Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Revised calibration of a long-term solar extreme ultraviolet irradiance data set
(USC Thesis Other)
Revised calibration of a long-term solar extreme ultraviolet irradiance data set
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
REVISED CALIBRATION OF A LONG-TERM SOLAR EXTREME ULTRAVIOLET
IRRADIANCE DATA SET
by
Seth Wieman
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(ASTRONAUTICAL ENGINEERING)
May 2015
ii
To the memory of my advisor Professor Darrell L. Judge
iii
Table of Contents
Acknowledgments.................................................................................................................. v
List of Figures ...................................................................................................................... vii
List of Tables ...................................................................................................................... xiii
Abstract ............................................................................................................................... xiv
1. Introduction ........................................................................................................................ 1
2. The Solar Extreme Ultraviolet and Soft X-ray Spectral Ranges ..................................... 10
2.1 Solar Spectrum ........................................................................................................... 10
2.2 Solar Variability ......................................................................................................... 12
2.3 History of EUV measurements .................................................................................. 19
2.4 Solar EUV proxies ..................................................................................................... 20
3. EUV and SXR irradiance and the geospace environment ............................................... 23
3.1 EUV irradiance and the study of the ionosphere ....................................................... 23
3.2 The role of EUV irradiance in the study of the thermosphere ................................... 30
4. Current space weather applications for which SEM data are specifically relevant ......... 38
4.1 The S10 index and the Jacchia Bowman thermospheric density model .................... 38
4.2 Understanding inter-minima EUV changes ............................................................... 41
5. SOHO/CELIAS/SEM ...................................................................................................... 45
5.1 Instrument Overview ................................................................................................. 45
5.2 SEM Sounding Rocket Calibration ............................................................................ 47
5.2.1 SEM clone sounding rocket instrument .............................................................. 50
5.2.2 Neon Rare Gas Ionization Cell ........................................................................... 52
5.3 SEM 26-34 nm channel ............................................................................................. 54
5.3.1 Revised Channel response function .................................................................... 54
5.3.2. Data processing algorithm ................................................................................. 57
5.4 SEM 0.1-50 nm channel ............................................................................................ 59
5.4.1 Channel response function .................................................................................. 59
5.4.2 Zeroth-order data processing algorithm .............................................................. 61
iv
6. SDO/EVE ......................................................................................................................... 63
6.1 SDO/EVE/ESP ........................................................................................................... 64
6.2 SDO/EVE/MEGS ...................................................................................................... 64
6.3 A comparison of EVE/MEGS spectra to SOLERS22 ............................................... 66
7. Resolving Differences between SOHO/SEM and SDO/EVE ......................................... 68
7.1 Comparisons of the 26-34 nm daily average time series ........................................... 68
7.2 Comparison of 26-34 nm measurements under flare conditions ............................... 73
7.3 Comparison of the daily average 0.1-7 nm and 7-50 nm time series ........................ 76
8. Pre-SDO Time Dependent reference spectra ................................................................... 80
8.1 Evaluation of FISM reference spectra ....................................................................... 82
8.2 Evaluation of Solar2000/SIP reference spectra ......................................................... 84
8.3 Evaluation of a system of discrete MEGS-A spectra ................................................. 86
8.4 Spectra based on SOHO CDS measurements and DEM modeling ........................... 93
8.5 Inter-comparison of FISM, Solar2000, ESP algorithm, and CDS spectra ................. 96
9. Re-processing the SEM data throughout the SOHO Mission ........................................ 103
9.1 Reprocessing the SEM sounding rocket data and degradation model ..................... 103
9.2 Comparison to EVE/ESP measured degradation ..................................................... 108
9.3 Re-processed SEM irradiance time series for the entire SOHO mission................. 113
9.4 SEM comparison with solar indices ........................................................................ 117
10. Calculating SEM equivalent irradiances based on EVE measurements ...................... 122
11. Conclusions .................................................................................................................. 126
References .......................................................................................................................... 129
Appendix: List of Related Publications ............................................................................. 139
Papers ............................................................................................................................. 139
Conference and Workshop Talks ................................................................................... 139
v
Acknowledgments
This work was carried out during a time of transition at USC’s Space Sciences Center (SSC)
which has given me the opportunity to work under the guidance of two highly experienced and
productive mentors, the late Professor Darrell L. Judge and his successor as SSC director,
Professor Leonid Didkovsky. They have both been great sources of wisdom regarding the
science, instrumentation, engineering, and business elements of our research. I am extremely
grateful for the opportunities that they have given me and the knowledge that they have shared.
I also deeply appreciate their concern for aspects of life outside of work.
I would like to thank the chair of my committee, Professor Mike Gruntman for his kind and
friendly encouragement throughout this work. I also thank committee members Professor
Joseph Wang, Professor Ed Rhodes, and Professor Joseph Kunc for their interest in this
research and the effort they put in to help improve it.
I am also grateful to the late Dr. Howard Ogawa, a long-time Research Scientist at SSC, Solar
EUV Monitor (SEM) Co-Investigator, and a pioneer in techniques for measuring solar EUV
irradiance. Even long into retirement he remained an enthusiastic source of both historical
perspective and novel insight.
I want to thank Marjorie Judge for her kindness and support throughout my time at SSC. She
would often pick up the phone when I would call Professor Judge to report complications that
had arisen during a critical phase of a project. Seasoned no doubt by decades of experience
with such situations, she has always provided a voice of calm and reassurance.
Don McMullin of the Space Systems Research Corporation and Andrew Jones of the
Laboratory for Atmospheric and Space Physics at the University of Colorado, Boulder have
both been major contributors to the SEM program during their time at SSC and since, and I
want to thank them for their many discussions regarding the SEM data. I would also like to
thank Tom Woods, Frank Eparvier, Phil Chamberlin and the rest of the SDO EVE Team for a
very productive collaboration. I am grateful to our colleagues at NIST: Rob Vest, Mitch Furst,
Alex Farrell, and Charles Tarrio, for their support of the SEM and EVE instrument
calibrations. I would also like to acknowledge colleagues and collaborators, too numerous to
list, in the SOHO, SDO and EUV irradiance communities without whose efforts this work
would not have been possible.
I thank my Astronautics course instructors, Dr. Michael Gabor for his Orbital Mechanics class,
and Prof. Mike Gruntman for his class in Spacecraft Design. I would especially like to thank
Dr. W. Kent Tobiska, whose Space Environment and Spacecraft Interactions class and work in
this area had a major influence on this dissertation.
vi
I thank Nancy Lin of SSC and Dell Cuason, Norma Perry, and Marrietta Penoliar of the
Astronautical Engineering Department for their logistical help throughout the completion of
this work, and fellow graduate student Vaibhav Gupta with whom I have had the pleasure of
many discussions about strategies for completing our degrees.
Finally, I would like to thank my family – I am most indebted to my wife, Nina, for her
constant support, advice, and sound scientific judgment. Were it not for her I would not have
taken on this endeavor. My parents, Leslie and Howard, and mother-in-law, Liza, have been a
tremendous source of support. My children, Leonard, Ronni, and Natalie have helped me to
maintain a healthy perspective throughout this effort, and my brother Ben and his wife Laura
have also helped to keep me in good humor about it.
SOHO is a project of International cooperation between the European Space Agency and the
National Aeronautics and Space Administration. This research has been supported in part by
University of Colorado award 153-5979, and NASA grants NNX08BA12G, and
NNX08AM94G.
vii
List of Figures
Figure 1.1 Photon flux time series in the 26-34 nm band pass measured with the Version
3.1 SOHO/SEM first-order channels (small black circles) and in the 0.1-50 nm bandpass
measured with the SOHO/SEM zeroth-order channel (small gray circles). Sounding rocket
calibration measurements include: RGIC full 5-57.5 nm band (triangles), SEM sounding
rocket clone 26-34 nm band (squares), RGIC scaled to 26-34 nm (diamonds), and
EVE/ESP sounding rocket clone (large gray circles) ....................................................................... 2
Figure 1.2 Comparison of 26-34nm irradiance measured by SOHO CDS (x), SOHO
SEM(*), and SOHO EIT(+) (after Thompson, McMullin, and Newmark, 2002). ........................... 5
Figure 1.3 Comparison of 26-34nm irradiance measured by TIMED SEE and SOHO SEM
from Woods et al., 2005. ................................................................................................................... 5
Figure 2.1 Solar photon spectrum for the soft x-ray through infrared region. The
SOHO/SEM broadband measurements include the 0.1-50 nm band (SEM 0th order; shown
in pink) and the 26-34 nm band (SEM 1st order; shown in orange)............................................... 11
Figure 2.2 Temperature profile of the solar atmosphere (From Yang et al., 2009). The peak
emission temperatures of several ion emission lines are shown. .................................................... 11
Figure 2.3 Concurrent images of the Sun from the Atmospheric Imaging Assembly (AIA)
aboard SDO. Each image is filtered to include wavelengths originating from a different
layer of the solar atmosphere. ......................................................................................................... 15
Figure 2.4 SOHO SEM 26-34 nm irradiance time series for the rise of Solar Cycle 24.
Strong solar rotation variability is evident particularly starting on Julian Day 2456100. .............. 16
Figure 2.5 – EUV and SXR irradiances observed during an X6.9 class flare. The 26-34 nm
channel exhibits impulsive phase variability while the 0.1-50 nm signal (which saturates at
~.02 W/m
2
) includes both gradual phase and impulsive phase components. ................................. 18
Figure 2.6 Solar cycle, solar rotation and flare related (impulsive and gradual phase)
variability of EUV and SXR emissions versus wavelength............................................................ 19
Figure 3.1 Densities of neutral and ionized species and electrons in the Earth’s upper
atmosphere. The composition is for quiet Sun daytime conditions (with the exception of He
profile which is based on a night time measurement). Density values below 250 km are
from sounding rocket borne mass spectrometer measurements taken over White Sands
Missile Range, NM. (from Johnson, C. Y.: 1969, Ion and neutral composition of the
ionosphere, Ann. IQSY, 5, reprinted in Kelly, 1989) ...................................................................... 24
Figure 3.2 – Energy Deposition rates in the thermosphere due to absorption of solar EUV
and SXR photons from Viereck (2004a) as a function of wavelength. Photons at 30.4 nm
and 28.4 nm due to solar He-II and Fe XV emissions respectively deposit energy at a
notably high rate. Emissions from these two spectral lines are measured by the SEM 1st
order (26-34 nm) channel................................................................................................................ 33
viii
Figure 3.3 – Flow diagram showing the path by which EUV photon energy is either
transferred to the neutral gas or radiatively lost (from Stolarski, 1975) ......................................... 33
Figure 3.4 Thermosphere temperature profiles associated with a) low EUV irradiance under
solar minimum conditions and b) with high solar irradiance under solar maximum
conditions. Results from Tobiska’s energy conservation-based model (thin curves) are
compared with the MSIS 83 empirical model (thick curves). Plots are from Tobiska, 1998,
with labels added to identify the profiles. ....................................................................................... 34
Figure 3.5 Thermosphere number densities corresponding to the temperature profiles of
Figure 3.4 for a) solar minimum and b) solar maximum conditions (from Tobiska 1988). ........... 36
Figure 4.1 – A comparison of the JB2008 (red diamonds and fit curve) thermospheric
density model (which uses the SEM based S10 index of EUV irradiance) with the F10.7
based NRLMSIS (grey diamonds and fit curve) and Jacchia 70 (purple diamonds and fit
curve). The models are compared based on agreement with the HASDM (described in text)
“observed” density values. Plotted values are HASDM/model for 400km density values
binned according to F ̅ 10 (labeled F10B) solar activity level for the period from 2001-2007
(from Bowman et al. 2008). JB2008 shows better agreement with HASDM (ratios closer to
unity) at most levels of activity, but most notably at lower F ̅ 10 levels. ......................................... 40
Figure 4.2 Comparison of changes during Solar Cycle 23 in a) thermospheric density at an
altitude of 400 km (after Emmert et al., 2010) and b) Global Mode Total Electron Content
(from Lean et al. 2011a). For the density values in a) the top panel shows daily values and
the bottom panel shows 81-day average values; the green dashed line in the bottom channel
shows the long-term trend of decreasing thermospheric density associated with increasing
atmospheric CO2 concentrations described in Emmert et al., (2008). For the TEC time
series in b) daily values are shown. ................................................................................................ 42
Figure 5.1 Opto-mechanical layout of the SOHO/CELIAS/SEM .................................................. 46
Figure 5.2 – Raw data from the SEM clone instrument obtained during a sounding rocket
calibration flight. ChA and ChC are the SEM first-order (26-34 nm) channels. ChB refers
to the zeroth-order (0.1-50 nm) channel. The difference signal levels for the two first-order
channels is due to differences in dark current levels between the two photodiode detectors
as well as asymmetry in the transmission grating +1 and -1 order diffraction efficiencies. .......... 51
Figure 5.3 – Schematic of the RGIC absolute detector .................................................................. 52
Figure 5.4 – Transmission of the residual atmosphere above ~160 km based on RGIC flight
data (blue circles) and the NRLMSISe-00 thermospheric density model. The above-
atmosphere irradiance, I0 is determined as the value which provides the best fit between the
measurements and the model. ......................................................................................................... 54
Figure 5.5 SOHO/SEM first-order response function used for irradiance calculation in
Version 3.1 (black curve partially obscured by the gray curve which is identical over the
wavelength range from 26 to 34 nm) and in the updated version reported here (gray line).
The SOLERS22 spectrum is shown for reference (dotted line). .................................................... 56
ix
Figure 5.6 – Source of additional sensitivity “peaks” in the SEM response profile (gray
curves on either side of the primary response peak in Figure 5.5). The green lines outline
the path of ~40 nm photons diffracted in the first-order or ~20 nm photons diffracted in the
second order and reflected off of the inner surfaces of the SEM housing. Sensitivity related
to this effect is limited to wavelengths with diffraction angles between 10.65°
(corresponding to ~17 nm, purple lines) and 13.5° (corresponding to ~47 nm, red lines). ............ 56
Figure 5.7 – Soft x-ray (top panel) and EUV (bottom panel) portions of the SOHO/SEM
zeroth-order response function used for irradiance calculations in both Version 3.1 and in
the updated version reported here (black line). The SOLERS22 spectrum is shown for
reference (dotted line, units are arbitrary but consistent between the top and bottom panels). ...... 60
Figure 6.1 A comparison of the SOLERS22 fixed reference spectrum (dotted line) to
measured SDO/EVE reference spectra (black line) for low (4/30/2010, top panel), medium
(3/21/2013, middle panel), and high (11/11/2011, bottom panel) levels of solar activity –
the SDO/EVE spectra shown are the daily average spectra for the specified day. In the top
panel the SOHO/SEM first-order response function (dashed line) is shown for reference.
Time-independent, characteristic differences exist between the spectra including a smaller
contribution from the wavelength bins centered at 28.5 nm and 30.5 for the EVE/MEGS
spectrum for all activity levels. ....................................................................................................... 67
Figure 7.1 a) A comparison of the updated SOHO/SEM 26-34 nm irradiance time series
(black curve) with SOHO/SEM Version 3.1 (dotted black curve), SDO/EVE/MEGS (dotted
gray curve) Version 2 spectra integrated over the 26-34 nm bandpass, and EVE ESP ch9
(gray curve). b) Ratio of the updated SOHO/SEM daily average irradiances over those of
SOHO/SEM Version 3.1. c) Ratios of the SOHO/SEM Version 3.1 daily average
irradiances over the daily average SDO/EVE/ESP Channel 9 irradiances compared to
similar ratios based on the updated SOHO/SEM. d) Ratios of the SOHO/SEM Version 3.1
daily average irradiances over the daily average integrated SDO/EVE/MEGS spectra
compared to similar ratios based on the updated SOHO/SEM. Mean ratio levels are shown
as dashed lines in panels c and d. Mean ratios for Version 3.1 SEM are about 1.26 for the
comparison with ESP (panel c) and 1.28 for the comparison with MEGS, but are much
closer to unity (~1.01 in panel c and ~ 1.03 in panel d) for the updated SEM. The broader
response function and the use of MEGS reference spectra for the updated SOHO/SEM
irradiances result in significantly better agreement with both SDO/EVE channels.
