellite imagery. The distance (D (hor) ) from the TRIAXUS to the GPS system onboard the ship at any given time was calculated using the length of the sea-cable pulling the vehicle (cableout) and the depth of the vehicle (Z) as follow: 83 This distance, along with the geographic position of the ship and its true course were used to calculate a more accurate geographic position of the vehicle. Data Processing and Analysis AC-s data processing: All the measurements made during this study correspond to total absorption (a) and attenuation (c), where a is the sum of the absorption by the dissolved (a g ) and particulated (a p ) fractions, minus water absorption (a w ) removed during data processing (a = a g + a p - a w ). The AC-s instrument was calibrated with Milli- Q treated water at the beginning of the cruise and each time that the towed vehicle was pulled out of the water. Data from the AC-s instrument was collected at 4Hz and a median value was obtained for each one-second interval. A time offset correction of 1.7 seconds was applied to synchronize the AC-s and the CTD data for further AC-s processing. Temperature and salinity corrections were performed according to Sullivan et al. (2006) and the coefficients derived in there for AC-s corrections; the proportional method (Zaneveld et al., 1994) was used for scattering corrections of the absorption values with 700 nm as the reference wavelength for absorption and scattering. Total scattering (b (λ) ) was computed by subtracting a (λ) from the corresponding c (λ) . Beam C attenuation and fluorescence processing: Dark counts were measured for the transmissometer and fluorometers before the deployment of the vehicle and every ! !!!"! !! ! ! ! !! !!"#$%&'( ! ! 84 time the vehicle was retrieved from the water in order to determine deviations from the calibration. The data from these fluorometers is reported as raw volts. Chlorophyll estimation from absorption: Chlorophyll concentration (C a ) was estimated from the difference between the total absorption at 676 nm and 650 nm measured with the AC-s; this difference was then divided by a chlorophyll specific absorption coefficient assumed to be 0.014 m 2 •mg -1 at 676 nm by other studies on cultured phytoplankton and field measurements (Boss et al., 2004; Sullivan et al., 2005; Dierssen et al., 2009): C a(acs) (mg•m -3 ) = (a (676) – a (650) )/0.014 (21) This chlorophyll concentration calculated from absorption correlated well with the measured chlorophyll fluorescence (R-squared = 0.73, N = 73,456, RMSE = 0.0039). Satellite Images processing Daily MODIS-Aqua images for the month of February were obtained from the NASA GSFC-LAADS system (http://ladsweb.nascom.nasa.gov), including a match-up image coinciding in time with the in situ campaign in San Bernardino. The downloaded Level 1 images and their corresponding products for atmospheric correction were processed by the Naval Research Laboratory (NRL) with their Automated Processing System (APS) software following the methods described in Ladner et al. (2007). Briefly, atmospheric corrections were applied using the short wave infrared channels (SWIR) 85 with near infrared channels (NIR) iterations and Stumpf 412 iterations for absorbing aerosol corrections. Water pixels contaminated with clouds, sun glint or marked as land were flagged and excluded from analysis (Ladner et al., 2007). A complete set of IOP products from the satellite data were derived following this procedure, including products for the Carder, QAA, GSM, and Gould algorithms, and the NASA chlorophyll OC3 algorithm for ocean color. For comparing with in situ measurements from San Bernardino, the derived satellite products for the mentioned algorithms included total absorption at 412 nm (a 412 ) as a proxy for CDOM and detritus, total attenuation at ~650 nm (c 650 ) as indicator of suspended particulate materials and the OC3 chlorophyll product. In situ vs. Satellite data Satellite images processed with the APS software were imported into the ENVI (ITT Visual Information Solution) image processing software. In situ geographic tracks of the TRIAXUS were also imported into ENVI and overlaid as vectors. Satellite data were extracted and exported from each pixel coinciding with an in situ observation; in situ data were averaged for the coincident pixel and were also exported. Both data sets were then imported into Matlab (Mathworks) for analysis. The depth-averaged in situ data were extracted within each pixel and compared to the satellite data using regression analysis to evaluate the performance of the satellite retrievals for different water column ranges. 86 4.5 RESULTS In situ observations The in situ sampling strategy revealed complex horizontal and vertical distributions for the considered variables along the San Bernardino channel (Figure 4.2). Two primary features can be distinguished along the transect: on the west side (from 0 to ~6 km) all the measured parameters were low, while higher values were observed in the east segment (~6 to ~12 km), This gradient across the channel reveals the presence of more absorbing materials towards the east side, as indicated by a 412 (>~0.05 m -1 ) (Fig. 4.2A) and more dissolve organic materials as measured by the CDOM fluorometer (>~0.2 v) (Fig. 4.2B); this pattern is consistent for the whole water column. Higher particulate concentrations, indicated by c 650 , also occurred on the east side of the strait, especially in the upper 40 m in the water column where total beam attenuation measured with the AC-s and the transmissometer were greater than ~0.2 m -1 , in contrast to the western side where the highest values of c 650 occurred below ~20 m and were less than ~0.15 m -1 . Phytoplankton biomass was higher in the east side of the transect especially in the eastern 2 kilometers of the transect, as indicated by chlorophyll concentrations estimated from AC-s absorption and chlorophyll fluorescence (Fig. 4.2E & F, 87 Figure 4.2. Spatial distribution of in situ water column measurements for: A) absorption at 412 nm (a 412 ), B) CDOM fluorescence (CDOM), C) total attenuation at 650 nm from the AC-s (c 650 ), D) total attenuation from the transmissometer (c 650(Trans) ), E) chlorophyll concentration from absorption (C a_acs ), and F) chlorophyll fluorescence (chl_f). Transects are oriented from west (left side) to east (right side). !" 88 respectively). Between 8 and 10 km, the chlorophyll maximum region was subsurface between 20 and 80 meters, having concentrations of 0.4-0.6 mg•m -3 (Figure 4.2E). In situ vs. Satellite data Three images from MODIS-Aqua were found to be cloud free over San Bernardino during the days the area was surveyed with the undulating vehicle. The first image was captured on February 16 during an early flood tide, two days before the in situ campaign started in the area. The second cloud free image from February 18 th (during late flood) coincided in time with the in situ transect with a total of 30 pixels overlaying the TRIAXUS transect. The last image was from February 19 acquired during ~high tide. In situ absorption at 412 nm and CDOM fluorescence correlated positively with the remotely sensed values of a 412 estimated with the Carder, QAA and GSM algorithms (Table 4.1). The integrated in situ values from 0 to 5 and 0 to 10 m deep better correlated with all three algorithms with coefficients of determination (R 2 ) higher than 0.70 for all cases. In general, the best correlations were obtained with the GSM algorithm where R 2 > 0.84 for the two shallowest depth ranges (Table 4.1). For the particulate fraction, the estimated total suspended sediments calculated with the Gould algorithm (TSS_Gould) correlated well with the observed total attenuation at 650 nm, c 650 , from the AC-s (R 2 > 0.83) and the transmissometer (R 2 > 0.87) for the shallowest two depth ranges (Table 4.2). These correlations with the satellite algorithm are lower when the in situ data is integrated from the surface to 15 or 20 meters. The derived satellite measurements of attenuation at 650 nm using the QAA algorithm also 89 showed a good correlation with the in situ c 650 , although only 4 valid points were successfully retrieved from the satellite image for this algorithm. In situ C a(acs) and chl_f correlated strongly with phytoplankton absorption at 443 nm (a ph(443) ) from the satellite algorithms (Table 3). They correlated most strongly with the Gould model, but all R 2 values were greater than 0.8 for depth ranges Table 4.1. Correlations between in situ observations and satellite estimated values for absorption at 412 nm (a 412 ), and between in situ CDOM fluorescence and satellite a 412 . Numbers in parenthesis are the sample number N. shallower than 10 meters. The coefficients of determination between a ph(443) and the in situ C a(acs) were all higher than 0.71 for at all depth ranges covered. The NASA OC3 In situ Depth Parameter (m) Carder QAA GSM a 412 0-5 0.772 0.794 0.879 P < 0.001 (10) P < 0.001 (13) P < 0.001 (13) 0-10 0.785 0.804 0.888 P = 0.001 (10) P < 0.001 (13) P < 0.001 (13) 0-15 0.722 0.636 0.834 P < 0.001 (13) P < 0.001 (17) P < 0.001 (17) 0-20 0.746 0.648 0.841 P < 0.001 (16) P < 0.001 (19) P < 0.001 (19) CDOM_f 0-5 0.815 0.819 0.845 P < 0.001 (10) P < 0.001 (13) P < 0.001 (14) 0-10 0.829 0.849 0.864 P < 0.001 (11) P < 0.001 (10) P < 0.001 (13) 0-15 0.740 0.608 0.737 P < 0.001 (13) P < 0.001 (17) P < 0.001 (17) 0-20 0.743 0.603 0.736 P < 0.001 (15) P < 0.001 (19) P < 0.001 (20) a 412 90 ocean color product correlated positively with in situ chlorophyll, especially for the surface to 10 meter depth interval (Table 4.3). Although the coefficients of determination were lower for depth ranges deeper than 10 meters, the correlation was still significant (p<0.05) for depths up to 20 m between both datasets. Two parameters can be compared 1:1 between the in situ dataset and the satellite products: the absorption in the short wavelength a 412 and chlorophyll concentration. The estimated a 412 from the remote sensing algorithms underestimated the in situ observed values along the channel with a bias up to ~60% (Fig. 4.3A). Table 4.2. Correlations between in situ observations of total attenuation at 650 nm (c 650 ) (from the AC-s and transmissometer) vs. the satellite estimated c 645 . Numbers in parenthesis are the sample number N. In situ Depth Parameter (m) QAA TSS Gould c 650 0-5 0.983 0.833 P = 0.009 (4) P = 0.004 (7) 0-10 0.955 0.831 P = 0.023 (4) P = 0.004 (7) 0-15 0.998 0.704 P < 0.001 (4) P = 0.002 (10) 0-20 0.989 0.722 P = 0.001 (4) P = 0.001 (11) c 650 (Trans) 0-5 0.050 0.877 P > 0.05 (4) P = 0.002 (7) 0-10 0.469 0.874 P > 0.05 (4) P = 0.002 (7) 0-15 0.997 0.690 P < 0.001 (4) P = 0.003 (10) 0-20 0.979 0.729 P = 0.001 (4) P = 0.001 (11) c 645 91 Table 4.3. Correlations between in situ observations of chlorophyll concentration (C a ) and chlorophyll fluorescence vs. the estimated absorption by phytoplankton (a ph(443) ) using the different ocean color models. Numbers in parenthesis are the sample number N. This underestimation occurs for low and high absorption values. The OC3 model underestimated in situ phytoplankton biomass, with a bias from the measured C a up to ~96% (Fig. 3B) when in situ C a is lower than ~0.2 m•gm -3 . Spatial-Temporal correlations The horizontal distribution of the satellite estimated parameters corresponded to the in situ observations from the surface to 10 meters depth along the transect (Fig. 4.4). It is assumed that the time required to complete the in situ transect (~1 hr) is quasi-synoptic, In situ Depth Parameter (m) Carder QAA GSM Gould OC3 Ca 0-5 0.925 0.884 0.864 0.955 0.834 P < 0.001 (10) P < 0.001 (13) P < 0.001 (12) P < 0.001 (5) P = 0.004 (7) 0-10 0.932 0.895 0.875 0.964 0.856 P < 0.001 (10) P < 0.001 (12) P < 0.001 (11) P < 0.001 (7) P = 0.003 (7) 0-15 0.896 0.742 0.869 0.809 0.710 P < 0.001 (13) P < 0.001 (18) P < 0.001 (16) P < 0.001 (11) P = 0.002 (10) 0-20 0.862 0.714 0.845 0.799 0.690 P < 0.001 (15) P < 0.001 (19) P < 0.001 (17) P < 0.001 (12) P = 0.002 (11) Chl_f 0-5 0.910 0.884 0.887 0.918 0.864 P < 0.001 (10) P < 0.001 (13) P < 0.001 (12) P = 0.001 (7) P = 0.002 (7) 0-10 0.911 0.894 0.903 0.917 0.900 P < 0.001 (10) P < 0.001 (14) P < 0.001 (12) P = 0.001 (7) P = 0.001 (7) 0-15 0.858 0.648 0.823 0.673 0.594 P < 0.001 (13) P < 0.001 (17) P < 0.001 (17) P = 0.004 (15) P = 0.009 (10) 0-20 0.806 0.595 0.761 0.645 0.534 P < 0.001 (15) P < 0.001 (19) P < 0.001 (18) P = 0.003 (11) P = 0.011 (11) a ph(443) 92 i.e., the changes in the cross-channel gradient are relatively stationary for that period of time. The similarity between the cross-channel distribution of the in situ data and the essentially instantaneous snapshot from the MODIS-Aqua sensor, indicates that this assumption is appropriate. Water masses observations with MODIS The higher spatial resolution (250 m) of the MODIS ocean color products enabled delimitate the distribution of phytoplankton biomass in the upper 10 meters Figure 4.3. A) Correlation between in situ vs. remotely sensed estimated dissolve absorption at 412 nm (a 412 ). B) Correlation between in situ chlorophyll concentration (C a_acs ) and the chlorophyll concentration estimated with the standard ocean color OC3 algorithm for MODIS (C a(sat) ). 