Divergence between the SOHO/SEM and SDO/EVE/ESP Ch9 irradiances in 2012 could be
related to higher sensitivity of the SEM to energetic particles compared to ESP or to errors
related to using daily average reference spectra on days with high solar activity (see text for
details). ............................................................................................................................................ 72
Figure 7.2 a) A comparison of the updated SOHO/SEM 26-34 nm irradiance time series
(black curve) with SOHO/SEM Version 3.1 (dotted curve), and SDO/EVE/MEGS Version
2 (gray curve) spectra integrated over the 26-34 nm bandpass with high time resolution
during the X1.7, N33W85 solar flare of Jan 27, 2012 . b) The ratios of the updated
SOHO/SEM irradiances over those of SOHO/SEM Version 3.1 vary by several percent
over the course of the flare due to the rapid change of solar spectral distribution during the
x
flare. c) Ratios of the SOHO/SEM Version 3.1 irradiances over the integrated
SDO/EVE/MEGS Ver. 2 spectra deviate significantly from their pre flare values, while
ratios of the daily average updated SOHO/SEM over the daily average integrated
SDO/EVE/MEGS spectra (d) show little change. .......................................................................... 75
Figure 7.3 a) A comparison of the 7-50 nm SOHO/SEM irradiance time series extracted
from the zeroth-order measurements for the updated version (black curve) with that for
Version 3.1 (dotted curve), and with daily average 7-50 nm MEGS integrated spectra (gray
curve). b) Ratios of the SOHO/SEM Version 3.1 daily average irradiances over the daily
average SDO/EVE/MEGS spectra integrated from 7-50 nm (dotted line) compared to
similar ratios based on the updated SOHO/SEM (solid black line). Mean ratio levels are
shown as dashed lines in panel b. The mean ratio for Version 3.1 SEM is about 0.84 but is
much closer to unity (~ 1.03) for the updated SEM. ...................................................................... 78
Figure 7.4 a) A comparison of the 0.1-7 nm SOHO/SEM irradiance time series extracted
from the zeroth-order measurements for the updated version (black curve) with that for
Version 3.1 (dotted curve), and with daily average 0.1-7 nm ESP measurements (gray
curve). b) Ratios of the SOHO/SEM Version 3.1 daily average irradiances over the daily
average SDO/EVE/ESP 0.1-7 nm measurements compared to similar ratios based on the
updated SOHO/SEM. Mean ratio levels are shown as dashed lines in panel b. The mean
ratio for Version 3.1 SEM is about 1.4 but is closer to unity (~ 1.15) for the updated SEM.
Because the soft x-ray response functions and reference spectra are modeled for both of the
SOHO/SEM versions and for the SDO/EVE/ESP QD channel (Didkovsky et al., 2012),
large differences among datasets are not unexpected in the 0.1-7 nm range. ................................. 79
Figure 8.1 a) A comparison of ESP 26-34 nm irradiances determined using FISM daily
reference spectra (blue line) to SEM Ver 3.1 (dotted line), EVE MEGS (black line), and
EVE ESP (gray line). Irradiance ratios for the FISM based SEM irradiances over SEM Ver
3.1, EVE ESP, and EVE MEGS are shown in panels b, c, and d respectively with mean
ratios shown as dashed lines in each panel. .................................................................................... 83
Figure 8.2 a) A comparison of ESP 26-34 nm irradiances determined using Solar2000 daily
reference spectra (blue line) to SEM Ver 3.1 (dotted line), EVE MEGS (black line), and
EVE ESP (gray line). Irradiance ratios for the FISM based SEM irradiances over SEM Ver
3.1, EVE ESP, and EVE MEGS are shown in panels b, c, and d respectively with mean
ratios shown as dashed lines in each panel. .................................................................................... 85
Figure 8.3 Histogram showing the number of days within each of the 11 activity levels
defined based on ESP daily average zeroth-order effective DN (as shown in Table 8.1). The
plot covers the interval from 30 April 2010 through 26 May 2014. From Didkovsky,
Wieman, and Woodraska, (2014b). ................................................................................................ 88
Figure 8.4 Time series of ESP zeroth-order effective DN (black curve) compared to SEM
zeroth-order effective DN (red curve). Due to differences in band pass and instrument
sensitivty the SEM zeroth-order effective DN values are higher than those from ESP and
have been scaled by a factor of 1:20 for this comparison. .............................................................. 89
xi
Figure 8.5 Correlation plot between SEM and ESP zeroth-order effective DN (blue circles)
shows a strong (rcor = 0.964) quadratic correlation between the data sets ...................................... 90
Figure 8.6 a) ESP 26-34 nm irradiances (blue line) determined using a system of 11
reference spectra established for processing EVE ESP data following the EVE MEGS-A
anomaly compared to SEM Ver 3.1 (dotted line), EVE MEGS (black line), and EVE ESP
(gray line). Irradiance ratios for the FISM based SEM irradiances over SEM Ver 3.1, EVE
ESP, and EVE MEGS are shown in panels b, c, and d respectively with mean ratios shown
as dashed lines in each panel. ......................................................................................................... 92
Figure 8.7 – Comparison of spectra from FISM, Solar2000, and the ESP data processing
algorithm with CDS-based spectra and SOLERS for a day near the maximum (MAX) of
solar cycle 23 (30 October 2001) and a day near the minimum (MIN) of solar cycles 23 and
24 (31 May 2010). The comparison includes the spectral range affecting the SEM
26-34 nm irradiance calculations. For solar MAX, the FISM spectrum provides the closest
match to the CDS ............................................................................................................................ 97
Figure 8.8 – Comparison of SEM 26-34 nm irradiance values determined based on
reference spectra from FISM (solid green line), Solar2000 (dashed red line), and the ESP
algorithm (dotted blue line). All three time series are in good agreement and virtually
indistinguishable in the plot suggesting that the SEM irradiance values do not depend
strongly on which of the three time-dependent reference spectra is used. ................................... 101
Figure 9.1 SOHO/SEM Version 3.1 photon flux values (black dots) are in nominally good
agreement with sounding rocket irradiance values determined from the SEM-clone (grey
squares) and the RGIC (gray diamonds) raw data using the Version 3.1 SEM response
function and SOLERS22 reference spectrum, but are systematically higher than the
EVE/ESP on orbit (green dots) and sounding rocket (grey circles) flux values. .......................... 104
Figure 9.2 – A comparison of modeled contaminant layer growth based on sounding rocket
measurements processed with the SOLERS22 reference spectrum (red curve) as used in
Version 3.1 of the SEM data and processed using the FISM reference spectrum (blue
curve) and the updated SEM response function (for the SEM clone measurements). The
updated model incorporates an additional sounding rocket flight (36.263 on July 24, 2012)
that was not included in the determination of the Version 3.1 degradation curve. ....................... 108
Figure 9.3 – Ratios of signal observed using an ESP redundant filter over that observed
with the primary filter provides a measurement of signal loss due to contamination of the
primary filter. Such measurements are taken as part of the EVE/ESP daily in-flight
calibration. .................................................................................................................................... 110
Figure 9.4 - Transmission of contaminant layer built up on the ESP primary filter based on
filter wheel degradation measurements (red squares) after 1100 days of operation compared
to Henke model transmission for a 33 nm thick layer of C. ......................................................... 112
Figure 9.5 - ESP filter #3 contaminant layer thickness versus time. The contaminant layer
continues to grow at an exponentially decreasing rate. ................................................................ 113
xii
Figure 9.6 – SEM time series recalculated using the FISM reference spectrum and the
updated response function and degradation model. The time series is in good agreement
with both the reprocessed sounding rocket data from the SEM-clone (grey squares) and the
RGIC (gray diamonds), and the EVE/ESP on-orbit (green dots) and sounding rocket
measurements (grey circles) in contrast to the Ver. 3.1 data shown in Figure 9.1. The inter-
minima change in photon flux is slightly lower for the recalculated time series (only ~12%
compared to ~15% according to the Version 3.1 data). ................................................................ 114
Figure 9.7 Top panel: A comparison of SOHO/SEM updated 26-34 nm irradiances to
irradiances in the same band from the ESP and MEGS channels of the SDO/EVE
instrument. Middle panel: daily ratio of SEM over MEGS (blue curve) plotted with dashed
lines showing the mean ratio (1.0013) and ratio STD (0.048). Bottom panel: daily ratio of
SEM over ESP (green curve) plotted with dashed lines showing the mean ratio (0.9995)
and ratio STD (0.0587). ................................................................................................................ 116
Figure 9.8 – Comparison of SEM 26-34 nm time series with the F10.7 index (F10.7 daily
and 81 day running mean values are averaged and linear fit to the SEM irradiances). The
well-documented lower sensitivity of F10.7 to variations around solar minimum is evident
as are differences during other intervals (e.g. around 2001 and between 2012 and 2013),
however with about the same signals for the peaks between 2011 and 2012. These time
intervals are isolated and not indicative of a steady long-term relative trend between time
series ............................................................................................................................................. 118
Figure 9.9 – Comparisons of SEM 26-34 nm time series with the daily Mg-II core to wing
ratio linear fit to the SEM data. The Mg-II index shows larger solar rotation variability and
differences are evident at isolated intervals, but the good agreement over much of the time
series (in the beginning and end in particular) does not suggest a steady long-term relative
trend. The Mg-II index shows signal levels similar to SEM for the two latest solar minima
(e.g. about 10% decrease for the 23/24 minimum as reported by Solomon, S.C. and Qian,
L.: 2011) and similar variability following the 2008 minimum, e.g. for the peaks between
2011 and 2012. .............................................................................................................................. 120
Figure 10.1 Reconstruction of SEM Version 3.1 irradiance values based on EVE ESP
channel 9 which has been found to be in good agreement with the updated SEM irradiance
values determined based on this work. ......................................................................................... 125
xiii
List of Tables
Table 3.1 Ionosphere regions and the processes by which they are formed (summarized
from Chamberlain and Hunten, 1987) ............................................................................................ 25
Table 5.1. SEM sounding rocket calibration measurements compared to concurrent on-orbit
measurements with SOHO/SEM in the 26-34 nm band. Also shown are measurements from
a similar bandpass channel of the EVE/ESP obtained from the SDO/EVE series of
sounding rocket calibration flights. The SEM Version 3.1 degradation model parameters
were established based on sounding rocket flights 36.147 through 36.236 shown in bold
type. ................................................................................................................................................. 49
Table 5.2. SEM sounding rocket calibration measurements using the RGIC compared to
concurrent on-orbit measurements with SOHO/SEM in the 0.1-50 nm band. The SEM
Version 3.1 degradation model parameters were established based on sounding rocket
flights 36.147 through 36.236 shown in bold type. ........................................................................ 50
Table 8.1: Ranges of ESP zeroth-order daily average effective DN corrected for zeroth-
order degradation used to select a reference spectrum based on activity level: ............................. 87
Table 8.2: Ranges of SEM zeroth-order daily average effective DN corrected for zeroth-
order degradation used to select a reference spectrum based on activity level: ............................. 91
Table 8.3 SEM 26-34 nm irradiances for a 30 October 2001 (solar MAX) and 31 May 2010
(solar MIN) determined based on reference spectra from SOHO CDS, FISM, Solar 2000,
the ESP data algorithm, EVE MEGS (for 2010 only), and SOLERS22....................................... 100
Table 9.1 – SEM clone and RGIC sounding rocket measurements recalculated using the
FISM time dependent reference spectrum (SEM-clone values are calculated using the
updated response function described in section 5.3.1) ................................................................. 106
xiv
Abstract
Solar irradiance measurements in the extreme ultraviolet (EUV) and soft x-ray (SXR)
spectral ranges are important to studies of solar variability and its impact on the geospace
environment and to operations affected by space weather. While the availability of solar EUV
measurements has increased over the last two decades, data from the Solar EUV Monitor (SEM),
part of the Charge, Element, and Isotope Analysis System (CELIAS) on board the Solar and
Heliospheric Observatory (SOHO) remain unique in that they are continuous with high time
cadence over a long time period which includes two solar minima, and their accuracy has been
maintained based on a long series of sounding rocket calibration underflights, resulting in their use
as the basis for a solar activity index for space weather operations. The purpose of this work is to
further improve the absolute calibration of the SEM EUV irradiance measurements using data
unavailable during earlier SEM calibrations. Solar EUV variability occurs over a range of
timescales, including periodic changes associated with the 11-year solar sunspot cycle (i.e. the 22-
year solar magnetic cycle). Thus, long-term stable irradiance measurements are important in order
to understand this variability and its relation to long-term changes in the geospace environment.
Maintaining the absolute accuracy of irradiance measurements is required to avoid data
inconsistencies and the misinterpretation of instrument-related bias when compiling long-term data
sets comprised of measurements from more than one instrument.
xv
Although analyses of the SEM absolute calibration have been ongoing since the launch of
SOHO and have incorporated measurements from a series of sounding rocket underflights, in
recent years additional data have become available which have created opportunities and
motivation for further refinement of the SEM measurements. Firstly, converting the SEM raw data
into irradiance values depends on the instrument response function and on the spectral distribution
of solar irradiance (i.e. reference spectrum) within the SEM sensitivity range, and both of these
parameters have been determined with greater accuracy than those used in SEM data processing to
date. Secondly, reliable EUV irradiance measurements in a spectral range overlapping that of SEM
are available from the Solar Dynamics Observatory’s EUV Variability Experiment (SDO/EVE)
which includes provisions for in-flight calibration and degradation monitoring, and these
measurements differ from concurrent SEM values (obtained using the original response function
and reference spectrum) by an amount that is consistent throughout the SDO and SOHO mission
overlap suggesting there may be a systematic offset in the SEM irradiance values. Thirdly, the
SEM measurements show lower (by ~15%) irradiance values for the minimum of Solar Cycles 22
and 23 compared to Solar Cycles 23 and 24 – this inter-minima change is consistent with the
response of the earth’s upper atmosphere over the same period with regard to some solar EUV-
driven processes (i.e. thermospheric neutral density) but inconsistent with others (i.e. global mode
ionospheric total electron content). Further investigation is thus required to determine whether the
lower EUV irradiance measured by SEM is real or an artifact of long-term instrument degradation
Resolving these three issues concerning the SEM data set is the objective of this
dissertation, and the included work has several key results. It demonstrates that the differences
between SOHO/SEM and SDO/EVE EUV irradiance measurements are resolved by reprocessing
xvi
the SEM raw data using an updated SEM instrument response function (introduced as part of this
work) and time-dependent solar reference spectra. Additionally, it provides an updated time and
wavelength dependent SEM instrument degradation function which is necessary to refine the
estimate of inter-minima change in EUV irradiance based on the SEM data. Finally, this work
provides a procedure for producing a SEM-equivalent EUV irradiance index based on SDO/EVE
measurements. The motivation for this final effort is that SOHO/SEM irradiance values based on
the previous response function and reference spectrum have already been adopted as a solar
irradiance index used for modeling thermospheric density, and because such space weather
operations are concerned with long-term consistency, it is desirable to continue providing an
irradiance index equivalent to that provided thus far by SEM using newer EUV instrumentation
after the SOHO mission has ended.
1. Introduction
Absolute solar irradiance measurements in the highly variable X-ray and EUV spectral
ranges have both fundamental research value and practical utility. Because solar photons in this
wavelength range are absorbed in the Earth’s upper atmosphere (unity optical depths occur at
altitudes between about 120 and 230 km – Viereck, 2004), observing them directly requires
satellite or rocket-borne instrumentation. Thus, such measurements were non-existent prior to the
space age and remained sparse through the 1980s. They have however, become increasingly
available over the last two decades, with EUV (EUV) and soft X-ray (SXR) irradiance instruments
aboard multiple satellites including, SOHO (Harrison, Sawyer, and Carter, 1995; Hovestadt et al,
1995), SDO (Woods et al., 2012), TIMED (Woods et al., 2005), SORCE (Woods and Rottman,
2005), ISS (Nikutowski et al., 2011), PROBA2 (Dominique et al., 2013), and GOES (Evans et al.,
2010). Among these, SOHO/CELIAS/SEM (Judge et al., 1998) 26-34 nm and 0.1-50 nm datasets
(Figure 1.1) are unique in that they include high time cadence (15 s), continuous (with the
exception of the brief SOHO mission interruptions of 1998) measurements that now span nearly 20
years and include two solar minima (data are available at http://www.usc.edu/dept/space_science/).
Furthermore, the absolute calibration of the SEM data is maintained based on measurements from
a long series of sounding rocket flights using a SEM clone instrument and a Neon Rare Gas
Ionization Cell (RGIC) absolute detector.
2
Figure 1.1 Photon flux time series in the 26-34 nm band pass measured with the Version 3.1
SOHO/SEM first-order channels (small black circles) and in the 0.1-50 nm bandpass measured
with the SOHO/SEM zeroth-order channel (small gray circles). Sounding rocket calibration
measurements include: RGIC full 5-57.5 nm band (triangles), SEM sounding rocket clone 26-
34 nm band (squares), RGIC scaled to 26-34 nm (diamonds), and EVE/ESP sounding rocket clone
(large gray circles)
The SEM data are the source of the S10.7 solar EUV irradiance proxy (Tobiska, Bouwer,
and Bowman, 2006; Bowman and Tobiska, 2008), and they are widely used for inter-comparison
with other EUV instrumentation (Thompson, McMullin, and Newmark, 2002; McMullin et al.,
3
2002; Woods et al., 2005; Wieman, Judge, and Didkovsky 2011; Didkovsky et al., 2010b) and as a
basis for validating EUV irradiance models (Chamberlin, Woods, and Eparvier, 2007; Haberreiter
et al., 2014; Haberreiter, 2013). Additionally, the SEM data are central to debates over whether
and by how much, solar EUV irradiance was lower during the minimum of Solar Cycles 23/24
compared to that of Solar Cycles 22/23. The SEM Version 3.1 data (the most recent version
available prior to the work reported here) suggest that 26-34 nm irradiance was 15% lower during
the latter minimum (Didkovsky et al, 2010a), an assertion that is supported by thermospheric data
(Solomon et al., 2011) but not by ionospheric global mode total electron content (TEC) data (Lean
et al., 2011, and Didkovsky and Wieman, 2014a, though the latter publication shows that spatially
resolved sectoral harmonic TEC values were lower for the solar cycles 23/24 minimum). For these
applications, and for continuation of the longstanding EUV record established by the SEM using
newer EUV instrumentation, it is important to understand the instrumental and data processing
factors that may affect the accuracy of the SEM absolute irradiances.
The current period of overlap between the SOHO and SDO missions has provided a good
opportunity to evaluate the accuracy of the SEM irradiance measurements based on comparisons
with concurrent, independently calibrated measurements from the Solar Dynamics Observatory’s
EUV Variability Experiment (SDO/EVE: Woods et al, 2012; Hock et al., 2012; Didkovsky et al.,
2012). The EVE includes the Extreme ultraviolet SpectroPhotometer (ESP) channel, a very similar
instrument to SOHO/SEM with a band pass nearly equivalent to the SEM 26-34 nm channel, as
well as the Multiple EUV Grating Spectrographs (MEGS) channels which provide high resolution
spectra in the 6-106 nm spectral range (due to an instrument malfunction, wavelengths shorter than
4
35nm are not available for dates after 26 May 2014). Comparisons with EVE have shown that
since the beginning of the SDO mission, SOHO/SEM Version 3.1 irradiances in the 26-34 nm
band have been about 20 to 25% higher (Didkovsky et al., 2010b; Wieman, Judge, and Didkovsky,
2011) compared to both the ESP, and the MEGS spectra integrated from 26-34 nm. The
SOHO/SEM irradiances have been found to be higher by a similar amount in earlier comparisons
with other EUV instruments, including the SOHO Coronal Diagnostic Spectrometer (CDS) in
Thompson, McMullin, and Newmark (2002) and the Thermosphere Ionosphere Mesosphere
Energetics Dynamics/Solar EUV Experiment (TIMED/SEE) in Woods et al., (2005). For these
earlier comparisons (Figures 1.2 and 1.3 respectively), the 20% higher SOHO/SEM irradiance was
within the combined calibration uncertainties (estimated uncertainties are 10%, 20%, and 25% for
SOHO/SEM, SOHO/CDS, and TIMED/SEE respectively), however the same cannot be said for
the comparisons with SDO/EVE due to the relatively low estimated uncertainties of about 10% for
ESP (Didkovsky et al., 2012) and between 5% and 20% depending on wavelength for MEGS
(Hock et al., 2012). Understanding these differences between the SOHO/SEM and SDO/EVE
irradiances is one of the motivations for this study.
5
Figure 1.2 Comparison of 26-34nm irradiance measured by SOHO CDS (x), SOHO SEM(*), and
SOHO EIT(+) (after Thompson, McMullin, and Newmark, 2002).
Figure 1.3 Comparison of 26-34nm irradiance measured by TIMED SEE and SOHO SEM from
Woods et al., 2005.
6
Since the latest release (Version 3.1) of the SOHO/SEM measurements new calibration
measurements of the SEM instrument response function and measurements of the time-dependent
solar spectral distribution (i.e. reference spectrum) have been obtained, both of which are key
parameters for calculating calibrated irradiances from the SEM raw data. The instrument response
function of the SEM instrument aboard SOHO was measured pre-launch at the National Institute
of Standards and Technology (NIST) Synchrotron Ultraviolet Radiation Facility (SURF III). More
recently, however, a revised calibration procedure used for measuring the response function of the
SEM sounding rocket clone instrument at NIST revealed additional sensitivity of the SEM optical
design to wavelengths outside of the nominal first-order 26-34 nm band pass. The nominally
identical on-orbit SOHO/SEM most likely shares this out of band sensitivity and presented in this
work is a SOHO/SEM response function that has been revised accordingly based on the SEM
clone measurements (note, the SEM sounding rocket clone instrument will from here be referred to
as the “SEM clone” and “SOHO/SEM” or “SEM” will refer to the SEM instrument on orbit aboard
SOHO).
For much of the SOHO mission, measured solar spectral distributions in the EUV and soft
x-ray range have not been available. As an alternative, the SOLERS22 spectrum (Woods et al.,
1998), a single modeled solar spectrum has been used to date in SOHO/SEM data processing
(including Version 3.1). The SOLERS22 spectrum was adopted for processing SOHO/SEM data
because it is widely known and available (McMullin et al., 2002) and because it represents a solar
spectrum midway between solar minimum and solar maximum conditions and is therefore reliable
over a broad range of activity levels (Judge, et. al. 2002). However, knowing the spectral
distribution for deriving irradiance from broadband measurements is a long-standing problem
7
(Wende, 1972; Acton, Weston, and Bruner, 1999) and it has been suggested that differences in the
SOHO/SEM irradiances compared to those of other EUV instruments may be related to choice of
reference spectrum (Thompson, McMullin, and Newmark, 2002; Woods et al., 2005; Didkovsky et
al., 2010b; Wieman, Judge, and Didkovsky 2011).
This assertion was tested by Wieman, Judge, and Didkovsky (2011), by recalculating the
SOHO/SEM dataset for the period of overlap with SDO using directly measured reference spectra
from MEGS (i.e. measured concurrently with the corresponding SEM observations) in place of
SOLERS22. That study showed however that while the choice of reference spectrum does indeed
have an effect on the calculated irradiance values, the differences between SEM and EVE (and
earlier EUV instruments) cannot be explained based on reference spectrum alone. One of the
objectives of this dissertation is to resolve the differences observed between SEM and SDO/EVE by
reprocessing the SOHO/SEM data using both the more recently available SDO/EVE/MEGS
reference spectra and the broader response function derived from the latest NIST calibration of
the SEM clone. This approach for resolving the differences is validated through a series of
comparisons with both spectrally resolved and broadband measurements from SDO/EVE.
A second objective is to apply the new response function and the approach of using time –
dependent reference spectra to recalculate the entire SOHO/SEM data set back to 1996. For
reprocessing the earlier (i.e. prior to SDO) SOHO/SEM data a time-dependent reference spectrum
needs to be established and the SOHO/SEM time series calculated using the EVE reference spectra
serves as a standard against which to evaluate such alternative spectra. Both empirical (e.g., FISM:
Chamberlin, Woods, and Eparvier, 2007; or Solar2000/SIP: Tobiska, Bouwer, and Bowman, 2006)
8
and physics-based (e.g., Del Zanna et al., 2001; Thompson, McMullin, and Newmark, 2002)
modeled spectra are compared with the MEGS spectra based on how they affect the calculation of
SEM irradiances for the period of overlap with SDO. A change in reference spectrum also requires
the reprocessing of the SEM sounding rocket calibration measurements which will affect the SEM
long-term degradation model
1
. Thus, a reevaluation of the SEM long-term degradation is intrinsic
to this recalculation of the long-term dataset and can affect estimates of the differences in
minimum EUV irradiances during the minimum of Solar Cycles 22/23 compared to the minimum
of Solar Cycles 23/24.
A third objective of this work is to determine a procedure for reproducing SEM
Version 3.1 irradiance values which have already been adopted for space weather operations from
newer EUV instrumentation that is likely to remain in service longer than SEM. The SEM data are
the source of the S10 solar EUV irradiance proxy (Tobiska, Bouwer, and Bowman, 2006) used in
the Jacchia-Bowman 2008 Thermospheric Density model (JB2008) used by the US Air Force for
satellite operations (Bowman and Tobiska, 2008). However, it is uncertain how much longer
SOHO will be in operation and a different source for the S10 index will be needed once its mission
is over. Because the JB2008 model has been parameterized to use the SEM Version 3.1 26-34 nm
measurements as input, changing the source of this data could compromise the output of the
model, even if the new source is of greater accuracy. The EUV SpectroPhotometer (ESP) channel
1
Intuitively it might seem that a change in reference spectrum would have the equivalent
effect on both the sounding rocket and on-orbit measurements and thus not change degradation
estimates; however, the sounding rocket measurements include a wavelength dependent correction
for atmospheric absorption and incorporate irradiances from the RGIC, which has a different
spectral response function, and are thus affected differently by the change in reference spectrum.
Therefore, the distribution of sounding rocket measurements and the degradation model based
upon them are affected by the change in reference spectrum.
9
of the SDO/EVE instrument is very similar in design to SEM and efforts are underway to establish
S10.7 index values from the ESP measurements (Bouwer, Jones, and Tobiska, 2013). As an
extension of the aforementioned work to resolve differences between the SOHO/SEM and
SDO/EVE/ESP irradiance measurements (i.e., the first dissertation objective), a procedure for
reliably reproducing SEM Version 3.1 irradiance values from SDO/EVE/ESP measurements has
been developed and is presented here.
10
2. The Solar Extreme Ultraviolet and Soft X-ray Spectral Ranges
2.1 Solar Spectrum
The solar photon spectrum from the soft x-ray through infrared spectral range is shown in
Figure 2.1 (after Woods et al., 2009) in spectral irradiance units of W m
-2
nm
-1
. The SEM measures
irradiance integrated over two spectral bands within the EUV and SXR spectral ranges; one from
about 0.1-50 nm and one from about 26-34 nm. Photon irradiances in this range are several orders
of magnitude below peak emissions in the visible range. In contrast to the visible emissions which
are thermal in nature (approximately matching the emission spectrum of a blackbody at ~5800K)
and emanate from the solar surface, the EUV and SXR spectra are characterized by numerous
isolated emission lines and originate higher in the solar atmosphere. The emission lines that make
up much of the EUV and SXR spectral ranges are characteristic of elements in very high ionization
states that are sensibly only present at temperatures much hotter (Klimchuk, 2006) than the ~5800
K blackbody temperature suggested by the visible spectrum. Observations of these emissions
suggest a solar atmosphere temperature profile as shown in Figure 2.2 from Yang et al., 2009. The
layers of the solar atmosphere, delineated by this temperature profile, include the photosphere (the
visible “surface” of the sun), the chromosphere, the transition region, and the corona. The peak
emission temperatures of several important EUV and SXR emission lines, including the 30.4 nm
(304Å) line of singly ionized He and the 28.4 nm (284Å) Fe-XV line central to the SEM
measurements, are also shown in Figure 2.2
11
Figure 2.1 Solar photon spectrum for the soft x-ray through infrared region. The SOHO/SEM
broadband measurements include the 0.1-50 nm band (SEM 0th order; shown in pink) and the
26-34 nm band (SEM 1st order; shown in orange).
Figure 2.2 Temperature profile of the solar atmosphere (From Yang et al., 2009). The peak
emission temperatures of several ion emission lines are shown.
12
2.2 Solar Variability
Irradiance in the EUV and SXR portions of the spectrum is highly variable with time
compared to the visible portion (Woods et al., 2006). The variability observed in EUV and SXR
irradiance occurs on many timescales tied to the changing structure of the magnetic field which
originates within but extends beyond the surface of the Sun. Several sources of variability and the
timescales over which they can be observed are described below.
Solar Cycle Variability (timescale: ~ 1 decade) – Solar magnetic fields are generated by
the “solar dynamo” (e.g. Tobias, 2002) – the motion of highly ionized plasma within the Sun’s
convection zone. Although many details of the physical mechanism governing the evolution of this
field remain unclear (Parker, 2009), its cyclic nature has been evident for centuries based on
observations of the number of sunspots, or dark regions that form on the visible photosphere. The
number of sunspots visible on the surface varies with a period of about 11 years from 0 to as high
as ~200 (Tascione, 2010). This periodic behavior is referred to as the “sunspot cycle” (or more
generally as the “solar cycle” since many aspects of the Sun’s behavior including irradiance, polar
magnetic fields, and particle emissions, have also been found to follow the same cycle). Sunspots
are dark regions that form on the sun where high magnetic field strengths (associated with field
lines that pierce the photosphere in the form of loops) inhibit convection, locally reducing
temperature and thus emission intensity. Around the time that sunspot numbers reach their peak,
referred to as “solar maximum,” magnetic fields at the Sun’s poles weaken and reverse polarity
following the Hale cycle or the 22-year solar magnetic cycle (i.e. magnetic fields at a given pole
will go from positive to negative and back to positive over the course of two sunspot cycles). Solar
13
minima, times at which sunspot numbers are at their lowest, demarcate the beginning and end of a
sunspot cycle which have been assigned numbers dating back to the 1700’s. The SEM began
operating at the end of Solar Cycle 22/beginning of Solar Cycle 23.
Solar EUV and SXR irradiances are strongly correlated with sunspot number and thus
show a similar 11 year periodicity. Minima in the irradiance levels of both SEM bands around
1996 and 2008 correspond to the minima of Solar Cycles 22/23 and 23/24. Sunspots are not
observed at all locations on the solar disk, but are confined to mid-latitudes in the northern and
southern hemispheres. At a given time there is typically some north-south asymmetry in terms of
number of sunspots and strength of polar fields, and peak sunspot numbers and polar field
reversals do not typically occur in both hemispheres simultaneously, so a given solar maximum
will often include two local maxima as is evident for Solar Cycle 23 in the SEM data between the
years of ~2000 and 2002.
Solar Rotation Variability (timescale: ~ 1 month) – The distribution of emitting plasma
in the Sun’s upper atmosphere is far less uniform than in the photosphere. The plasma-β parameter,
βplasma, the ratio of gas pressure within a plasma relative to the magnetic pressure is given by:
𝛽 𝑝𝑙𝑎𝑠𝑚𝑎 =
𝑛 ∙ 𝑘 ∙ 𝑇 𝐵 2
2𝜇 0
⁄
where n is ion number density, k is Boltzman’s constant, T is the plasma temperature, B is
magnetic field strength, and μ0 is the magnetic permeability of free space. In regions where βplasma
has a low value the dynamics and density distribution are governed by the magnetic field present
rather than by gas pressure. Over a magnetically active region, βplasma is about equal to 1 in the
photosphere and upper corona, but drops to a value significantly less than 1 in the transition region
14
and lower corona (Gary, 2001). As a result emergent magnetic field lines, evident in the visible
photosphere by the aforementioned dark sunspots, form bright regions of confined plasma (e.g.,
coronal loops or magnetic flux loops) in the corona. Thus the Sun has a much different appearance
for example, when viewed at photospheric (visible) wavelengths compared to coronal (EUV/SXR)
wavelengths. These differences are illustrated in Figure 2.3 which includes filtered images from
the Atmospheric Imaging Assembly (AIA) aboard SDO (Boerner et al., 2010) representing the
photosphere, chromosphere, transition region and corona.
15
Figure 2.3 Concurrent images of the Sun from the Atmospheric Imaging Assembly (AIA) aboard
SDO. Each image is filtered to include wavelengths originating from a different layer of the solar
atmosphere.
Active regions can remain static considerably longer than the ~27 day rotational period of
the Sun (Hurlburt and DeRosa, 2008) and often revolve in and out of our view from Earth several
times before dissipating. Variation in the number, brightness and size of viewable active regions
photosphere
chromosphere
transition region corona
active regions
active regions
active regions
active regions
16
with solar rotation produces a strongly periodic ~27 day change in irradiance. In the photosphere,
active regions appear as relatively minor blemishes, so solar rotation produces little change in
visible irradiance. In the corona however a significant portion of the emitting plasma is
concentrated in active regions, thus EUV and SXR irradiance can vary significantly with rotation.
The 27-day periodicity is clearly evident in the SEM data. In Figure 2.4 SEM 26-34 nm data are
shown for an interval of notably strong and consistent rotational variability starting around Julian
Day 2456100.
Figure 2.4 SOHO SEM 26-34 nm irradiance time series for the rise of Solar Cycle 24. Strong solar
rotation variability is evident particularly starting on Julian Day 2456100.
Flare variability (timescale: seconds to hours) – EUV and SXR irradiances are also
highly variable on short timescales during solar flares. Flares are isolated events, so unlike solar
cycle and solar rotation related changes, flare variability is not periodic. They occur when energy
17
stored in the aforementioned magnetic flux loops is rapidly released in the forms of
electromagnetic radiation across a broad spectral range (X-ray to radio), particle acceleration and
often mass ejection. The fractional increase in EUV and SXR irradiance during a flare can be quite
significant. For example, one of the most common classifications for flare magnitude, the National
Oceanic and Atmospheric Administration’s X-ray flare class
(http://www.swpc.noaa.gov/index.html) is based on X-ray irradiance in the 1.0 to 8.0 Å spectral
band. The timing of flare-related irradiance increase for a given flare is not uniform across all EUV
and SXR wavelengths. Transition region emissions tend to rise rapidly, peak early, and decay
quickly in what is called the impulsive phase of the flare. Soft x-ray coronal emissions generally
start earlier but rise slowly, peak later, and decrease slowly in what is called the gradual phase.
Figure 2.5 shows the response of the SEM channels during an X6.9 flare on August 9, 2011. The
26-34 nm channels is dominated by the He-II transition region line and thus it has a nominally
impulsive phase time profile. The 0.1-50 nm channel includes contributions from both transition
region emissions as well as coronal SXR emissions and thus shows the early slow rise and gradual
decay characteristic of gradual phase flaring superimposed on a fast rising and falling impulsive
phase peak. Saturation of the 0.1-50 nm detector is evident from the plateau at maximum
irradiance levels.