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 a 412(in situ) (m -1 ) a 412(sat) (m -1 ) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 C a(sat) (mg . m -3 ) C a_acs(in situ) (mg . m -3 ) A B Carder QAA GSM Gould OC3 93 Figure 4.4. Horizontal distribution along the channel of in situ and estimated products from MODIS: A) absorption at 412 (a 412 ), B) beam attenuation c 650 and c 645 , C) in situ chlorophyll concentration (C a_acs ) and satellite phytoplankton absorption at 443 (a ph443 ), and D) in situ chlorophyll concentration (C a_acs ) and the chlorophyll concentration estimated with the standard ocean color OC3 algorithm for MODIS (C a(sat) ). 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.000 0.025 0.050 0.075 0.100 0.000 0.025 0.050 0.075 0.100 a 412(sat) (m -1 ) a 412(in situ) (m -1 ) 0.00 0.25 0.50 0.75 1.00 0.05 0.10 0.15 0.20 0.25 C 645(sat) (m -1 ) C 650(in situ) (m -1 ) 0.00 0.01 0.02 0.03 0.04 0.0 0.2 0.4 0.6 0.8 a ph443(sat) (m -1 ) C a_acs(in situ) (mg . m -3 ) C a(sat) (mg . m -3 ) C a_acs(in situ) (mg . m -3 ) 124.14 124.18 124.26 124.22 Longitude (Deg) 124.14 124.18 124.26 124.22 Longitude (Deg) A B C D In situ Carder QAA GSM Gould OC3 94 across the San Bernardino strait. In general, dissolved absorption, a 412 , was higher in interior waters of San Bernardino than in exterior oceanic water (Fig. 4.5). On February 16 th during an early flood a marked gradient was observed outside San Bernardino, with remaining interior waters still outside the channel, and open ocean water entering the channel (Fig. 4.5A); the GSM model revealed the presence of this intrusion from the open ocean along the coast in the west side of the channel (Fig. 4.5A). During a late flood (February 18 th ) higher concentration of dissolved materials from the interior are mixed with oceanic water, resulting in values of a 412 ~0.03 m -1 for the three models used (Fig. 4.5B). Higher concentrations of dissolved materials are present inside the San Bernardino strait during high tide (February 19 th ), though there is a gradient from high concentrations to lower concentrations towards the open ocean (Fig. 4.5C). Higher absorptions at 443 nm and chlorophyll concentrations are observed inside the San Bernardino Strait than northward toward the Pacific Ocean during the early flood (Fig. 4.6 & 4.7, respectively). The Carder algorithm showed for February 16 th interior waters with higher a ph(443) outside the channel towards east (Fig. 4.6A). During the late flood and high tide, a ph(443) values outside the channels are lower (Fig. 4.6B & C). The same water mass was observed for C a using the OC3 model, which also showed the presence of lower C a from oceanic waters along the west side of the channel (Fig. 4.7A). During loate flood and high tide, oceanic waters with lower phytoplankton biomass during enter the channel (Fig. 4.7B & C). 95 Figure 4.5. Total absorption at 412 nm (a 412 ) estimated with the GSM algorithm for: A) February 16, B) February 18, and C) February 19. The yellow line represents the in situ transect. . 96 Figure 4.6. Absorption by phytoplankton at 443 nm (aph(443) estimated with the Carder mode for: A) February 16, B) February 18, and C) February 19. 97 Figure 4.7. Chlorophyll concentration (C a ) estimated with the OC3 MODIS algorithm for: A) February 16, B) February 18, and C) February 19. 98 4.6 DISCUSSION In this study we successfully gather quasi-synoptic in situ data coinciding with MODIS satellite overpass for the first time from a narrow strait in the Philippines archipelago. This in situ methodology characterized the distribution of the inherent optical properties contributing to the ocean color in the area of the San Bernardino Strait. We found that the 250 m resolution ocean color satellite models estimate the patterns of absorption and concentration of suspended materials quite well, and distinguish differences in optical properties across the strait at a broader spatial and temporal scale. Data retrieved directly from the 250 m resolution MODIS bands has being successfully used to estimate the concentration of suspended particles in coastal areas (e.g. Hu et al., 2004; Miller & Mckee 2004; Petus et al., 2010). The 250-m resolution ocean color products have been also previously validated (e.g. Ladner et al., 2007), but the present study is the first attempt to validate them with quasi-synoptic in situ data in a highly dynamic archipelago where strong tidal currents and complex circulation patterns have been observed (Meñez et al., 2006; Jones et al., 2011; Ohlmann, 2011). Bailey and Werdell (2006) suggested averaging neighbor pixels around the location of the field station (5x5 pixel box) to correct for individual pixels containing errors or no data during validation procedures of ocean color products. This method is commonly applied to MODIS standard products with 1 km spatial resolution, though we found that this method was not the most appropriate for evaluating MODIS 250-m spatial resolution products in the ~15 km across San Bernardino Strait. In an effort to retrieve as much information as 99 possible from the satellite imagery we extracted data from individual 250-m pixels coinciding geographically with the TRIAXUS sampling locations. The calculated coefficients of correlation between the in situ and the satellite retrievals were > 0.5 for all parameters (Tables 4.1 to 4.3), even when the in situ dataset was integrated down to 20 m deep. Both datasets showed the same spatial and temporal trends when compared within a period of ~1 hr along the sampling transect. The major differences between these two datasets occur when comparing their absolute values, where remotely sensed a 412 and Ca underestimated the measured in situ values. The deviation found from the 1:1 linear relation could be overcome with the calculated linear fits: ! !"#(!"##) = ! !"#(!"#) + 0.0022 0.635 and ! !(!"##) = ! !(!"#) + 0.1912 1.017 where a 412(sat) is the estimated absorption at 412 from the remote sensing ocean color model, a 412(corr) is the tuned or corrected absorption, C a(sat) is the concentration of chlorophyll estimated with the OC3 algorithm, and C a(corr) is the tuned chlorophyll concentration value. These approximations are appropriate for the period of time (and 100 tidal stage) this San Bernardino area was studied and must be further fine-tuned with future validation campaigns. The underestimation of in situ values by remote sensors is generally observed in coastal areas due to the presence of regional atmospheric features. For short wavelengths, absorbing aerosols can bias satellite retrievals of water reflectance at ~412 (Werdell et al., 2007; Wang et al., 2010), consequently underestimating the dissolved fraction using the standard ocean color algorithms. Suspended sediments (also common in coastal areas) magnify this bias as they reflect more strongly in the near infrared (Wang et al., 2010). Validation of satellite imagery for numerous coastal regions have resulted in local algorithms to overcome the biases that result from the standard ocean color algorithms, designed for open ocean waters. For example in the Baltic Sea, CDOM retrievals using the standard Carder and GSM algorithms for MODIS resulted in a upward bias of 70% (Kowalczuk et al., 2010). In the Baltic Sea, CDOM absorption can be as much as 80% of the total absorption of blue light (Kowalchuk, et al., 2005). In France, MODIS total suspended sediments retrievals of sediments from the Adour River were also validated with in situ data but in this case data bias was dependent on turbidity levels (Petus et al., 2010). In the Sea of Japan, chlorophyll concentrations estimated with the standard OC3 algorithm underestimate measured in situ concentration when C a > 3 mg•m -3 (Salyuk et al., 2010). Along the Patagonian continental shelf, chlorophyll retrievals using the MODIS OC3 model underestimated in situ measured values by 7 to 32% (Dogliotti et al., 2009). In contrast to other satellite ocean color validations in coastal areas where only surface samples (<10 m depth) were used (Dogliotti et al., 2009, Kowalczuk et al., 2010, 101 Petus et al., 2010, Salyuk et al., 2010), we found significant correlations with in situ inherent optical properties down to ~20 m depth in the San Bernardino Strait. Although, our results demonstrate a promising tool for studying coastal processes and water component at a broader spatial scale and down to 20 m depth, extrapolation to deeper areas in the water column, especially for complex areas like straits, should not be attempted without appropriate validation. Subsurface processes can change the concentration and distribution of suspended materials in the water. In the case of San Bernardino strait, we observed a large decrease in the concentration of particles and C a below ~ 40 m (Fig. 4.2). We also observed the intrusion of oceanic water during flooding tide, and residual inward water masses outside the straits from a previous ebb stage (Fig. 4.5, 4.6 & 4.7); these processes contribute to the materials horizontally and vertically mixing during tidal changes. For San Bernardino, this tidally driven vertical mixing was reported by Jones et al. (2011), showing the aspiration of deeper particles and nutrients to surface waters around underwater sills, consequently enhancing phytoplankton development. In conclusion, we have demonstrated that towed vehicles provide reliable in situ measurements for validating and tuning ocean color algorithms applied to high spatial and temporal resolution coastal satellite imagery. The accurate geo positioning of towed vehicles, plus their quasi-synoptic sampling capacities make them ideal for monitoring spatial and temporal variations of water column optical properties and biogeochemical processes in dynamic coastal systems. We validated MODIS 250 m resolution data for the San Bernardino strait for the dissolved (a 412 ), particulate (TSS and c 650 ) and 102 chlorophyll estimators (C a and a 443 ). Although, we found a bias for the estimated C a and a 412 , high correlations between in situ and satellite-derived inherent optical properties indicate the possibility of using these retrievals for future studies. Furthermore, we demonstrated that GSM model provides better estimates of dissolved organic matter for the region of the San Bernardino Strait, and the Carder model better estimates C a distributions. The combination of these models, plus the quasi-synoptic in situ methodology allowed the study of the spatial temporal distribution of dissolve and particulate materials and phytoplankton biomass, highlighting the complexity and dynamics of different water masses circulation in coastal areas like straits. 103 SUMMARY Ocean color and in situ bio-optical measurements have advanced the understanding of the components and dynamics of the planktonic community in the open ocean, nevertheless challenges of interpreting optical measurements from in situ and satellite sensors in coastal areas still exist. The high complexity of suspended and dissolved materials in coastal waters due to terrigenous inputs, resuspension of sediments or algae blooms among others cause confounding scattering and absorption signals hard to differentiate without hyperspectral sensors. In addition, rapid dynamics occurring in coastal areas require high spatial and temporal resolution optical measurements to study the distribution of water biogeochemical components. In order to contribute to the knowledge in marine optics from coastal areas, the spatial and temporal variations of in situ hyperspectral optical properties and hydrographic features were studied in two contrasting island environments: a dynamic strait in the Philippine Islands (San Bernardino strait) and a shallow coral reef area from Bermuda Islands. In addition, standard ocean color remote sensing algorithms were compared with in situ optical measurements from San Bernardino strait in order to validate these algorithms in a tidally driven narrow strait. The main objectives of this thesis deal with: 1. Characterize the inherent optical properties (IOPs) of biogeochemical components found in San Bernardino Strait (Philippines) and Great Sound Bay (Bermuda); and 2. Use quasi-synoptic hyperspectral 104 in situ optical data to evaluate satellite ocean color products in the highly dynamic waters of San Bernardino strait. Based on hydrographic and optical measurements, it was observed that during the ebb tide in the San Bernardino strait the outflowing water was characterized by higher chlorophyll concentrations originating in the more productive areas inside the strait. Conversely, oceanic waters entering into the strait during the flood phase of the tides were characterized by lower chlorophyll concentration, lower total water attenuation and smaller sized suspended particles, accompanied by higher temperatures, lower salinities, lower densities, and higher oxygen concentration. Cluster analyses categorized the biogeochemical optical components finding 3 absorption and 4 scattering groups in San Bernardino strait. These groups mapped closely to water masses previously identified by Jones et al (2011) for the same area using standard temperature and salinity observations. This optical characterization of water masses enabled us to more clearly identify transport and mixing and gave additional insight than temperature and salinity alone. Derivative analysis corroborated the presence of phytoplanktonic species