18
Figure 2.5 – EUV and SXR irradiances observed during an X6.9 class flare. The 26-34 nm channel
exhibits impulsive phase variability while the 0.1-50 nm signal (which saturates at ~.02 W/m
2
)
includes both gradual phase and impulsive phase components.
The level of variation associated with each of the above modes is highly wavelength
dependent within the EUV and SXR ranges. The expected fractional change associated with each
of these modes is plotted versus wavelength in Figure 2.6 from Chamberlin et al., (2008).
19
Figure 2.6 Solar cycle, solar rotation and flare related (impulsive and gradual phase) variability of
EUV and SXR emissions versus wavelength.
2.3 History of EUV measurements
Solar EUV and SXR emissions are absorbed in the upper atmosphere so irradiance
measurements in this spectral range require space-based instrumentation, such measurements
however have not been routine throughout the space age. Numerous short-term sounding rocket
based measurements have intermittently provided “snapshots” of both spectral and broadband
integrated irradiance (Hall et al., 1969; Carlson et al., 1984, Ogawa et al., 1986, Woods and
20
Rottman, 1990), but only limited information about variability on different timescales. Intervals of
continuous EUV and SXR data were provided by the Orbiting Solar Observatory (OSO – Hall and
Hinteregger, 1970) and Atmospheric Explorer (AE – Hinteregger et al., 1981) series of satellite
missions during the 1960s and 1970s. Spectral data from the AE-E satellite contributed to the
SOLERS22 reference spectrum (Woods et al., 1998) used to date for processing the SOHO/SEM
data (the dependence of SOHO/SEM irradiance values on reference spectrum is a central topic of
this work).
The period after the AE- mission ended in 1980 through the start of the SOHO mission at
the end of 1995 is referred to as the “EUV hole” (Woods et al., 2005) because only a few isolated
EUV measurements (e.g., Schmidtke et al., 1992) were made during this time. This dry spell has
ended gradually over the last two decades with EUV and SXR instruments aboard multiple
satellites including, SOHO (Harrison, Sawyer, and Carter, 1995; Hovestadt et al, 1995),
SNOE(Bailey et al., 2000), TIMED (Woods et al., 2005), SORCE (Woods and Rottman, 2005),
ISS (Nikutowski et al., 2011), PROBA2 (Dominique et al., 2013), GOES (Evans et al., 2010). The
spectral and temporal coverage of EUV irradiances provided by the SDO/EVE (Woods et al.,
2012, Hock et al., 2012, Didkovsky et al., 2012) instrument suite operating since 2010 is
unprecedented.
2.4 Solar EUV proxies
Because EUV irradiance measurements have only in the last couple of decades become
available with some consistency, many scientific investigations and space weather operations that
21
depend on such data have instead adopted more readily available solar activity proxies. A few of
the commonly used proxies are discussed briefly below:
Sunspot number, R –The aforementioned sunspot number provides some quantification of
solar activity, although because accounting for the presence of a sunspot does not carry with it any
information about the size of the associated active region or the brightness of emissions in the
overlying atmospheric layers, the relationship to EUV and SXR irradiance is indirect. Sunspot
number will often fall to zero during solar minimum while irradiance levels continue to vary.
Although sunspot number has in the past been adopted in investigations of the effects of solar
variability on the Earth’s atmosphere (e.g. Allen, 1948), it is no longer commonly used for such
applications.
10.7 cm radio flux, F10.7 – One of the most widely used proxies, the F10.7 index, is the
solar radio flux at 10.7 cm wavelength. Microwave flux at this wavelength is not absorbed in the
Earth’s atmosphere and thus can be readily measured from the ground. Emissions at this
wavelength originate from locations of intense magnetic fields in the chromosphere and overlying
atmospheric regions like EUV and SXR emissions (Foukal, 1998) but also include a weaker, more
diffuse component distributed over the solar disk (Tapping, 1987). The 10.7 cm flux correlates
well with various wavelengths within the EUV and SXR ranges as well as with upper atmospheric
conditions known to depend on EUV and SXR irradiance (Tobiska, 1988). It is commonly used in
upper atmosphere modeling (Hedin et. al., 1977; Jacchia, 1970; Bowman, 2008). Under conditions
of low solar activity the F10.7 index does not correlate well with EUV and SXR emissions – with
decreasing activity levels it tends to level off near a minimum value while the shorter wavelength
22
photon emissions continue to decrease (e.g., Tapping and DeTracey, 1990; Woods et al., 2000;
Tobiska, 1988). The use of the F10.7 index in place of direct EUV measurements has been cited as
a source of uncertainty in upper atmosphere modeling efforts (Lean et al., 2009).
Magnesium II core to wing ratio, Mg-II – The Mg II (Viereck, 2004b) core to wing is
based on the correlation of changes in the spectral shape of the Mg-II absorption feature (~280 nm)
with EUV and SXR irradiance. Mg-II cwr must be measured from space but because it based on
relative changes of a spectral feature it is not sensitive to instrument degradation the way absolute
irradiance measurements are. Long term records of Mg-II incorporate measurements from multiple
spacecraft using spectroscopic instruments with different spectral resolutions. This use of different
instrumentation has been cited as a source of inconsistency in the long-term Mg-II record (Lean et
al., 2009).
23
3. EUV and SXR irradiance and the geospace environment
Solar EUV irradiance data are essential to efforts to understand, model, and predict space
weather. EUV and SXR photons from the Sun are absorbed, mostly through photoionization, in
Earth’s upper atmosphere (altitudes above ~ 90 km), thus producing the ionosphere and initiating
processes that heat neutral species in the thermosphere (Stolarski, et al., 1975). As a result, space
weather conditions of scientific and practical significance such as temperature, neutral and electron
density profiles in the upper atmosphere directly depend on solar irradiance in this spectral range
(Roble et al. 1995, Tascione 2010, Tsurutani, et. al. 2005, Solomon 2010, Lean, et. al. 2011b). This
chapter provides an overview of this connection between solar EUV irradiance and the geospace
environment.
3.1 EUV irradiance and the study of the ionosphere
Solar photons in the EUV and SXR spectral ranges have sufficient energy to ionize the
major neutral species (Tobiska, 1988) in the Earth’s thermosphere (altitudes above ~ 90 km). The
ionosphere, created as a result of this process, consists of electrons and ions that are within the
thermosphere (and extend above it) but for a given altitude, are present with much lower number
densities than the neutral species. Number density profiles for both neutral and ionized species in
the upper atmosphere are shown in Figure 3.1 as an example of day time distributions during low
solar activity. Electron densities do not vary uniformly with altitude, but instead have a partly
stratified profile with distinct layers as shown by the corresponding curve (solid line labeled e
-
) on
the left side of the plot in Figure 3.1. There are in general four layers (though not all may be
24
present at a given time) designated the D, E, F1, And F2 layers which form through the ionization
of the several neutral species present in the thermosphere by ionizing solar photons of various
wavelengths. Table 3.1 (after Chamberlin, 1978) summarizes the ionospheric processes involved
in the formation of each of these layers.
Figure 3.1 Densities of neutral and ionized species and electrons in the Earth’s upper atmosphere.
The composition is for quiet Sun daytime conditions (with the exception of He profile which is
based on a night time measurement). Density values below 250 km are from sounding rocket borne
mass spectrometer measurements taken over White Sands Missile Range, NM. (from Johnson, C.
Y.: 1969, Ion and neutral composition of the ionosphere, Ann. IQSY, 5, reprinted in Kelly, 1989)
25
Table 3.1 Ionosphere regions and the processes by which they are formed (summarized from
Chamberlain and Hunten, 1987)
Region Peak electron
density*, Nm
(cm
-3
)
Nominal height
of peak (km)
Primary ionization processes
D ~10
4
90
ionization of NO by Lyman-α and SXR photons
(intermediate EUV wavelengths are absorbed in
higher layers)
E ~10
5
110
Ionization of O2 by EUV (between ~40 nm and
100 nm) and of O, O2, and N2 by SXR photons
F1 ~3∙10
5
200
Ionization of O by EUV (between ~10 nm and
100
F2 ~10
6
300
Same as F1, but remains present at night due to
lower recombination rates
* Peak values occur at noon; at night the D and F1 layers virtually disappear and electron
densities in the E and F2 layers decrease by a factor of ~10.
This general layer structure of the ionosphere, and to some degree its variability, can be
explained based on a simple model atmosphere introduced by Chapman (1931). This model
assumes monochromatic ionizing photons from the Sun, a single species atmosphere, and planar
26
earth/isobaric surfaces, and invokes the more general assumption that the change in flux of photons
as they pass through an absorbing medium is proportional to the flux, the mass absorption
coefficient, and the mass density of the medium (the latter two parameters are expressed here in
terms of absorption cross-section, and number density, respectively for consistency with other
portions of the text) according to:
𝑑 Φ
𝐸𝑈𝑉 𝑑 ℎ
= Φ
𝐸𝑈𝑉
(ℎ) ∙ 𝑛 (ℎ) ∙ 𝜎 𝑎𝑏𝑠 ∙
1
𝑐𝑜𝑠𝜒 (3.1)
where Φ is photon flux, n is neutral density, h is altitude, σabs is the total photoabsorption
cross-section and χ is the solar zenith angle (when the Sun is not at zenith, dh/cos χ is the optical
path length for photons traversing the altitude increment dh). The wavelength dependence of σabs
is ignored based on the assumption of monochromatic ionizing photons.
The atmospheric neutral density is assumed to decrease exponentially with height and can
be expressed as:
𝑛 (ℎ) = 𝑛 0
∙ 𝑒 −
ℎ
𝐻 (3.2)
where n0 is the density at the base of the atmosphere, and H is the atmospheric scale height
(assumed constant in the Chapman model
2
), which represents the increment in altitude over which
the neutral density decreases by a factor of e. Combining 3.1 and 3.2 and solving the differential
equation for the attenuated photon flux as a function of altitude (and zenith angle, χ) gives:
2
This assumption is not entirely valid since intrinsic to it is the further assumption that
atmospheric temperature does not vary with height. This simplification significantly reduces the
accuracy with which the Chapman model represents the real atmosphere
27
Φ
𝐸𝑈𝑉
(ℎ, 𝜒 ) = Φ
𝐸𝑈𝑉 ∞
∙ 𝑒 [−𝜎 𝑎𝑏𝑠 ∙𝑛 0
∙𝐻 ∙𝑠𝑒𝑐𝜒 ∙𝑒 (−ℎ/𝐻 )
]
(3.3)
where Φ
𝐸𝑈𝑉 ∞
is photon flux above the atmosphere, i.e. the quantity measured by SEM. If
the number of ions produced per photon absorbed is equal to β, then the volumetric ion production
rate, I (e.g. with units: ions per cm
3
per second) at a height h due to solar photons incident at a
zenith angle, χ is given by:
𝐼 (ℎ, 𝜒 ) = 𝛽 ∙ 𝑐𝑜𝑠𝜒 𝑑 Φ
𝐸𝑈𝑉 𝑑 ℎ
(3.4)
Taking the derivative of (3.3) and substituting into (3.4) gives:
𝐼 (ℎ, 𝜒 ) = 𝛽 ∙ 𝜎 𝑎𝑏𝑠 ∙ Φ
𝐸𝑈𝑉 ∞
∙ 𝑛 0
∙ 𝑒 [−
ℎ
𝐻 − 𝜎 𝑎𝑏𝑠 ∙𝑛 0
∙𝐻 ∙𝑠𝑒𝑐𝜒 ∙𝑒 (−ℎ/𝐻 )
]
(3.5)
The rate of change of electron density in the ionosphere is equal to this ion production rate
minus the rate of ion loss through recombination. The rate of recombination can be treated as
proportional to the square of the electron density (Chapman,1931; Mitra, 1959) and the rate of
change of electron density, Ne, can be expressed as:
𝑑 𝑁 𝑒 𝑑𝑡
= 𝐼 − 𝛼 𝑒𝑓𝑓 ∙ 𝑁 𝑒 2
(3.6)
where αeff is the effective recombination coefficient which, when multiplied by Ne
2
,
provides the recombination rate. Electron density profiles can thus be determined for equilibrium
28
conditions (Tascione 2010) where the ion production and loss rates are equal by combining (3.5)
and (3.6):
𝑁 𝑒 (ℎ, 𝜒 ) =
√
𝛽 ∙𝜎 𝑎𝑏𝑠 ∙Φ
𝐸𝑈𝑉 ∞
∙𝑛 0
∙𝑒 [−
ℎ
𝐻 − 𝜎 𝑎𝑏𝑠 ∙𝑛 0
∙𝐻 ∙𝑠𝑒𝑐𝜒 ∙𝑒 (−ℎ/𝐻 )
]
𝛼 𝑒𝑓𝑓 (3.7)
Equation (3.7) shows the fundamental dependence of the ionospheric electron density
profile on solar EUV irradiance according to the Chapman model. Although the Chapman model
includes many simplifying assumptions that limit the accuracy with which it represents the real
ionosphere (e.g., the model does not account for magnetospheric effects or particle collisions, and
density scale heights and effective recombination coefficients are not actually constant with
altitude) the general structure of the ionosphere including the electron density profiles associated
with the aforementioned D, E, F1, and F2 ionization regions can be reasonably approximated
(Tascione, 2010) by summing appropriately defined Chapman layers. Furthermore, EUV
irradiance or a correlated proxy (e.g., the F10.7 radio index) remains a primary input for most of
the current, more comprehensive models of the ionosphere and upper atmosphere (Tobiska, 2011,
Lean, et. al. 2011b).
Various properties of the ionosphere can be determined from ionosonde measurements and
from measured delays in radio signals from Global Positioning System (GPS) satellites. With the
ionosonde technique, originally developed in the 1920’s by Breit and Tuve (1926), the reflections
of pulsed radio signals transmitted upward at various frequencies are measured. Radio signals are
reflected off of ionospheric layers up to a certain critical frequency, beyond which they penetrate
the layer without being reflected. This critical frequency is related to the plasma frequency (a
29
function of electron density) of the layer, thus, the frequency of the transmitted signals and their
transit time provide information about the electron density and height of the various ionospheric
layers. Another quantity that is often used to characterize ionospheric conditions is Total Electron
Content (TEC). TEC is a measure of the total number of electrons in a column of a specific cross-
sectional area from the ground to the top of the ionosphere, typically expressed in units of TECU,
where 1 TECU corresponds to a total of 10
12
electrons in a 1 cm
2
column (Mannucci, et al. 1998).
Techniques have been developed (Lanyi and Roth, 1988; Mannucci, et al., 1998, Hernández-
Pajares, et al., 1999, 2009) for creating temporally and spatially resolved TEC maps based on data
from geodetic Global Position System (GPS) receivers which provide information on ionosphere-
induced delays in radio signals from GPS satellites.
The response of the ionosphere (i.e., as determined using ionosonde and GPS techniques)
to variability in solar EUV and soft x-ray irradiance has been studied extensively. EUV irradiance
has been shown to be a dominant driver of ionospheric conditions including TEC and peak
electron density over a large range of time scales from minutes [i.e. solar flare variability as shown
in Tsurutani, et al. (2005), and Qian et. al, (2011)] to weeks [i.e. solar rotation variability as shown
in Liang (2008) , and Afraimovich et al. (2008)] to decades [i.e. solar cycle variability as shown in
Liu et al. (2006), and Lean et al. (2011a, 2011b)]. The SOHO/SEM EUV irradiance data, which
are available with high time-cadence, and radiometric accuracy continuously over an extended
period (>19 years at the time of this writing), are often used in these studies (Tsurutani et al., 2005;
Afraimovich et al., 2008; Liu et al., 2006; and Lean et al., 2011b). While ionospheric variability is
complex (Kelly, 1989; Anderson and Fuller-Rowell, 1999, Lean et al., 2011b) and depends on
magnetospheric conditions, interactions with the varying neutral thermosphere (also driven by
30
EUV irradiance as discussed below) in which it is embedded, and other sources of ionization (e.g.
auroral ionization caused by charged particle preciptitation from the magnetosphere), EUV and
soft x-ray irradiance measurements remain of primary importance to ionospheric studies.
Through its influence on the ionosphere, solar EUV and SXR irradiance has an effect on
GPS navigation and radio communications. Changes in TEC affect the delay of radio signals
transmitted from GPS satellites. The aforementioned geodetic GPS receiver stations provide a
measure of this delay and thus can account for such changes in dual-frequency GPS systems, but
for single-frequency GPS users, the changing delay remains a source of inaccuracy (Jensen and
Mitchell, 2011). Peak ionosphere electron densities are sufficient to reflect radio signals of up to
about 30 MHz and can thus provide a means of extending the range of ground-based radio
communications at such frequencies by using the ionosphere as a wave guide where there is no
direct line of sight between the transmitter and receiver. Such transmissions relying on reflection
off of the ionosphere F-layer can be absorbed if D-layer ionization increases rapidly due to a solar
flare related increase in SXR irradiance – an effect known as shortwave fadeout (Tascione, 2010).
Thus EUV and SXR data are of practical value for operations relying on such radio signal
propagations.