typical of interior coastal areas and the association of specific absorption spectral groups with the “Interior surface water” endpoint classification. Ocean color satellite data from San Bernardino strait correlated well with in situ biogeochemical components. Even though these ocean color models underestimated absolute values for chlorophyll concentration and absorption by dissolved materials, they 105 accurately mapped the spatial distribution of biogeochemical variables in the strait during different tidal stages. In the Bermuda Islands, a well defined gradient of biogeochemical components was observed from inshore to the open, were high chlorophyll concentration, high dissolved absorption and high concentration of small particles were present close to land. Cluster analyses well discriminated hyperspectral absorption and scattering observations in 3 groups each, also distributed spatially from inland to out shore. The effects water level on the distribution of water constituents was observed with a significant inverse correlation with dissolved absorption at 400 nm and chlorophyll concentration, and a positive correlation with volume concentration and origin of suspended particles. Due to the high productivity of coastal area and its importance in the global ocean there is a need to better characterize water optical properties in these areas and, hence improve our understanding of how the coastal ocean respond to short-term events. The study of the spatial-temporal variability of optical properties in coastal areas and the physical factors that determine their distribution will help understanding how natural and anthropogenic disturbances affect coastal systems. This will lead to better monitoring of long-term changes due to climate and sea level change. Statistical analyses such as cluster analyses not generally used in marine optics showed to be a powerful tool for discriminating absorption and scattering spectral shapes and amplitudes. Cluster analyses can also help discriminating water masses based on the spectral characteristics of the water constituents. This technique combined with derivative analyses on hyperspectral data can also be used for identifying phytoplankton 106 populations in coastal areas specially when algal blooms formed by dominant species are present. Statistical analyses of in situ data combined with satellite products at 250 m resolution open the possibility of using satellite imagery with higher spatial resolution for future modeling and monitoring of optical properties not only in coastal areas but also in the open ocean when studying mixing of water masses, algal blooms and the distribution of biogeochemical components. 107 REFERENCES Ackleson SG (2006) Optical determinations of suspended sediment dynamics in western Long Island Sound and the Connecticut River plume. J. Geophys. Res. 111:C07009. Arnone RA, Wood M, Gould Jr. RW (2004) The Evolution of Optical Water Mass Classification. Oceanography 17:14-15. Babin M, Stramski D, Ferrari GM, Claustre H, Bricaud A, Obolensky G, Hoepffner N (2003) Variations in the light absorption coefficients of phytoplankton, nonalgal particles, and dissolved organic matter in coastal waters around Europe. J. Geophys. Res. 108:3211. Bailey SW, Hooker SB, Antoine D, Franz BA, Werdell PJ (2008) Sources and assumptions for the vicarious calibration of ocean color satellite observations. Appl. Opt. 47:2035-2045. Bailey SW, Werdell PJ (2006) A multi-sensor approach for the on-orbit validation of ocean color satellite data products. Remote Sensing of Environment 102:12-23. Barnes DJ, Lough JM (1999) Porites growth characteristics in a changed environment: Misima Island, Papua New Guinea. Coral Reefs 18:213-218. Barth JA, Pierce SD, Castelao RM (2005) Time-dependent, wind-driven flow over a shallow midshelf submarine bank. J. Geophys. Res. 110:C10S05. Barth JA, Bogucki DJ (2000) Spectral light absorption and attenuation measurements from a towed undulating vehicle. Deep Sea Research Part I: Oceanographic Research Papers 47:323-342. Barth JA, Cowles TJ, Pierce SD (2001) Mesoscale physical and bio-optical structure of the Antarctic Polar Front near 170W during austral spring. J. Geophys. Res. 106:13879-13902. Behrenfeld MJ, Falkowski PG (1997) Photosynthetic rates derived from satellite-based chlorophyll concentration. Limnology and Oceanography 42:1-20. 108 Bidigare RR, Morrow JH, Kiefer DA (1989) Derivative analysis of spectral absorption by photosynthetic pigments in the western Sargasso Sea. Journal of Marine Research 47:323-341. Birkeland C (1997) Life and death of coral reefs. Chapman & Hall. Bissett WP, Arnone RA, Davis CO, Dickey TD, Dye D, Kohler DDR, Gould Jr. RW (2004) From Meters to Kilometers: A Look at Ocean-Color Scales of Variability, Spatial Coherence, and the Need for Fine-Scale Remote Sensing in Coastal Ocean Optics. Oceanography 17:32-43. Blanco AC, Nadaoka K, Yamamoto T (2008) Planktonic and benthic microalgal community composition as indicators of terrestrial influence on a fringing reef in Ishigaki Island, Southwest Japan. Marine environmental research 66:520-35. Blondeau-Patissier D, Brando VE, Oubelkheir K, Dekker AG, Clementson LA, Daniel P (2009) Bio-optical variability of the absorption and scattering properties of the Queensland inshore and reef waters, Australia. J. Geophys. Res. 114:C05003. Bodungen B von, Jickells TD, Smith SR, Ward JAD, Hillier GB (1982) The Bermuda Marine Environment. Bermuda Biological Station, St Georges West, Bermuda. Boss E, Pegau WS, Lee M, Twardowski MS, Shybanov E, Korotaev G, Baratange F (2004) Particulate backscattering ratio at LEO 15 and its use to study particle composition and distribution. J. Geophys. Res. 109:C01014. Boss E, Pegau WS, Gardner WD, Zaneveld JRV, Barnard AH, Twardowski MS, Chang GC, Dickey TD (2001) Spectral particulate attenuation and particle size distribution in the bottom boundary layer of a continental shelf. J. Geophys. Res. 106:9509- 9516. Boss E, Slade WH, Behrenfeld M, DallʼOlmo G (2009) Acceptance angle effects on the beam attenuation in the ocean. Opt. Express 17:1535-1550. Boss E, Zaneveld JRV (2003) The effect of bottom substrate on inherent optical properties: Evidence of biogeochemical processes. Limnology and Oceanography 48:346-354. Bricaud A, Claustre H, Ras J, Oubelkheir K (2004) Natural variability of phytoplanktonic absorption in oceanic waters: Influence of the size structure of algal populations. J. Geophys. Res. 109:C11010. 109 Bricaud A, Stramski D (1990) Spectral absorption coefficients of living phytoplankton and nonalgal biogenous matter: A comparison between the Peru upwelling areaand the Sargasso Sea. Limnology and Oceanography 35:562-582. Capone DG (1996) Coral reef ecosystems in the context of the marine nitrogen cycle. In: Bjork M, Semesi A, Pedersen M, Bergman B (eds) Current trends in marine botanical research in the East African region. Ord & Vetande, Uppsala, pp 61-76. Carder KL, Chen FR, Lee ZP, Hawes SK, Kamykowski D (1999) Semianalytic Moderate-Resolution Imaging Spectrometer algorithms for chlorophyll a and absorption with bio-optical domains based on nitrate-depletion temperatures. J. Geophys. Res. 104:5403-5421. Chazottes A, Bricaud A, Crepon M, Thiria S (2006) Statistical analysis of a database of absorption spectra of phytoplankton and pigment concentrations using self- organizing maps. Appl. Opt. 45:8102-8115. Chazottes A, Crepon M, Bricaud A, Ras J, Thiria S (2007) Statistical analysis of absorption spectra of phytoplankton and of pigment concentrations observed during three POMME cruises using a neural network clustering method. Appl. Opt. 46:3790-3799. Ciotti AM, Lewis MR, Cullen JJ (2002) Assessment of the relationships between dominant cell size in natural phytoplankton communities and the spectral shape of the absorption coefficient. Limnology and Oceanography 47:404-417. Clark DK, Feinholz ME, Yarbroughb MA, Johnson BC, Browne SW, Kim YS, Barnes RA (2002) Overview of the radiometric calibration of MOBY. Proceedings of SPIE 4483:64-76. Cleveland JS, Perry MJ (1994) A model for partitioning particulate absorption into phytoplanktonic and detrital components. Deep Sea Research Part I: Oceanographic Research Papers 41:197-221. Coble PG, Hu C, Gould Jr. RW, Chang G, Wood AM (2004) Colored Dissolved Organic in the Coastal Ocean: An Optical Tool for Coastal Zone Environmental Assessment and Management. Oceanography 17:50-59. COSTANZA R, DʼARGE R, GROOT RD, FARBER S, GRASSO M, HANNON B, LIMBURG K, NAEEM S, OʼNEILL RV, PARUELO J, RASKIN RG, SUTTON P, BELT MVD (1997) The value of the worldʼs ecosystem services and natural capital. Nature 387:253-260. 110 Cunningham A, McKee D, Craig S, Tarran G, Widdicombe C (2003) Fine-scale variability in phytoplankton community structure and inherent optical properties measured from an autonomous underwater vehicle. Journal of Marine Systems 43:51-59. DallʼOlmo G, Westberry TK, Behrenfeld MJ, Boss E, Slade WH (2009) Significant contribution of large particles to optical backscattering in the open ocean. Biogeosciences 6:947-967. Del Castillo CE (2005) Remote Sensing of Organic Matter in Coastal Waters. In: Miller RL, Del Castillo CE, Mckee BA (eds) Remote Sensing of Coastal Aquatic Environments. Springer Netherlands, pp 157-180. Dierssen HM, Zimmerman RC, Burdige DJ (2009) Optics and remote sensing of Bahamian carbonate sediment whitings and potential relationship to wind-driven Langmuir circulation. Biogeosciences 6:487-500. Dogliotti AI, Schloss IR, Almandoz GO, Gagliardini DA (2009) Evaluation of SeaWiFS and MODIS chlorophyll-a products in the Argentinean Patagonian Continental Shelf (38° S–55° S). International Journal of Remote Sensing 30:251-273. Dupouy C, Neveux J, Ouillon S, Frouin R, Murakami H, Hochard S, Dirberg G (2010) Inherent optical properties and satellite retrieval of chlorophyll concentration in the lagoon and open ocean waters of New Caledonia. Marine Pollution Bulletin 61:503- 518. Dupouy C, Savranski T, Lefevre J, Despinoy M, Mangeas M, Fuchs R, Faure V, Ouillon S, Petit M (2010) Monitoring optical properties of the southwest tropical Pacific. In: Frouin RJ, Yoo HR, Won J-S, Feng A (eds) Remote Sensing of the Coastal Ocean, Land, and Atmosphere Environment. SPIE, p 78580J. Edinger EN, Jompa J, Limmon GV, Widjatmoko W, Risk MJ (1998) Reef degradation and coral biodiversity in indonesia: Effects of land-based pollution, destructive fishing practices and changes over time. Marine Pollution Bulletin 36:617-630. Eisner LB, Cowles TJ (2005) Spatial variations in phytoplankton pigment ratios, optical properties, and environmental gradients in Oregon coast surface waters. J. Geophys. Res. 110:C10S14. Eisner LB, Twardowski MS, Cowles TJ, Perry MJ (2003) Resolving phytoplankton photoprotective : photosynthetic carotenoid ratios on fine scales using in situ spectral absorption measurements. Limnology and Oceanography 48:632-646. 111 Esaias WE, Abbott MR, Barton I, Brown OB, Campbell JW, Carder KL, Clark DK, Evans RH, Hoge FE, Gordon HR, Balch WM, Letelier R, Minnett PJ (1998) An overview of MODIS capabilities for ocean science observations. IEEE Transactions on Geoscience and Remote Sensing 36:1250-1265. Faust MA, Norris KH (1985) In vivo spectrophotometric analysis of photosynthetic pigments in natural populations of phytoplankton. Limnology and Oceanography 30:1316-1322. Fichez R, Chifflet S, Douillet P, Gerard P, Gutierrez F, Jouon A, Ouillon S, Grenz C (2010) Biogeochemical typology and temporal variability of lagoon waters in a coral reef ecosystem subject to terrigeneous and anthropogenic inputs (New Caledonia). Marine Pollution Bulletin 61:309-322. Fortes MD (2004) Wetland Conservation and Management in the Philippines: Where are We Now? The Case of Seagrass and Mangrove. In: Asia MHWBT-WE in (ed) Elsevier, Amsterdam, pp 233-262. Franz BA, Werdell PJ, Meister G, Kwiatkowska EJ, Bailey SW, Ahmad Z, Mcclain CR (2006) MODIS Land Bands for Ocean Remote Sensing Applications. In: Proc. Ocean Optics XVIII. Montreal, pp 1-16. Gardner WD, Mishonov AV, Richardson MJ (2006) Global POC concentrations from in- situ and satellite data. Deep Sea Research Part II: Topical Studies in Oceanography 53:718-740. Garver SA, Siegel DA (1997) Inherent optical property inversion of ocean color spectra and its biogeochemical interpretation 1. Time series from the Sargasso Sea. J. Geophys. Res. 102:18607-18625. Gordon AL, Sprintall J, FField A (2001) Regional Oceanography of the Philippine archipelago. Oceanography 24:14-27. Hoitink AJF (2004) Tidally-induced clouds of suspended sediment connected to shallow- water coral reefs. Marine Geology 208:13-31. Horner-Devine AR (2009) The bulge circulation in the Columbia River plume. Continental Shelf Research 29:234-251. Horvath H (1993) Atmospheric light absorption--A review. Atmospheric Environment. Part A. General Topics 27:293-317. 