3.2 The role of EUV irradiance in the study of the thermosphere
The Chapman model atmosphere, with neutral density profile described by equation 3.2 –
an expression of barometric law for an isothermal atmosphere (Chamberlain, 1987), does not
accurately describe the density profile of the thermosphere, since the temperature of the
thermosphere is not uniform with height. The temperature of the mesopause at the base of the
31
thermosphere remains relatively stable at about 175K (Chamberlain, 1987) and the temperature of
the exosphere, which thermospheric temperatures approach asymptotically at higher altitudes, can
range from about 500K to about 2000K (Tascione 2010) depending largely on the level of EUV
irradiance. For the upper thermosphere, above about 100km there is little mixing of constituents
and each has an independent number density profile governed by diffusive equilibrium, which can
be determined by integrating the diffusion equation (e.g. after Jacchia, 1971; Chamberlain, 1987;
Bates, 1959):
𝑑𝑛
𝑛 (ℎ)
= −
𝑑𝑇
𝑇 (ℎ)
−
𝑀 ∙𝑔 (ℎ)
𝑘𝑇 (ℎ)
𝑑𝑧 (3.8)
Where n is the number density of an individual species, T is temperature, M is the species
molecular/atomic mass, g is the acceleration of gravity, k is Boltzmann’s constant, and h is
altitude. For this integration, a thermosphere temperature profile T(h), must be determined.
Because solar EUV is the primary source of thermospheric heating, EUV irradiance data
are essential to determining such temperature profiles. For example, Banks and Kockarts (1973)
and Chamberlain (1978) describe analytic approaches for establishing thermosphere temperatures
based on the energy conservation equation, in which they balance heat input to the thermosphere
(mainly from solar EUV) with radiative losses and conduction to the cooler mesopause and, as a
boundary condition, assume a static temperature near the base of the thermosphere. The EUV
heating term, 𝑄 𝐸𝑈𝑉
, has a similar form in both the Banks and Kockarts and Chamberlain energy
conservation equations. This term is shown here for an individual neutral gas constituent although
32
for solution of the energy conservation equation, heating values would be summed over all of the
major (i.e. O, O2, N2) and minor (e.g., CO2, NO, He, H) thermospheric species:
𝑄 𝐸𝑈𝑉
(ℎ) = ∑ 𝑛 (ℎ) ∙ 𝜎 𝑖 ∙ Φ
𝐸𝑈𝑉𝑖
(ℎ) ∙ 𝜀 (ℎ)
𝑖 (3.9)
where, n is the species number density, σi is the total absorption cross section. The
subscript i denotes a spectral interval within the SXR to EUV range over which the value of σ is
effectively constant and Φ EUVi is the photon flux within that spectral range. The parameter ε is the
thermospheric heating efficiency coefficient (Stolarski et al., 1975, Richards, 2012), which
describes the fraction of the energy which the EUV photons deposit upon absorption that
ultimately goes into heating the neutral species.
Figure 3.2 (from Viereck, 2004a) provides an example of the energy deposition rate of
EUV and SXR photons as a function of wavelength and altitude in the thermosphere; rates at a
given time and location depend on the thermospheric composition and variability of the solar
SXR/EUV spectral irradiance. This energy is the primary heat source in the thermosphere,
although not all of it goes into heating the neutral species (hence the use of the heating efficiency
parameter, ε, in equation 3.9). Most of the photons are absorbed by photoionization and thus their
energy goes initially to the resulting ion and photoelectron. Heating of the thermospheric neutrals
(and ions) subsequently occurs as a result of numerous exothermic reactions that occur among the
photoions and the energetic electrons (Richards, 2012). Figure 3.3 from Stolarski, 1975 diagrams
some of the paths by which the deposited photon energy either ends up heating the neutral species,
or is radiatively lost.
33
Figure 3.2 – Energy Deposition rates in the thermosphere due to absorption of solar EUV and SXR
photons from Viereck (2004a) as a function of wavelength. Photons at 30.4 nm and 28.4 nm due to
solar He-II and Fe XV emissions respectively deposit energy at a notably high rate. Emissions
from these two spectral lines are measured by the SEM 1st order (26-34 nm) channel.
Figure 3.3 – Flow diagram showing the path by which EUV photon energy is either transferred to
the neutral gas or radiatively lost (from Stolarski, 1975)
34
Ultimately, about half of the EUV and SXR energy goes into heating the thermosphere
(Richards, 2012), and the variability of irradiance in this spectral range results in the
aforementioned large range of exospheric temperatures. For example, Figure 3.4 from Tobiska,
1988 shows temperature profiles for the thermosphere for lower levels of EUV irradiance
corresponding to solar minimum and for higher levels corresponding to solar maximum. Tobiska’s
energy conservation-based model is similar to that described by Banks and Kockarts, 1973 and is
compared in the figure to profiles from MSIS 83, a prominent empirical model (Hedin, 1997).
From panels a and b in Figure 3.4 it is evident that for this example the difference in EUV
irradiance from solar minimum to solar maximum results in a factor of ~2 increase in temperature
in the upper thermosphere.
Figure 3.4 Thermosphere temperature profiles associated with a) low EUV irradiance under solar
minimum conditions and b) with high solar irradiance under solar maximum conditions. Results
from Tobiska’s energy conservation-based model (thin curves) are compared with the MSIS 83
empirical model (thick curves). Plots are from Tobiska, 1998, with labels added to identify the
profiles.
35
Associated with elevated temperatures in the thermosphere under increased EUV irradiance
is an expansion of the atmosphere which results in higher neutral densities at a given altitude. This
effect, a shift in the diffusive equilibrium-governed density profile, can be quantified for example
by integrating the diffusive equilibrium equation (equation 3.8) using two different temperature
profiles (e.g. a solar minimum profile from Figure 3.4a and a solar maximum profile from Figure
3.4b). Neutral density profiles determined in this manner for several species are compared in
Figure 3.5 where the density profiles in panel a) correspond to a solar minimum temperature
profile and the profiles in panel b) correspond to a solar maximum temperature profile. From this
figure it is evident that the difference in EUV irradiance between solar minimum and solar
maximum results in a factor of more than 10 increase in number density for the major neutral
species at altitudes near the top of the thermosphere. As shown in Figure 3.1 ion and electron
densities in the thermosphere are much (more than 1 to several orders of magnitude) lower than
neutral densities. Even allowing for increased photoionization during high solar activity only a
very small fraction of the thermosphere species are ionized, so the direct loss of neutrals through
photoionization (or gain through recombination) does not have a significant effect on the neutral
density profile compared to the changes related to neutral heating.
36
Figure 3.5 Thermosphere number densities corresponding to the temperature profiles of Figure 3.4
for a) solar minimum and b) solar maximum conditions (from Tobiska 1988).
The aforementioned MSIS empirical model (Hedin et al., 1977) is one of several (e.g., see
also Bates, 1959 and Jacchia, 1965) that approximate the theoretically derived temperature profiles
with an exponential function that describes the rise in temperature between a nominally fixed value
near the base of the thermosphere (an altitude of 120 km is typically used) and an exospheric
temperature value that varies empirically based on a measured index of solar activity. Because
continuous, direct EUV irradiance measurements were not available at the time these models were
developed, they have typically relied upon the F10.7 radio flux for this purpose. For example
Jacchia (1970) determines exospheric temperature, T∞ (in Kelvin), based on the F10.7 index
according to:
𝑇 ∞
= 379 + 3.24 ∙ 𝐹 10.7 + 1.3(𝐹 10.7 − 𝐹 10.7) (3.10)
37
where F10.7 is the daily index value, and F̅10.7 is the index averaged over 81 days (3 solar
rotations). The use of the F10.7 index in place of direct EUV measurements has been cited as a
source of uncertainty in thermospheric density models (Lean et al., 2009). Additionally, Bowman,
et al. (2008) report that the The Jacchia-Bowman 2008 (JB-2008) thermospheric density model,
which is based on the Jacchia diffusion equations but includes solar activity indices based on direct
SXR and EUV irradiance measurements (with those from SEM contributing heavily), achieves
significantly better accuracy than earlier models that rely on F10.7 as the sole indicator of EUV
irradiance. The JB-2008 model will be discussed further in section 4.1 on specific applications of
the SEM dataset.
Neutral densities below about 1000 km can impart sufficient drag on low earth orbit (LEO)
satellites to appreciably affect their trajectory, thus density data such as those provided by the
aforementioned models are important for tracking and predicting the orbits of satellites operating
within the thermosphere. For example, orbital perturbations due to changes in the thermospshere,
if unpredicted, may inhibit the ability of ground-based radar tracking stations to locate the LEO
satellites they are charged with tracking (Tascione, 2010). The dependence of thermospheric
density on solar variability as described above further highlights the importance of SXR and EUV
irradiance data for such operational efforts.
38
4. Current space weather applications for which SEM data are
specifically relevant
4.1 The S10 index and the Jacchia Bowman thermospheric density model
The Jacchia Bowman (JB2008) thermospheric density model (Bowman et al., 2008) adopts
a solar activity index, S10, based on measurements from the SEM 26-34 nm channel. The index
values are established by applying a linear scaling to the SEM photon flux measurements to
facilitate their use in a function developed initially to determine exospheric temperature based on
the F10.7 index (as in equation 3.10). The new exospheric temperature equation used in JB2008
also incorporates indices based on other data sets including the M10 chromospheric index based on
the Mg II core to wing ratio (Viereck et al., 2004b) and the Y10 index based on measurements
from the NOAA/GOES X-ray Spectrometer (XRS) instrument. Exospheric temperature, T∞, is
calculated as a weighted sum of these indices:
𝑇 ∞
= 392.4 + 3.227 𝐹 𝑆 + 0.298∆𝐹 10.7 + 2.259∆𝑆 10
+ 0.312 ∆𝑀 10
+ 0.718 ∆𝑌 10
(4.1)
where F ̅ s is the 81-day average of a corrected version of the F10.7 index – the correction is
a substitution of S10 values in place of F10.7 during periods of low solar activity where F10.7 is
known to reach a fixed minimum value even though EUV irradiance continues to vary. The delta
(Δ) in front of each of the indices indicates the difference between the daily and 81 day average
index values, e.g. ΔS10 = (S10 - S̅ 10). It is evident from 4.1 that S10 is one of the most heavily
39
weighted indices, second only to FS, the modified F10.7 index which, as noted above, depends on
S10 near solar minimum.
Monitoring of drag-related changes in satellite orbits either through radar tracking or on-
board accelerometers provides indirect yet reliable measurements (Storz et al., 2002; Emmert et
al., 2004, 2008) of thermospheric density which provide a basis for comparison and validation of
density models. The agreement between the JB2008 model and thermospsheric densities
determined in this manner is significantly better than what has been achieved by earlier models
based on F10.7 as the sole index of solar activity. Figure 4.1 from Bowman et al., 2008 shows a
comparison of JB2008 400 km density values over a range of solar activity levels (indicated by the
81-day mean F10.7 value labeled F10B in the figure) to those of several other thermospheric
density models including the High Accuracy Satellite Drag Model (HASDM – Storz et al., 2002).
The HASDM model determines density directly by continuously incorporating near real-time data
on drag related changes of satellite orbits, and thus provides a set of “observed” densities
considered to be the standard in these comparisons. Ratios of model/HASDM are significantly
closer to unity for JB2008 than for the F10.7-based Jacchia 70 and NRLMSIS (a later version of
the aforementioned MSIS) models over the range of activity levels compared.
For operational models such as JB2008, consistency of input data is often more important
than absolute accuracy (Tobiska, 2012). The primary aim of this dissertation is to improve the
accuracy of the SEM data, and part of that effort requires understanding the sources of differences
between SEM irradiance measurements and those from other EUV instrumentation. An additional
goal of this work, based on this understanding, is to establish a means of determining irradiance
40
values from newer EUV instrumentation that are in close agreement with those from SEM such
that they can be used in place of the SEM data for applications like the S10 index without loss of
consistency, after the SOHO mission ends.
Figure 4.1 – A comparison of the JB2008 (red diamonds and fit curve) thermospheric density
model (which uses the SEM based S10 index of EUV irradiance) with the F10.7 based NRLMSIS
(grey diamonds and fit curve) and Jacchia 70 (purple diamonds and fit curve). The models are
compared based on agreement with the HASDM (described in text) “observed” density values.
Plotted values are HASDM/model for 400km density values binned according to F ̅ 10 (labeled
F10B) solar activity level for the period from 2001-2007 (from Bowman et al. 2008). JB2008
shows better agreement with HASDM (ratios closer to unity) at most levels of activity, but most
notably at lower F ̅ 10 levels.
41
4.2 Understanding inter-minima EUV changes
Very low levels of solar activity persisted for an unusually long period of time during the
minimum of Solar Cycles 23/24 (occurring in 2008) compared to previous solar minima. However
there are significant discrepancies among the various indicators of solar activity over how much
lower EUV/SXR irradiance was during this recent minimum period. In particular, as described
above in section 3, both neutral and electron density in the upper atmosphere are strongly
dependent on EUV/SXR irradiance, however it is not clear that these properties had the same
response to the inter-minima change in EUV. Based on satellite drag measurements, globally
averaged thermospheric neutral densities (at a given altitude) were at record low levels during the
2008 minimum as described in Emmert et al. (2010) and Solomon et al. (2010, 2011) suggesting
that EUV levels were significantly lower than during previous minima. However, global mode
TEC levels were comparable to past minima (Lean et al., 2011a) suggesting little inter-minima
EUV change (“global mode” TEC refers to TEC values that are globally averaged without spatial
resolution as opposed to available higher order spherical harmonic TEC values which can, in the
case of sectoral harmonics for example, resolve differences in night side/day side ionosphere
TEC). In figure 4.2, plots showing the changes over Solar Cycle 23 of global mode TEC (from
Lean et. al, 2011a) and of global fixed altitude neutral density (from Emmert et al. 2010) illustrate
this difference in response. One possible explanation for this discrepancy is the long term increase
in atmospheric CO2 concentrations (Roble et al., 1989) – this increase, which tends to result in
higher temperatures in the lower atmosphere due to the capacity of CO2 to absorb infrared
radiation, has the opposite effect in the upper atmosphere where at lower number densities CO2
ceases to be an effective absorber but remains an effective emitter and increases the capacity of the
42
mesopause as a heat sink on the thermosphere. Thus, higher CO2 concentrations tend to decrease
the thermospheric temperature and fixed altitude density whereas according to coupled
thermospere-ionosphere models (Rishbeth et al., 1992), accompanying changes in ionosphere
electron densities are relatively small. While this increase in CO2 concentrations may partly
explain the apparent inconsistency in the response of the upper atmosphere, its effect on
thermospheric density at 400km is estimated based on satellite drag measurements (Emmert et al.,
2008) and supported by model predictions (Roble et al., 1989) to be a decrease of only about 2%-
5% per decade (Solomon et al., 2011). This rate of decrease does not account for the ~30%
decrease between minima reported by Solomon et al. (2011) and Emmert et al. (2010).
Figure 4.2 Comparison of changes during Solar Cycle 23 in a) thermospheric density at an altitude
of 400 km (after Emmert et al., 2010) and b) Global Mode Total Electron Content (from Lean et al.
2011a). For the density values in a) the top panel shows daily values and the bottom panel shows
81-day average values; the green dashed line in the bottom channel shows the long-term trend of
decreasing thermospheric density associated with increasing atmospheric CO2 concentrations
described in Emmert et al., (2008). For the TEC time series in b) daily values are shown.
43
This issue regarding the inter-minima change in EUV irradiance cannot be easily resolved
based on EUV proxies or direct EUV measurements. The only single-source EUV irradiance
measurements spanning all of Solar Cycle 23 are those from SEM. SEM shows irradiance in the
26-34 nm spectral range to be about 15% lower during the 2008 solar minimum compared to the
previous 1996 minimum (Didkovsky et al., 2010). Based on thermospheric response models
(Solomon et al.,2010, 2011) this level of EUV change is consistent with the ~30% difference in
thermospheric density determined between the two minima. Lean et al. 2011a asserts however that
such a large change in EUV is not consistent with the response of the ionosphere which showed
little difference in TEC between the two minima, and that “sensitivity drifts in the SEM
instrument” are a more likely explanation for the “spurious” difference between 1996 and 2008
solar minimum irradiance levels as measured by SEM. This assertion is supported by F10.7 index
values which are about the same for the two minima, however as noted above, the F10.7 is known
to “plateau” and vary non-linearly with EUV for periods of low solar activity (Bowman et al.
2008). The Mg-II index is lower for the 2008 minimum compared to previous minima, but only by
about 5% (Solomon et al., 2011) to 10% (Solomon and Qian, 2011) and thus does not provide
unequivocal support for either the SEM measurements or the assertions of Lean et al., 2011a.