112 Hu C, Cannizzaro J, Carder KL, Muller-Karger FE, Hardy R (2010) Remote detection of Trichodesmium blooms in optically complex coastal waters: Examples with MODIS full-spectral data. Remote Sensing of Environment 114:2048-2058. Hu C, Chen Z, Clayton TD, Swarzenski P, Brock JC, Muller-Karger FE (2004) Assessment of estuarine water-quality indicators using MODIS medium-resolution bands: Initial results from Tampa Bay, FL. Remote Sensing of Environment 93:423- 441. Hunter PD, Tyler AN, Presing M, Kovacs AW, Preston T (2008) Spectral discrimination of phytoplankton colour groups: The effect of suspended particulate matter and sensor spectral resolution. Remote Sensing of Environment 112:1527-1544. IOCCG (2006) Remote sensing of inherent optical properties: fundamentals, tests of algorithms, and applications. In: Lee ZP (ed) Reports of the International Ocean- Colour Coordinating Group, No. 5. IOCCG, Dartmouth, Canada, p 126. Jay DA, Pan J, Orton PM, Horner-Devine AR (2009) Asymmetry of Columbia River tidal plume fronts. Journal of Marine Systems 78:442-459. Jensen HS, McGlathery KJ, Marino R, Howarth RW (1998) Forms and availability of sediment phosphorus in carbonate sand of Bermuda seagrass beds. Limnology and Oceanography 43:799-810. Johnson KS, Berelson WM, Boss E, Chase Z, Claustre H, Emerson S, Gruber N, Körtzinger A, Perry MJ, Riser SC (2009) Observing Biogeochemical Cycles at global with profiling floats and gliders. PROSPECTS FOR A GLOBAL ARRAY. Oceanography 22:216-225. Jones BH, Lee CM, Toro-Farmer G, Boss ES, Gregg MC, Villanoy CL (2011) Tidally Driven Exchange in an Archipelago Strait: Biological and Optical Responses. Oceanography 24:142-155. Jouon A, Ouillon S, Douillet P, Lefebvre JP, Fernandez JM, Mari X, Froidefond J-M (2008) Spatio-temporal variability in suspended particulate matter concentration and the role of aggregation on size distribution in a coral reef lagoon. Marine Geology 256:36-48. Karp-Boss L, Azevedo L, Boss E (2007) LISST-100 measurements of phytoplankton size distribution : evaluation of the effects of cell shape. Limnology and Oceanography: Methods:396-406. Kirk JTO (1994) Light and photosynthesis in aquatic ecosystems. Cambridge University Press. 113 Kirkpatrick G, Millie DF, Moline MA, Schofield O (2000) Optical discrimination of a phytoplankton species in natural mixed populations. Limnology and Oceanography 45:467-471. Kowalczuk P, Olszewski J, Darecki M, Kaczmarek S (2005) Empirical relationships between coloured dissolved organic matter (CDOM) absorption and apparent optical properties in Baltic Sea waters. International Journal of Remote Sensing 26:345-370. Kowalczuk P, Darecki M, Zabłocka M, Górecka I (2010) Validation of empirical and semi-analytical remote sensing algorithms for estimating absorption by Coloured Dissolved Organic Matter in the Baltic Sea from SeaWiFS and MODIS imagery *. Oceanologia 52:171-196. Ladner SD, Sandidge JC, Lyon PE, Arnone RA, Gould RW, Lee ZP, Martinolich PM (2007) Development of finer spatial resolution optical properties from MODIS. In: Frouin RJ, Lee Z (eds) Coastal Ocean Remote Sensing. SPIE, San Diego, CA, USA, pp 668013-668019. Larcombe P, Ridd PV, Prytz A, Wilson B (1995) Factors controlling suspended sediment on inner-shelf coral reefs, Townsville, Australia. Coral Reefs 14:163-171. Lee Z, Carder KL, Arnone RA (2002) Deriving Inherent Optical Properties from Water Color: a Multiband Quasi-Analytical Algorithm for Optically Deep Waters. Appl. Opt. 41:5755-5772. Maritorena S, Siegel DA, Peterson AR (2002) Optimization of a semianalytical ocean color model for global-scale applications. Appl. Opt. 41:2705-2714. McCulloch M, Fallon S, Wyndham T, Hendy E, Lough J, Barnes D (2003) Coral record of increased sediment flux to the inner Great Barrier Reef since European settlement. Nature 421:727-730. McKee D, Chami M, Brown I, Calzado VS, Doxaran D, Cunningham A (2009) Role of measurement uncertainties in observed variability in the spectral backscattering ratio: a case study in mineral-rich coastal waters. Appl. Opt. 48:4663-4675. Meñez LAB, Villanoy CL, David LT (2006) Movement of Water Across Passages Connecting Philippine Inland Sea Basins. Science Diliman 18:2:10-17. Miller RL, McKee BA (2004) Using MODIS Terra 250 m imagery to map concentrations of total suspended matter in coastal waters. Remote Sensing of Environment 93:259- 266. 114 Millie D, Schofield O, Kirkpatrick G, Johnsen G, Evens T (2002) Using absorbance and fluorescence spectra to discriminate microalgae. European Journal of Phycology 37:313-322. Millie D, Schofield O, Kirkpatrick G, Johnsen G, Tester PA, Vinyard BT (1997) Detection of harmful algal blooms using photopigments and absorption signatures: A case study of the Florida red tide dinoflagellate, Gymnodinium breve. Limnology and Oceanography 42:1240-1251. Mobley CD (1994) Light and water: radiative transfer in natural waters. Academic Press. Mobley CD, Stramski D, Bissett WP, Boss E (2004) Optical Modeling of Ocean Waters: Is the Case 1 - Case 2 Classification Still Useful? Oceanography 17:60-67. Morel A, Gentili B, Claustre H, Babin M, Bricaud A, Ras J, Tieche F (2007) Optical properties of the clearest natural waters. Applied optics 52:217-229. Morel A, Prieur L (1977) Analysis of variations in ocean color. Limnology and Oceanography 22:709-722. Morel A (1988) Optical Modeling of the Upper Ocean in Relation to Its Biogenous Matter Content (Case I Waters). Journal of Geophysical Research 93:10749-10768. Morris B, Barnes J, Brown F, Markham J (1977) The Bermuda Marine Environment. A report of the Bermuda inshore waters investigations 1976-1977. Bermuda Biological Station, St Georges West, Bermuda. Muller-Karger FE, Hu C, Andréfouët S, Varela R, Thunell R (2005) The Color of the Coastal Ocean and Applications in the Solution of Research and Management Problems. In: Miller RL, Del Castillo CE, Mckee BA (eds) Remote Sensing of Coastal Aquatic Environments. Springer Netherlands, pp 101-127. Murdoch T, Glasspool A, Outerbridge M, Ward J, Manuel S, Gray J, Nash a, Coates K, Pitt J, Fourqurean J, Barnes P, Vierros M, Holzer K, Smith S (2007) Large-scale decline in offshore seagrass meadows in Bermuda. Marine Ecology Progress Series 339:123-130. Nencioli F, Chang G, Twardowski M, Dickey TD (2010) Optical Characterization of an Eddy-induced Diatom Bloom West of the Island of Hawaii. Biogeosciences 7:151- 162. Ogston AS, Storlazzi CD, Field ME, Presto MK (2004) Sediment resuspension and transport patterns on a fringing reef flat, Molokai, Hawaii. Coral Reefs:559-569. 