Long-term calibration of the SEM measurements is maintained based on periodic sounding
rocket measurements (and an instrument degradation model based upon them) that provide a time-
dependent correction factor which, applied in data processing, maintains agreement between the
on-orbit and rocket-based measurements. The “sensitivity drifts” proposed in Lean et. al., 2011a
would suggest some shortcoming in this method for maintaining the SEM calibration. The
recalibration of the SEM dataset performed as part of this work includes the introduction of new
44
values for the instrument response function and time dependent solar reference spectra. The effect
of these new data processing parameters on the sounding rocket measurements and the
accompanying degradation model are investigated in this work to evaluate the possibility that the
inter-minima change in SEM irradiance may be due to inaccurately assessed long-term changes in
SEM sensitivity.
45
5. SOHO/CELIAS/SEM
SOHO's Solar EUV Monitor (SEM), designed and built at the University of Southern
California, is a highly stable transmission grating spectrometer that has been measuring the full
disk absolute solar flux in two bandpasses, 26-34 nm and 0.1-50 nm since shortly after its launch
in December of 1995. The SEM 26-34 nm bandpass is dominated by the scientifically important
He II 30.4 nm line. Irradiance measurements, without spatial resolution, are important to scientific
investigations of the Sun and its variability as well as for the atmospheric studies and operational
modeling efforts described in Chapters 3 and 4.
5.1 Instrument Overview
The SEM instrument (Hovestadt et al., 1995; Judge et al., 1998), shown in figure 5.1, uses
a highly stable freestanding transmission grating (Schattenburg and Anderson, 1990; Scime et al.,
1995; Gruntman, 1995; Wieman et al., 2007) and radiation-hard silicon photodiode detectors
(Funsten et al., 2004; Krumrey, et. al. 2000) with high efficiency in the EUV/soft x-ray spectral
range. Aluminum thin film filters, one freestanding at the entrance slit and one deposited on the
surface of each of the three photodiode detectors, prevent the detection of visible light. Two of the
detectors are positioned symmetrically in the grating diffraction pattern to detect photons in the 26-
34 nm band pass diffracted in the +1 and -1 orders, and the third detector is positioned to detect the
zeroth-order with its band pass constrained by the aluminum thin films to effectively 0.1-50nm. An
electric field maintained by a mesh grid held at high voltage in front of the entrance slit deflects
charged particles (at low to moderate energies for which flux levels are appreciable) preventing
46
them from reaching the detectors and contaminating the photon signal; some isolated signal
contamination persists due to infrequent higher energy particle events.
Figure 5.1 Opto-mechanical layout of the SOHO/CELIAS/SEM
Analog photodiode photocurrent signals are measured with low-current electrometers and
converted to digital (square wave) frequencies with a linear voltage-to-frequency converter.
Onboard the spacecraft these output frequencies are counted with a 0.25 second gate time and
averaged over 15 seconds before being telemetered. These 15-second average frequencies
constitute the SEM raw data numbers (DN) and include signal due to solar photons as well as a
“dark” signal resulting from a small constant electrometer bias and detector thermal Johnson noise.
Because these DN values represent electrometer output frequencies they are typically expressed in
counts/second and are sometimes referred to as “count rates,” although this term is somewhat
misleading as the DN are NOT equivalent to the rate of individual photon counts. For this
dissertation, the term “DN” is used to refer to the SEM channel output frequencies, and the term
“effective DN” is used to refer to the portion of these DN related to solar photon signal.
47
The absolute response function of the SEM instrument (with the exception of the soft x-ray
portion of the 0.1-50 nm channel) was measured prior to the launch of SOHO at NIST SURFIII
(Furst, Graves, and Madden, 1993; Vest et al., 1999) on beam-line 9 which is equipped with a
monochromator that allows the channel response functions to be measured with 1nm resolution
over a spectral range from about 15 to 49 nm. The response function of the zeroth-order channel
for soft x-ray wavelengths shorter than 15 nm is modeled based on photoabsorption and
transmission values for the relevant materials found in Henke, Gullikson, and Davis (1993).
5.2 SEM Sounding Rocket Calibration
Calibration of the SEM instrument has been maintained over the course of the SOHO
mission through periodic sounding rocket measurements made using a clone of the SEM
instrument (Judge, 2002) and a RGIC absolute detector (Carlson et al., 1984; Ogawa and Judge,
1986). The time series of SEM Version 3.1 photon flux values for the 26-34 nm and 0.1-50 nm
bands shown in Figure 1.1 include corrections for degradation of the on-orbit SEM that have been
applied to keep them in agreement with flux values from the periodic SEM clone and RGIC
measurements. The RGIC measures photon flux in a band from approximately 5 to 57 nm
corresponding to the range of wavelengths readily absorbed by neon through photoionization. For
comparison with the SEM first-order channels, RGIC flux values have been scaled to the 26-34 nm
band by multiplying by the ratio of the 26-34 nm integrated flux over the 5-57 nm integrated flux
from the SOLERS22 reference spectrum. The RGIC flux values remain in their native ~5-57 nm
band pass for comparison with the zeroth-order channel (further details of the RGIC measurements
are in section 5.2.2 below).
48
The correction for degradation was introduced when early sounding rocket measurements
indicated that SOHO/SEM sensitivity was gradually decreasing with time (Judge, 2002). The
reduction in sensitivity has been attributed to the buildup and subsequent UV-photon-driven
polymerization of hydrocarbons on the aluminum thin film filters. A finite contaminant source,
present with an exponentially diminishing pressure, is assumed. The degradation is accordingly
modeled as a contaminant layer that grows in thickness, τ, with time according to:
𝜏 (𝑡 ) = 𝑎 + 𝑏 ∙ 𝑒 −𝑡 /𝑐 (5.1)
where t is time from the beginning of the mission and a, b, and c are parameters that are
determined to minimize discrepancies between sounding rocket and corresponding on-orbit
measurements. The exact composition of the contaminant is unknown but pure carbon is used to
define wavelength dependence in the degradation model and has resulted in good agreement with
sounding rocket measurements as shown in Figure 1.1. The degradation factor is applied as a
wavelength dependent transmission value between 0 and 1 determined from carbon
photoabsorption cross-sections, σC(λ), as:
𝑓 𝑑𝑒𝑔𝑟𝑎𝑑 (𝜆 , 𝑡 ) = 𝑒 −𝜎 𝐶 ∙𝜏 (5.2)
Sounding rocket irradiance measurements and concurrent measurements from the SEM on-
board SOHO are shown in table 5.1 for the 26-34 nm channel and in table 5.2 for the 0.1-50 nm
channel. While the comparisons in tables 5.1 and 5.2 include sounding rocket flights through 2012,
the degradation model parameters (a, b, and c from equation 5.1) for Version 3.1 were determined
based on sounding rocket flights 36.147 through 36.236 (i.e., the flights after 2006 were not
49
included in the optimization of the Version 3.1 parameters). The sounding rocket and on-orbit
SEM irradiance values in these tables are determined from raw data using the SOLERS22
reference spectrum data pre-launch SEM response function used in Version 3.1 as opposed to the
updated data processing approach described in Chapters 7 and 8 of this dissertation.
Table 5.1. SEM sounding rocket calibration measurements compared to concurrent on-orbit
measurements with SOHO/SEM in the 26-34 nm band. Also shown are measurements from a
similar bandpass channel of the EVE/ESP obtained from the SDO/EVE series of sounding rocket
calibration flights. The SEM Version 3.1 degradation model parameters were established based on
sounding rocket flights 36.147 through 36.236 shown in bold type.
Date
Sounding
Rocket
Flight
number
Calibration
Instrument
Calibration 26-
34 nm flux
measurement
(ph/cm2/s)
Corresponding 26-
34 nm SOHO/SEM
V3.1 measurement
(ph/cm2/s)
Ratio:
SOHO/SEM
over Calibration
measurement
26 June 1996 36.147
SEM clone 1.32E+10
1.21E+10
0.92
RGIC 1.15E+10 1.05
11 August
1997
36.164
SEM clone 1.28E+10
1.42E+10
1.11
RGIC 1.36E+10 1.04
18 August
1999
36.181
SEM clone 2.09E+10
2.22E+10
1.06
RGIC 2.24E+10 0.99
6 August 2002 36.202
SEM clone 2.29E+10
2.28E+10
1.00
RGIC 2.43E+10 0.94
5 December
2003
36.211
SEM clone 1.75E+10
1.78E+10
1.02
RGIC 1.67E+10 1.07
3 August 2005 36.227
SEM clone 1.53E+10
1.57E+10
1.03
RGIC 1.52E+10 1.03
7 November
2006
36.236
SEM clone 1.20E+10
1.26E+10
1.05
RGIC 1.22E+10 1.03
14 April 2008 36.240
EVE/ESP
clone
8.60E+9 9.5E+9 1.10
3 May 2010 36.258
EVE/ESP
clone
1.05E+10 1.18E+10 1.12
23 March 2011 36.275
EVE/ESP
clone
1.25E+10 1.37E+10 1.10
23 June 2012 36.286
EVE/ESP
clone
1.21E+10 1.31E+10 1.08
24 July 2012 36.263
SEM clone 1.51E+10
1.48E+10
0.98
RGIC 1.51E+10 0.98
50
Table 5.2. SEM sounding rocket calibration measurements using the RGIC compared to
concurrent on-orbit measurements with SOHO/SEM in the 0.1-50 nm band. The SEM Version 3.1
degradation model parameters were established based on sounding rocket flights 36.147 through
36.236 shown in bold type.
Date
Sounding Rocket
Flight number
Calibration 5-
57 nm flux
measurement
(ph/cm2/s)
Corresponding
0.1-50 nm
SOHO/SEM V3.1
measurement
(ph/cm2/s)
Ratio:
SOHO/SEM over
Calibration
measurement
26 June 1996 36.147 2.34E+10 2.31 E+10 0.99
11 August 1997 36.164 2.85E+10 2.73E+10 0.96
18 August 1999 36.181 4.57E+10 4.48E+10 0.98
6 August 2002 36.202 4.95E+10 4.58E+10 0.93
5 December 2003 36.211 3.54E+10 3.50E+10 0.99
3 August 2005 36.227 3.13E+10 3.08E+10 0.98
7 November 2006 36.236 2.48E+10 2.43E+10 0.98
24 July 2012 36.263 3.10E+10 2.98E+10 0.96
5.2.1 SEM clone sounding rocket instrument
The SEM clone, nominally identical to SOHO/SEM, is preserved in an Ultra High Vacuum
storage tank at the USC Space Sciences Center between rocket flights. An “end-to-end” calibration
of the clone instrument is performed on beam line 9 at the NIST Synchrotron Ultraviolet Radiation
Facility (SURF) shortly before and shortly after each flight. Figure 5.2 shows SEM clone raw data
from a typical rocket flight. As with SOHO/SEM EUV flux values are determined based on EUV
flux distribution (from SOLERS22) within the SEM band pass and the NIST-measured,
wavelength-dependent instrument quantum efficiencies. For the sounding rocket flights a
51
correction (between 1-8% depending on date of flight) for EUV absorption in the residual
atmosphere above the rocket apogee is applied. This correction is determined from the
NRLMSISe-00 (Picone et al., 2002) thermospheric density model, the absorption cross sections of
N2, O2 and O and the SOLERS22 reference spectrum. The JB2008 thermospheric density model is
not used to determine this atmospheric correction because although the JB2008 produces better
agreement with satellite drag measurements in terms of total mass density, it does not directly
provide information about the number density of individual species.
Figure 5.2 – Raw data from the SEM clone instrument obtained during a sounding rocket
calibration flight. ChA and ChC are the SEM first-order (26-34 nm) channels. ChB refers to the
zeroth-order (0.1-50 nm) channel. The difference signal levels for the two first-order channels is
due to differences in dark current levels between the two photodiode detectors as well as
asymmetry in the transmission grating +1 and -1 order diffraction efficiencies.
52
5.2.2 Neon Rare Gas Ionization Cell
The Neon Rare Gas Ionization Cell measures absolute solar flux in the ~5-57.5 nm spectral
range (as determined by the ionization region of Ne). The instrument is similar to the radiometric
standard used at the NIST SURF facility prior to 2005, so in addition to the calibrated SEM clone,
we also fly our version of the calibration standard.
The RGIC (Fig. 5.3) is an absolute detector which measures photo-ion current
generated by the total absorption of incoming EUV in a neon target gas. During measurements, the
cell is periodically filled with Ne working gas to a pressure sufficient to completely absorb
incoming EUV within the Ne ionization region (photoionization is the only appreciable absorption
process for neon) and is then allowed to vent through the windowless front aperture.
Figure 5.3 – Schematic of the RGIC absolute detector
53
Electron-ion pairs generated in the photoionization process are collected and the ion current
is measured using a highly stable electrometer, which is calibrated before and after each rocket
flight. Flux values are determined based on the ion current vs. target gas pressure profiles
measured as the cell vents. Approximately 12 target gas fill-vent cycles occur over the course of
one rocket flight allowing for several flux measurements near rocket apogee as well as several on
the upward and downward legs of the flight. The RGIC has no optical surfaces to degrade and
because the working gas is continuously cycled through the cell, the instrument does not suffer
from changes in responsivity related to accumulation of contaminant gas.
As with the SEM clone instrument, a correction is made for the small amount of
absorption in the residual atmosphere above the rocket apogee. In the case of the RGIC, the above-
atmosphere flux level (I0 in Fig. 5.4) is determined as the value which provides the best fit between
RGIC measurements made near apogee and an atmospheric absorption model (similar to the
previously mentioned NRLMSIS/SOLERS22 model used to correct the SEM clone)
54
Figure 5.4 – Transmission of the residual atmosphere above ~160 km based on RGIC flight data
(blue circles) and the NRLMSISe-00 thermospheric density model. The above-atmosphere
irradiance, I0 is determined as the value which provides the best fit between the measurements and
the model.
5.3 SEM 26-34 nm channel
5.3.1 Revised Channel response function
The Version 3.1 response function of the SOHO/SEM first-order (26-34 nm) channel
measured prior to launch is plotted as a black line in Figure 5.5. The SOLERS22 reference solar
spectrum (dashed line) is also included for comparison. For the early SOHO/SEM and SEM clone
calibrations a portion of the housing was removed as a precaution to prevent excessive mechanical
stresses in the free-standing grating due to gas pressure gradients that would otherwise develop
across the grating during the repeated evacuation and venting of the beamline test chamber with
each calibration. During a 2007 post flight calibration of the SEM sounding rocket clone
55
instrument (by which time the mechanical stability of the gratings had been established with
greater certainty) the complete housing was used and was found to produce two additional peaks in
the first-order response function, one on either side of the primary 26-34 nm band pass. This effect,
illustrated in Figure 5.6, is attributed to the grazing incidence reflection of longer wavelength
photons (and shorter wavelength photons diffracted in the second order) off the inner surface of the
housing and into the first-order detectors. The 2007 SEM clone calibration indicates that the SEM
sensitivity to wavelengths outside the nominal 26-34 nm band pass – particularly to peaks in
emission between 19 and 25 nm – is greater than originally assessed. This additional sensitivity,
not accounted for by the response function determined pre-flight, results in a systematic calibration
error in the Version 3.1 irradiance values.
For the revised version of the SEM irradiances reported here, modified first-order response
functions (the +1 and -1 order response functions are slightly different) have been modeled for the
SOHO/ SEM based on the SEM clone calibration measurements taken with the complete housing
in contrast to all previous calibrations performed with the cover removed. The modeled
SOHO/SEM first-order response functions, shown for one of the channels as a gray line in Figure
5.5, is obtained by substituting the SEM clone response values (the average of the two, +1 and -1,
first-order channels of the SEM clone) from 13-26 nm and from 34-47 nm into the SOHO/SEM
response function after scaling the values by the ratio of the peak efficiency for SOHO/SEM over
that for the SEM clone.
56
Figure 5.5 SOHO/SEM first-order response function used for irradiance calculation in Version 3.1
(black curve partially obscured by the gray curve which is identical over the wavelength range
from 26 to 34 nm) and in the updated version reported here (gray line). The SOLERS22 spectrum
is shown for reference (dotted line).
Figure 5.6 – Source of additional sensitivity “peaks” in the SEM response profile (gray curves on
either side of the primary response peak in Figure 5.5). The green lines outline the path of ~40 nm
photons diffracted in the first-order or ~20 nm photons diffracted in the second order and reflected
off of the inner surfaces of the SEM housing. Sensitivity related to this effect is limited to
wavelengths with diffraction angles between 10.65° (corresponding to ~17 nm, purple lines) and
13.5° (corresponding to ~47 nm, red lines).
57
5.3.2. Data processing algorithm
Shown below is the algorithm for determining 26-34 nm irradiance from SEM raw data
(i.e. DN) based on the SOLERS22 reference spectrum. The approach for substituting the MEGS
reference spectrum is described in Chapter 6.
The SEM 26-34 nm flux, ΦSEM1 (expressed in photon flux units [ph cm
-2
sec
-1
] in Version
3.1), is the mean of the two symmetric plus and minus first-order channels. The flux in each
channel is calculated from its detector’s DN according to:
Φ
𝑆𝐸𝑀 1
= 𝑘 1
𝐷𝑁
𝑆𝐸𝑀𝑐 ℎ1
−𝐼 𝑏𝑘𝑔𝑟𝑑 𝐴 ∙∫ 𝜂 1
∙𝜙 𝑆 22
∙𝑓 𝑑𝑒𝑔𝑟𝑎𝑑 ∙𝑑𝜆 ∙𝑓 1𝐴𝑈
𝜆 2
𝜆 1
∫ 𝜙 𝑆 22
𝑑𝜆
𝜆 2
𝜆 1
(5.3)
where:
k1 , defined below based on the SOLERS22 reference spectrum, is a correction for the SEM
sensitivity band which extends slightly beyond 26-34 nm (including second order contributions
from wavelengths near 17 nm),
Ibkgrd is the background signal due to diode/electrometer dark current and residual light leaks,
DNSEMch1 is the first-order channel raw DN,
A is the entrance aperture area,
58
η1 is the SEM first-order channel efficiency from NIST calibrations (described above in section
5.3.1) performed before the SOHO 1995 launch,
ϕS22 is the reference spectrum solar flux (i.e. SOLERS22 for SOHO/SEM Version 3.1),
fdegrad is a time and wavelength dependent degradation factor (described above in section 5.2)
based on sounding rocket calibration flight through 2006 for SEM Version 3.1,
f1AU is a correction to normalize observations to a distance of 1 AU, and
λ1 to λ2 is the range of wavelengths over which the SEM first-order channel is sensitive, which
extends slightly beyond the reported 26-34 nm band. The specific values of λ1 and λ2 vary slightly
depending on which of the first-order channels (i.e. +1 or -1 diffraction order) is considered and
whether the pre-flight or modified response function is used.
The SEM response function, η1, as well as the degradation term, fdegrad, are both functions
of wavelength and thus it is necessary to determine a weighted mean value for these terms with the
weight factors equal to the relative intensity of spectral bins within a reference solar spectrum. The
SOLERS22 reference spectrum has been used throughout the SOHO mission including the most
recent release, Version 3.1. Additionally, the value, k1, which corrects for SEM first-order
sensitivity that extends slightly beyond the reported 26-34 nm bandpass and includes some second
order contribution from wavelengths near 17 nm, is also determined based on the SOLERS22
reference spectrum according to:
𝑘 1
=
∫ 𝜙 𝑆 22
𝑑𝜆
34𝑛𝑚
26𝑛𝑚
∫ 𝜙 𝑆 22
𝜆 2
𝜆 1
𝑑𝜆
(5.4)
59
Thus, the weighted average efficiency, the weighted average degradation factor, and the
second order/out of band correction are all dependent on the spectral shape, but not the absolute
irradiance values of the reference spectrum. Equation (5.3) can be simplified by substituting the
right-hand side of Equation (5.4) in for k1, leaving:
Φ
𝑆𝐸𝑀 1
=
(𝐷𝑁
𝑆𝐸𝑀𝑐 ℎ1
−𝐼 𝑏𝑘𝑔𝑟𝑑 )∙∫ 𝜙 𝑆 22
𝑑𝜆
34
26
𝐴 ∙∫ 𝜂 1
∙𝜙 𝑆 22
∙𝑓 𝑑𝑒𝑔𝑟𝑎𝑑 ∙𝑑𝜆 ∙𝑓 1𝐴𝑈
𝜆 2
𝜆 1
(5.5)
The quantity in the numerator, (DNSEMch1 – Ibkgrd), is the portion of the total DN that is
related to solar photons and will be referred to as “effective DN” throughout the text.
5.4 SEM 0.1-50 nm channel
5.4.1 Channel response function
The SOHO/SEM 0.1-50 nm response function is constrained spectrally by the aluminum
thin film filters in front of the entrance slit and deposited on the zeroth-order detector. The profile,
shown in Figure 5.7, is non-uniform over the reported spectral range with most of the signal
coming from the EUV band between 15 and 50 nm and the soft x-ray band shorter than 5 nm.
Unlike the first-order channel detectors, the zeroth-order detector is not within the path of photons
reflected at grazing incidence off of the inside of the SEM housing and therefore the zeroth-order
response function remains nominally the same whether or not the complete housing is installed.
Thus, for the work reported the original zeroth-order response function that has been used
throughout the SOHO/SEM mission will continue to be used.
60
Figure 5.7 – Soft x-ray (top panel) and EUV (bottom panel) portions of the SOHO/SEM zeroth-
order response function used for irradiance calculations in both Version 3.1 and in the updated
version reported here (black line). The SOLERS22 spectrum is shown for reference (dotted line,
units are arbitrary but consistent between the top and bottom panels).
61
5.4.2 Zeroth-order data processing algorithm
Shown below is the Version 3.1 algorithm for determining 0.1-50 nm irradiance from SEM
zeroth-order channel DN based on the SOLERS22 reference spectrum. The revised approach
adopted as part of this work, for which time-dependent reference spectra (for example those from
EVE MEGS) are used in place of SOLERS22, is described in Chapter 6.
Solar soft x-ray fluxes are known to have greater variability with time than EUV fluxes, but
because the SOLERS22 spectrum is fixed with time, it does not capture this relative variability.
Thus it is not expected (Judge, et. al., 2002) that SOLERS22 provides a reliable reference spectrum
over the full SEM zeroth-order (0.1-50 nm) range of sensitivity for all levels of solar activity. To
account for changes in the relative contribution of soft x-ray versus EUV photons over time,
different time-dependent scaling factors, α and χ are applied to the soft x-ray (0.1.-5 nm) and EUV
(5-50 nm) portions of the adopted reference spectrum respectively:
𝜙 𝑆𝐸𝑀 0
(𝜆 ) = {
𝛼 ∙ 𝜙 𝑆 22
(𝜆 ), 𝜆 = 0.1 − 5𝑛𝑚
𝜒 ∙ 𝜙 𝑆 22
(𝜆 ), 𝜆 = 5 − 50𝑛𝑚
(5.6)
where ϕS22 is the SOLERS22 reference spectrum for Version 3.1, and the reported SEM
zeroth-order 0.1-50 nm photon flux, ΦSEM0, is the integral of the scaled reference spectrum:
Φ
𝑆𝐸𝑀 0
= ∫ 𝜙 𝑆𝐸𝑀 0
𝑑𝜆
50
0.1
(5.7)
The scaling factor for the EUV wavelengths, χ, is calculated such that the constructed
spectrum, when integrated over the 26-34 nm bandpass is equal to ΦSEM1, the flux measured in the
first-order channels according to:
62
𝜒 =
Φ
𝑆𝐸𝑀 1
∫ 𝜙 𝑆 22
𝑑𝜆
34𝑛𝑚
26𝑛𝑚
(5.8)
The EUV portion of the reference spectrum scaled by χ, and the zeroth-order response
function, η0(λ) for λ = 5-50 nm, provide the estimated portion of the SEM zeroth-order raw data
number, DNSEMch0, that is related to EUV photons. The remaining portion of the zeroth-order
effective DN – those related to soft x-ray photons – are then used with the response function, η0(λ)
for λ = 0.1-5 nm, to determine the scaling factor, α, according to:
𝛼 =
(
𝐷𝑁
𝑆𝐸𝑀𝑐 ℎ0
−𝐼 𝑏𝑘𝑔𝑟𝑑 0
𝑓 1𝐴𝑈
∙𝐴 )−∫ 𝜒 ∙𝜙 𝑆 22
∙𝜂 0
∙𝑓 𝑑𝑒𝑔𝑟𝑎𝑑 𝑑𝜆
50𝑛𝑚
5𝑛𝑚
∫ 𝜙 𝑆 22
∙𝜂 0
∙𝑓 𝑑𝑒𝑔𝑟𝑎𝑑 𝑑𝜆
5𝑛𝑚
0.1𝑛𝑚
(5.9)
where fdegrad is based on the same carbon growth model described in section 2.1 and used
for calculating degradation in the first-order channels.
63
6. SDO/EVE
The EVE, one of three instruments launched aboard the Solar Dynamics Observatory,
provides solar spectral irradiance measurements in the EUV and SXR ranges which are
unprecedented in terms of spectral resolution, time cadence, accuracy, and precision. The EVE
consists of 5 instrument channels including:
The Multiple EUV Grating Spectrograph (MEGS) A and B channels which
together provide high spectral resolution (~.02 nm) measurements over the 6-
100 nm range. A power supply malfunction occurring May 26, 2014 has
indefinitely suspended MEGS-A operations thus limiting MEGS spectral coverage
to 35-100 nm.
The EUV SpectroPhotometer (ESP) channel, designed and built at the USC
Space Sciences Center, is very similar to SEM in terms of its long-term stable
design, but includes a greater number of spectral band passes covering the range
from ~0.1-37 nm.
The Solar Aspect Monitor (SAM) channel, a filtered pinhole camera that
provides low resolution images of the sun in the 0.1-7 nm SXR spectral range.
Some resolution of spectra within this range can be achieved through photon event
detection. Operation of the SAM instrument was also suspended as a result of the
May 2014 power supply malfunction.
The MEGS P channel, a photometer channel similar to those incorporated into
ESP, but is configured to detect Lyman-α (121.6 nm) irradiance.
64
The EVE channels are calibrated pre-flight at NIST/SURF and their calibration is
maintained through periodic sounding rocket measurements using an EVE clone instrument.
Additionally, EVE has provisions for in-flight measurements of degradation related to filter
contamination, change in detector sensitivity and dark current (Hock et al., 2012; Didkovsky et al.,
2012).
6.1 SDO/EVE/ESP
The EVE/ESP channel is very similar in design to the SEM and is based on the same type
of free-standing transmission grating. ESP also includes several improvements over the SEM
design, including the aforementioned capacity for in-flight degration monitoring, a greater number
of band passes and a greater telemetry bandwidth allocation which allows it to provide
measurements with a factor of 60 higher time resolution than SEM. One of the ESP band passes is
approximately the same as the SEM 26-34 nm channel. Because of the good calibration of ESP
maintained through sounding rocket measurements and on-board monitoring, and the similarity of
its spectral measurements, the ESP is an excellent source of inter-comparison during the period of
overlap between the SOHO and SDO missions. This work relies heavily on this inter-comparison
as discussed in Chapter 7.
6.2 SDO/EVE/MEGS
While SOLERS 22 does not capture variations in the solar spectral distribution with time,
continuous time varying spectra are available from the EVE/MEGS channels. Because the issue of
calibration has been so thoroughly addressed in the EVE program, the EVE/MEGS measured
spectra were, prior to the loss of MEGS A, the most reliable source of reference spectra to use for
65
processing SOHO/SEM data. In addition to using them for this purpose for processing the SEM
datasets presented in the next chapter, the EVE/MEGS spectra are also integrated over the SEM
band passes to provide an additional source of intercomparison.
The high resolution MEGS spectra cover the entire range of SOHO/SEM first-order
channel sensitivity, so these spectra, ϕMEGS(t,λ), can be substituted directly into the first-order
irradiance equation (5.5) in place of ϕS22(λ). Because the MEGS spectra do not cover the soft x-ray
range shorter than 6 nm, for the zeroth-order SOHO/SEM channel a different reference spectrum
must be used to cover this short wavelength spectral range. Since measured soft x-ray spectra are
not readily available, for this work the SOLERS22 is used for this purpose and the following
scaled reference spectrum is adopted (in place of equation 5.6 above) for calculating zeroth-order
irradiances:
𝜙 𝑆𝐸𝑀 0
(𝜆 ) = {
𝛼 ∙ 𝜙 𝑆 22
(𝜆 ), 𝜆 = 0.1 − 5𝑛𝑚
𝜒 ∙ 𝜙 𝑆 22
(𝜆 ), 𝜆 = 5 − 7𝑛𝑚
𝜒 ∙ 𝜙 𝑀𝐸𝐺𝑆 (𝑡 , 𝜆 ), 𝜆 = 7 − 50𝑛𝑚
(6.1)
Note, although the shortest wavelength included in the SDO/EVE spectra is at 6 nm, a
lower boundary of 7.0 nm is used as a convention to accommodate alternate irradiance conversion
algorithms which use measurements in the nominal 0.1-7 nm carbon titanium filter band pass (e.g.
SDO/EVE/ESP quad diode, TIMED/SEE/XPS) to derive soft x-ray reference spectra. Because
solar irradiance is low in this range, the slight shift of this boundary has little effect on the
irradiance calculation.
66
6.3 A comparison of EVE/MEGS spectra to SOLERS22
In Figure 6.1, The SOLERS22 reference spectrum is compared to daily average SDO/EVE
Ver. 2 spectra for days of low (top panel), moderate (middle panel), and high (bottom panel) levels
of activity. Also shown in the top panel is the SOHO/SEM updated response function for one of
the SOHO/SEM first-order channels (scaled by an arbitrary factor to fit the plot). In addition to
time dependence, there are some characteristic, time-independent differences between the EVE
spectra and the SOLERS22 spectrum. For example, for all activity levels, the wavelength bins
centered at 28.5 nm and 30.5 nm (which include the Fe XV coronal and He-II transition region
emission lines respectively) make a lesser contribution to the EVE spectral distribution than they
do for the SOLERS22 spectrum. When substituting the EVE spectra for SOLERS22 in the
irradiance calculation, these differences affect the weighted average instrument response and
degradation factor, and the k1 correction factor in equation (5.3).
67
Figure 6.1 A comparison of the SOLERS22 fixed reference spectrum (dotted line) to measured
SDO/EVE reference spectra (black line) for low (4/30/2010, top panel), medium (3/21/2013,
middle panel), and high (11/11/2011, bottom panel) levels of solar activity – the SDO/EVE spectra
shown are the daily average spectra for the specified day. In the top panel the SOHO/SEM first-
order response function (dashed line) is shown for reference. Time-independent, characteristic
differences exist between the spectra including a smaller contribution from the wavelength bins
centered at 28.5 nm and 30.5 for the EVE/MEGS spectrum for all activity levels.
68
7. Resolving Differences between SOHO/SEM and SDO/EVE
In this chapter a series of comparisons is performed between SOHO/SEM and SDO/EVE
measurements to demonstrate that the revised first-order response function presented in Chapter 5
and the approach of using MEGS measured reference spectra can largely resolve differences
observed between the current (Version 3.1) irradiances and those measured by EVE.
7.1 Comparisons of the 26-34 nm daily average time series
Figure 7.1 compares the updated SOHO/SEM and Version 3.1 first-order (26-34 nm) daily
average irradiances to both the MEGS and ESP channels of EVE. The SEM Version 3.1 data,
which are normally published in photon flux units (photons/cm
2
/sec), have been converted to
irradiance units (W/m
2
) using the SOLERS22 reference spectrum to describe the energy
distribution within the SEM band
*
. For these comparisons the EVE/MEGS Version 2 absolute
spectra are integrated over the 26-34 nm band in contrast to their use as a reference spectrum in
SOHO/SEM data processing, for which they are normalized and only provide information about
the relative spectral distribution. The EVE ESP data are converted to irradiances based on NIST-
measured channel response functions and adopted reference spectra (the reference spectra used in
ESP data processing are discussed in section 8.3) using an algorithm similar to the one described in
section 5.3.2 for the SEM first-order channels. The comparison with EVE/ESP uses ESP Ver. 2
data from channel 9 which measures in the 26.7-33.8 nm band pass. The ESP irradiances are
scaled to account for the slight difference in band pass by a factor equal to the ratio of the
*
Because the conversion between Photon Flux and Irradiance units is dependent on the spectral distribution,
the SEM data, for which the presumed spectral distribution is known, are converted to match the units of the ESP data
(thus avoiding assumptions about the reference spectrum adopted to process the ESP data – assumptions that would
have to be made to convert ESP irradiance values to photon flux units).
69
SDO/EVE/MEGS integrated irradiances for the two band passes (the SEM band over the ESP
band) for the corresponding day, according to:
Φ
𝐸𝑆𝑃 ′
= Φ
𝐸𝑆𝑃 ∙
∫ 𝜙 𝑀𝐸𝐺𝑆 𝑑𝜆
34
26
∫ 𝜙 𝑀𝐸𝐺𝑆 𝑑𝜆
33.8
26.7
(7.1)
The daily ratio values in Figure 7.1b for the updated SOHO/SEM over Version 3.1 range
from about 0.79 to about 0.83 over the time series shown and depend on the MEGS spectral
distribution for the corresponding day. Solar cycle variability of these distributions results in a
slight long term trend in the ratio values apparent from the slope of the linear fit line of about 10
-5
day
-1
. Since the only time dependent difference between the updated and Version 3.1 algorithms is
the change in reference spectrum with solar activity level, this trend is expected to reverse with the
transition to the solar cycle 24-25 minimum. For the SOLERS22 reference spectrum, irradiance for
the 30.4 nm He II line is characteristically higher than for the MEGS spectra. Since this
wavelength is near the peak of the SEM response function, the use of SOLERS22 results in a
higher value for the weighted mean instrument response and thus lower values for the calculated
irradiance according to Equation (5.3). However, due to the broader SEM response function, the
updated SEM irradiances include a larger correction (i.e. lower k1 values based on Equation 5.4)
for SEM signal outside of the reported SEM 26-34 nm band pass. The latter effect is more
significant and results in the approximately 20% decrease for the updated irradiance values, which
puts them in much better agreement with both the integrated MEGS spectra and the ESP ch9
irradiances (compare ratios in Figure 7.1 panels c and d). Mean ratios are reduced from about 1.25
70
to about 1.01 for the comparison with ESP Ch9 and from about 1.27 to about 1.03 for the
comparison with MEGS.