115 Ohlmann JC Drifter Observations of Small-Scale Flows in the Philippine Archipelago. Oceanography 24:122-129. Oliver MJ, Schofield O, Bergmann T, Glenn S, Orrico C, Moline M (2004) Deriving in situ phytoplankton absorption for bio-optical productivity models in turbid waters. J. Geophys. Res. 109:C07S11. Otis DB, Carder KL, English DC, Ivey JE (2004) CDOM transport from the Bahamas Banks. Coral Reefs 23:152-160. Ouillon S, Douillet P, Lefebvre JP, Le Gendre R, Jouon A, Bonneton P, Fernandez JM, Chevillon C, Magand O, Lefevre J, Le Hir P, Laganier R, Dumas F, Marchesiello P, Bel Madani A, Andrefouet S, Panche JY, Fichez R (2010) Circulation and suspended sediment transport in a coral reef lagoon: The south-west lagoon of New Caledonia. Marine Pollution Bulletin 61:269-296. Para J, Coble PG, Charrière B, Tedetti M, Fontana C, Sempéré R (2010) Fluorescence and absorption properties of chromophoric dissolved organic matter (CDOM) in coastal surface waters of the northwestern Mediterranean Sea, influence of the Rhône River. Biogeosciences 7:4083-4103. Paulay G (1997) Diversity and distribution of reef organisms. In: Birkeland C (ed) Life and death of coral reefs. Chapman & Hall, pp 298-353. Perry MJ, Sackmann BS, Eriksen CC, Lee CM (2008) Seaglider observations of blooms and subsurface chlorophyll maxima off the Washington coast. Limnology and Oceanography 53:2169-2179. Peterson JO, Peterson WT (2008) Influence of the Columbia River plume (USA) on the vertical and horizontal distribution of mesozooplankton over the Washington and Oregon shelf. ICES Journal of Marine Science 65:477-483. Petrenko AA, Jones BH, Dickey TD, LeHaitre M, Moore C (1997) Effects of a sewage plume on the biology, optical characteristics, and particle size distributions of coastal waters. J. Geophys. Res. 102:25061-25071. Petus C, Chust G, Gohin F, Doxaran D, Froidefond J-M, Sagarminaga Y (2010) Estimating turbidity and total suspended matter in the Adour River plume (South Bay of Biscay) using MODIS 250-m imagery. Continental Shelf Research 30:379- 392. Piniak G a, Storlazzi CD (2008) Diurnal variability in turbidity and coral fluorescence on a fringing reef flat: Southern Molokai, Hawaii. Estuarine, Coastal and Shelf Science 77:56-64. 116 Rao YR, Murthy CR, Sinha PC (2008) The coastal ocean. In: Murthy CR, Sinha P, Rao YR (eds) Modelling and monitoring of coastal marine processes. Springer, pp 3-10. Reynolds RA, Stramski D, Wright VM, Woźniak SB (2010) Measurements and characterization of particle size distributions in coastal waters. J. Geophys. Res. 115:C08024. Roesler CS, Perry MJ, Carder KL (1989) Modeling in situ phytoplankton absorption from total absorption spectra in productive inland marine waters. Limnology and Oceanography 34:1510-1523. Rogers CS (1990) Responses of coral reefs and reef organisms to sedimentation. Marine Ecology Progress Series 62:185-202. Salomonson VV, Barker J, Knight E (1833) Spectral Characteristics of the Earth Observing System (EOS) Moderate Resolution Imaging Spectroradiometer (MODIS). SPIE 2480:142-152. Salyuk P, Bukin O, Alexanin A, Pavlov A, Mayor A, Shmirko K, Akmaykin D, Krikun V (2010) Optical properties of Peter the Great Bay waters compared with satellite ocean colour data. International Journal of Remote Sensing 31:4651-4664. Schofield O, Arnone RA, Bissett WP, Dickey TD, Davis CO, Finkel Z, Oliver M, Moline MA (2004) Watercolors in the Coastal Zone: what can we see? Oceanography 17:24-31. Schowengerdt RA (2007) Remote sensing: models and methods for image processing. Academic Press. Short F, Carruthers T, Dennison W, Waycott M (2007) Global seagrass distribution and diversity: A bioregional model. Journal of Experimental Marine Biology and Ecology 350:3-20. Smith SR (1998) Bermuda. In: Kjerfve B (ed) CARICOMP – Caribbean coral reef, seagrass and mangrove sites. Coastal region and small island papers 3. UNESCO, Paris,. Stramski D, Bricaud A, Morel A (2001) Modeling the Inherent Optical Properties of the Ocean Based on the Detailed Composition of the Planktonic Community. Appl. Opt. 40:2929-2945. Sullivan JM, Twardowski MS, Donaghay PL, Freeman SA (2005) Use of optical scattering to discriminate particle types in coastal waters. Applied optics 44:1667- 1680. 117 Sullivan JM, Twardowski MS, Zaneveld JRV, Moore CM, Barnard AH, Donaghay PL, Rhoades B (2006) Hyperspectral temperature and salt dependencies of absorption by water and heavy water in the 400-750 nm spectral range. Appl. Opt. 45:5294-5309. Szmant A (2002) Nutrient enrichment on coral reefs: Is it a major cause of coral reef decline? Estuaries and Coasts 25:743-766. Twardowski MS, Claustre H, Freeman SA, Stramski D, Huot Y (2007) Optical backscattering properties of the “clearest” natural waters. Biogeosciences 4:1041- 1058. Twardowski MS, Boss E, Macdonald JB, Pegau WS, Barnard AH, Zaneveld JRV (2001) A model for estimating bulk refractive index from the optical backscattering ratio and the implications for understanding particle composition in case I and case II waters. J. Geophys. Res. 106:14129–14142. Twardowski MS, Boss E, Sullivan JM, Donaghay PL (2004) Modeling the spectral shape of absorption by chromophoric dissolved organic matter. Marine Chemistry 89:69- 88. Twardowski MS, Lewis MR, Barnard AH, Zaneveld JRV (2005) In-Water Instrumentation and Platforms for Ocean Color Remote Sensing Applications. In: Miller RL, Del Castillo CE, Mckee BA (eds) Remote Sensing of Coastal Aquatic Environments. Springer Netherlands, pp 69-100. Unesco (1997) Phytoplankton pigments in oceanography: guidelines to modern methods. UNESCO Publishing. Volpe V, Silvestri S, Marani M (2011) Remote sensing retrieval of suspended sediment concentration in shallow waters. Remote Sensing of Environment 115:44-54. Wang J-J, Lu XX, Liew SC, Zhou Y (2010) Remote sensing of suspended sediment concentrations of large rivers using multi-temporal MODIS images: an example in the Middle and Lower Yangtze River, China. International Journal of Remote Sensing 31:1103-1111. Werdell PJ, Franz B a, Bailey SW, Harding Jr. LW, Feldman GC (2007) Approach for the long-term spatial and temporal evaluation of ocean color satellite data products in a coastal environment. Proceedings of SPIE 6680:12. Woźniak B, Dera J (2007) Light absorption in sea water. Springer. 118 Xiong X, Wenny BN, Barnes WL (2009) Overview of NASA Earth Observing Systems Terra and Aqua moderate resolution imaging spectroradiometer instrument calibration algorithms and on-orbit performance. Journal of Applied Remote Sensing 3:032501. Zanardi-Lamardo E, Moore CA, Zika RG (2004) Seasonal variation in molecular mass and optical properties of chromophoric dissolved organic material in coastal waters of southwest Florida. Marine Chemistry 89:37-54. Zaneveld JRV, Kitchen JC, Moore CC (1994) Scattering error correction of reflecting- tube absorption meters. In: Jaffe JS (ed) Ocean Optics XII. SPIE, Bergen, Norway, pp 44-55. Zaneveld JRV, Twardowski MJ, Barnard A, Lewis MR (2005) Introduction to Radiative Transfer. In: Miller RL, Del Castillo CE, Mckee BA (eds) Remote Sensing of Coastal Aquatic Environments. Springer Netherlands, pp 1-20. Zhao S, Wei J, Yue H, Xiao T (2010) Picophytoplankton abundance and community structure in the Philippine Sea, western Pacific. Chinese Journal of Oceanology and Limnology 28:88-95.

Linked assets