An apparent long term trend in the SEM over ESP ratios in Figure 7.1c suggests that both
SOHO/SEM versions are showing a greater increase with solar cycle 24 than EVE/ESP Ch9.
However this trend is not consistent throughout the entire SDO mission. For example, linear fits to
the ratios over the first half of the mission suggest an opposite trend. The source of such trends is
not yet clear. One possible explanation is that the inaccuracies introduced by using daily average
reference spectra to calculate irradiance for days in which significant changes in spectral
distribution occur are not likely to be the same for SOHO/SEM as they are for EVE/ESP since the
two instruments have different response functions. A second possibility is that the divergence is
related to the higher susceptibility of the SOHO/SEM instrument to signal contamination due to
energetic particles since the SEM instrument housing is much thinner and more easily penetrated
by lower energy particles (Didkovsky et al., 2007) than that of EVE/ESP. The divergence between
the SOHO/SEM and EVE/ESP datasets appears to increase with the transition from lower to
higher levels of solar activity, which is consistent with both of these explanations. A third source
of divergence could be related to changes in dark currents for one or both of the instruments. For
SOHO/SEM the dark currents are assumed to be stable and for most of the EVE/ESP version 2
data a dark current model which accounts for changes in temperature only is used, yet in-flight
dark current measurements with EVE/ESP have shown evidence of small changes in dark currents
with both temperature and time.
71
Sharp drops followed by gradual rises in the SEM / EVE (MEGS) ratios of Figure 7.1d
(e.g., on julian days 2455364, 2455463, 2455999) are due to abrupt increases of EVE/MEGS
irradiance values following CCD bake-outs performed to restore sensitivity in the MEGS B
channel.
72
Figure 7.1 a) A comparison of the updated SOHO/SEM 26-34 nm irradiance time series (black curve)
with SOHO/SEM Version 3.1 (dotted black curve), SDO/EVE/MEGS (dotted gray curve) Version 2
spectra integrated over the 26-34 nm bandpass, and EVE ESP ch9 (gray curve). b) Ratio of the updated
SOHO/SEM daily average irradiances over those of SOHO/SEM Version 3.1. c) Ratios of the
SOHO/SEM Version 3.1 daily average irradiances over the daily average SDO/EVE/ESP Channel 9
irradiances compared to similar ratios based on the updated SOHO/SEM. d) Ratios of the SOHO/SEM
Version 3.1 daily average irradiances over the daily average integrated SDO/EVE/MEGS spectra
compared to similar ratios based on the updated SOHO/SEM. Mean ratio levels are shown as dashed
lines in panels c and d. Mean ratios for Version 3.1 SEM are about 1.26 for the comparison with ESP
(panel c) and 1.28 for the comparison with MEGS, but are much closer to unity (~1.01 in panel c and ~
1.03 in panel d) for the updated SEM. The broader response function and the use of MEGS reference
spectra for the updated SOHO/SEM irradiances result in significantly better agreement with both
SDO/EVE channels. Divergence between the SOHO/SEM and SDO/EVE/ESP Ch9 irradiances in 2012
could be related to higher sensitivity of the SEM to energetic particles compared to ESP or to errors
related to using daily average reference spectra on days with high solar activity (see text for details).
73
7.2 Comparison of 26-34 nm measurements under flare conditions
Because the updated SOHO/SEM irradiance measurements are based on time-dependent
reference spectra, they should be less susceptible to loss of accuracy due to rapid changes in solar
spectral distribution that occur during a solar flare than the Version 3.1 measurements. They are
expected therefore to be better correlated with MEGS integrated spectra under such conditions. For
example, in Figure 7.2 updated and Version 3.1 SOHO/SEM measurements are compared to
MEGS spectra integrated from 26-34 nm for a period of three hours surrounding a Jan 27, 2012 X-
class solar flare (X1.7, N33W85). The time cadence for the measurements is 10 sec for the MEGS
spectra and 15 sec for the SOHO/SEM irradiances, thus near real-time (i.e. with an offset of 5 sec
or less) reference spectra are available for the SOHO/SEM irradiances calculated using our
updated approach. Near the peak of the flare, the ratio of updated over Version 3.1 SOHO/SEM
irradiance in Figure 7.2b shifts rapidly by about 4% compared to its pre-flare level. During this
time the updated SOHO/SEM values which are calculated using the time varying reference spectra
track the EVE/MEGS integrated spectra much more closely than Version 3.1. The ratio of Version
3.1 SOHO/SEM irradiances over MEGS integrated irradiance shifts by nearly 5% (Figure 7.2c) at
the flare peak while the ratio for the updated irradiances shifts by only about 1 or 2% (Figure
7.2d). This small shift is possibly due to remaining inaccuracy in the SOHO/SEM response
function for which the out of band sensitivity was not measured directly, but derived from
measurements of the SEM clone. Nonetheless, the relative consistency of the ratio throughout the
course of the flare suggests that the updated parameters are a significant improvement over those
from Version 3.1.
74
For the three hour period shown in Figure 7.2, the correlation between Version 3.1 and
MEGS is quite good, with correlation coefficient, rcor = 0.98, in spite of the shift in ratio values
around the peak of the flare evident in Figure 7.2c. Nonetheless, for the updated SEM the
correlation with MEGS improves to rcor = 0.99. While this improvement is small, it is consistent
throughout the period of SDO and SOHO mission overlap. For 10 out of the 14 X-class flares
occurring during this period the updated SOHO/SEM data results in moderately higher correlation
coefficients compared to Version 3.1, and for the remaining 4 the correlation coefficients are
equal. Abnormally low correlation coefficients were found for some flares (e.g., for the 23 October
2012 flare, rcor = 0.76 for Version 3.1 and 0.83 for the updated version) and are attributed to the
aforementioned contamination of the SOHO/SEM signal by flare related energetic particle fluxes
which rise near the end of the three hour period covered in the correlation. SOHO/SEM is
susceptible to signal contamination from protons of energy 10MeV and greater (Didkovsky et al.,
2007), yet neither Version 3.1 nor the updated version is corrected for such contamination.
Nonetheless, the correlation coefficients for the updated SOHO/SEM are higher than those for
Version 3.1 in such cases where both coefficients are abnormally low.
75
Figure 7.2 a) A comparison of the updated SOHO/SEM 26-34 nm irradiance time series (black
curve) with SOHO/SEM Version 3.1 (dotted curve), and SDO/EVE/MEGS Version 2 (gray curve)
spectra integrated over the 26-34 nm bandpass with high time resolution during the X1.7, N33W85
solar flare of Jan 27, 2012 . b) The ratios of the updated SOHO/SEM irradiances over those of
SOHO/SEM Version 3.1 vary by several percent over the course of the flare due to the rapid
change of solar spectral distribution during the flare. c) Ratios of the SOHO/SEM Version 3.1
irradiances over the integrated SDO/EVE/MEGS Ver. 2 spectra deviate significantly from their pre
flare values, while ratios of the daily average updated SOHO/SEM over the daily average
integrated SDO/EVE/MEGS spectra (d) show little change.
76
7.3 Comparison of the daily average 0.1-7 nm and 7-50 nm time series
The SEM zeroth-order (0.1 – 50 nm) response function remains unchanged from Version
3.1, however because the algorithm for determining zeroth-order irradiances depends on the first-
order measurements (i.e. based on the scaling factor χ, defined in equation 5.8) it is affected by the
updates to the first-order response function. Additionally, the weighted mean zeroth-order response
function and degradation correction are dependent on the choice of reference spectrum. Thus
comparisons between the SOHO/SEM zeroth-order band with SDO/EVE can provide additional
validation for these updated data processing parameters.
None of the individual SDO/EVE channels includes the full 0.1-50 nm band measured by
the SOHO/SEM zeroth-order channel. However irradiance within a given portion of the SEM
zeroth-order band can be calculated based on the adopted reference spectrum. We calculate
irradiance in the 7-50 nm band from the SOHO/SEM measurements based on both the updated and
Version 3.1 data processing parameters for comparison with integrated MEGS absolute spectra.
This comparison is shown in Figure 7.3. We also compare the 0.1-7 nm portion of the SEM zeroth-
order band to the EVE/ESP zeroth-order quad diode (QD) channels in Figure 7.4.
For both of these comparisons SOHO/SEM irradiances are in significantly better agreement
with the EVE measurements when calculated using the updated data processing parameters. In the
7-50 nm band, the mean ratio for SOHO/SEM over MEGS integrated spectra is about 1.03 for the
updated irradiance values compared to about 0.84 for the Version 3.1 values. For the 0.1-7 nm
comparisons with the ESP QD channels the mean ratio is about 1.15 based on the updated
SOHO/SEM irradiances and about 1.4 for Version 3.1. Because the soft x-ray response functions
77
and reference spectra are modeled for both of the SOHO/SEM versions and for the SDO/EVE/ESP
QD channel (Didkovsky et al., 2012), large differences among datasets are not unexpected in the
0.1-7 nm range. For example, Judge (2002) demonstrated that the results of derivations of soft x-
ray fluxes from SOHO/SEM measurements vary significantly depending on which of two modeled
reference spectra is used. This result highlights the need for solar spectral irradiance measurements
in the soft x-ray range.
78
Figure 7.3 a) A comparison of the 7-50 nm SOHO/SEM irradiance time series extracted from the
zeroth-order measurements for the updated version (black curve) with that for Version 3.1 (dotted
curve), and with daily average 7-50 nm MEGS integrated spectra (gray curve). b) Ratios of the
SOHO/SEM Version 3.1 daily average irradiances over the daily average SDO/EVE/MEGS
spectra integrated from 7-50 nm (dotted line) compared to similar ratios based on the updated
SOHO/SEM (solid black line). Mean ratio levels are shown as dashed lines in panel b. The mean
ratio for Version 3.1 SEM is about 0.84 but is much closer to unity (~ 1.03) for the updated SEM.
79
Figure 7.4 a) A comparison of the 0.1-7 nm SOHO/SEM irradiance time series extracted from the
zeroth-order measurements for the updated version (black curve) with that for Version 3.1 (dotted
curve), and with daily average 0.1-7 nm ESP measurements (gray curve). b) Ratios of the
SOHO/SEM Version 3.1 daily average irradiances over the daily average SDO/EVE/ESP 0.1-7 nm
measurements compared to similar ratios based on the updated SOHO/SEM. Mean ratio levels are
shown as dashed lines in panel b. The mean ratio for Version 3.1 SEM is about 1.4 but is closer to
unity (~ 1.15) for the updated SEM. Because the soft x-ray response functions and reference
spectra are modeled for both of the SOHO/SEM versions and for the SDO/EVE/ESP QD channel
(Didkovsky et al., 2012), large differences among datasets are not unexpected in the 0.1-7 nm
range.
80
8. Pre-SDO Time Dependent reference spectra
The comparisons in the previous chapter demonstrate that differences of about 20%
between SOHO/SEM and SDO/EVE measurements of 26-34 nm irradiance, which have persisted
throughout the SDO mission, can be significantly reduced, to about 5% by applying a revised and
more accurate response function, and measured time-dependent reference spectra to the processing
of the SEM data. Identifying the sources of these differences, and eliminating them, thus fulfills
the first major objective of this work. The second objective of using this same approach to
recalculate the entire SOHO/SEM data set, which includes all of solar cycle 23, is of significant
value, but requires the use of an alternative time-dependent reference spectrum because of the
absence of relevant SDO/EVE/MEGS measurements.
Such a recalculation could, for example, provide verification of, or new information
regarding, lower EUV irradiance during the minimum of solar cycles 23/24 compared to that of
solar cycles 22/23. Assuming that the solar spectral distribution under solar minimum conditions is
consistent from one minimum to the next, it might seem initially that the relative comparison of
these two minima should not be affected by whether a time dependent or fixed reference spectrum
is used. However, the same fixed reference spectrum currently used in the SOHO/SEM Version
3.1 data processing algorithm is also used for processing the sounding rocket measurements which
were performed at different times/activity levels throughout the solar cycle. A change in reference
spectrum affects the 26-34 nm irradiance values calculated based on RGIC sounding rocket
measurements differently than those based on SEM clone and SOHO/SEM measurements because
the instruments have different response functions. Since the adopted reference spectrum affects the
81
distribution of sounding rocket measurements, the degradation model based upon them
(Equations 5.1 and 5.2) will also be affected.
In this chapter several different sources of reference spectra are evaluated in order to
identify one which is suitable for reprocessing the pre-SDO SEM data. Comparisons with EVE-
MEGS and EVE ESP during the period of overlap with SDO – like those performed in the
previous chapter to validate the updated SEM algorithm based on MEGS-A spectra – are the basis
of these evaluations. A combination of modeled, measured and semi-empirical reference spectra
are compared including:
1. the Flare Irradiance Spectral Model (FISM) described in Chamberlin, Woods, and
Eparvier, (2007),
2. The Solar2000/Solar Irradiance Platform proxy based spectra described in Tobiska,
W. K. et al. (2000),
3. a system of discrete spectra based on EVE MEGS measured spectra representing
different levels of activity similar to what is being used in the SDO EVE ESP
algorithm following the anomaly which has indefinitely suspended MEGS A
operations.
A brief description and preliminary evaluation of each of the above sources of reference
spectra are presented in sections 8.1 through 8.3. While there are very few direct measurements of
solar EUV irradiance spectra available from the early years of the SOHO mission, spectra have
been determined from measurements with the SOHO from the Coronal Diagnostic Spectrometer
(CDS) combined with spectral models based on Differential Emission Measures (DEM). The CDS
82
measurements and associated DEM modeling are described briefly in section 8.4. In section 8.5,
the FISM, Solar2000, and ESP algorithm reference spectra are inter-compared with each other and
with discrete solar minimum (solar cycle 23/24) and solar maximum (solar cycle 23) spectra from
CDS.
8.1 Evaluation of FISM reference spectra
The FISM empirical irradiance model (Chamberlin, Woods, and Eparvier, 2007) calculates
deviations from a nominal “quiet Sun” spectrum due to solar cycle, solar rotation and flare-related
variations. Figure 8.1 shows the comparison of SEM 26-34 nm irradiances with those from EVE
ESP and from EVE MEGS-A similar to the comparison of Figure 7.1, but the “updated SEM”
irradiances based on MEGS-A reference spectra have been replaced with SEM irradiances
determined based on FISM reference spectra.
83
Figure 8.1 a) A comparison of ESP 26-34 nm irradiances determined using FISM daily reference
spectra (blue line) to SEM Ver 3.1 (dotted line), EVE MEGS (black line), and EVE ESP (gray
line). Irradiance ratios for the FISM based SEM irradiances over SEM Ver 3.1, EVE ESP, and
EVE MEGS are shown in panels b, c, and d respectively with mean ratios shown as dashed lines in
each panel.
The SEM irradiances calculated based on FISM are in much better agreement with the ESP
measurements than the Ver 3.1 irradiances, but are on average about 4% higher than the EVE ESP
values and 9% higher than the EVE MEGS values. As noted above however, the choice of
reference spectrum will have some effect on irradiances determined based on the sounding rocket
calibration measurements and on the degradation model derived from them. This comparison with
84
the EVE measurements is thus preliminary in nature in the sense that the observed differences in
absolute irradiance (of less than 10%) may be substantially reduced once the degradation model is
revised. The FISM model therefore should not be excluded as a possible source of reference
spectra based solely on this comparison.
8.2 Evaluation of Solar2000/SIP reference spectra
The Solar2000/SIP spectra described in Tobiska, W. K. et al. (2000) are available with 1-
day time cadence and are determined by evaluating deviations from a nominal “Quiet Sun”
spectrum based on proxy measurements where the relevant proxy is selected based on the source
region of a given spectral interval. For example if a 1 nm spectral bin is at a wavelength dominated
by emissions from the chromosphere, variability is determined based on Lyman-α measurements,
whereas if a bin is comprised primarily of coronal emissions, its variability is determined based on
the F10.7 index. Figure 8.2 shows the comparison of SEM 26-34 nm irradiances with those from
EVE ESP and from EVE MEGS-A similar to the comparisons of Figures 7.1 and 8.1, but in this
case the updated SEM irradiances (blue curve in panel a) being compared have been determined
based on Solar2000 daily reference spectra.
85
Figure 8.2 a) A comparison of ESP 26-34 nm irradiances determined using Solar2000 daily
reference spectra (blue line) to SEM Ver 3.1 (dotted line), EVE MEGS (black line), and EVE ESP
(gray line). Irradiance ratios for the FISM based SEM irradiances over SEM Ver 3.1, EVE ESP,
and EVE MEGS are shown in panels b, c, and d respectively with mean ratios shown as dashed
lines in each panel.
86
The SEM irradiances calculated based on Solar2000 are also in much better agreement with
the EVE measurements than the Ver 3.1 irradiances, and with average difference of about 1% and
6% relative to ESP and MEGS respectively, they are (based on this preliminary comparison) in
slightly better agreement with EVE than the FISM based SEM irradiances compared in Figure 8.1.
8.3 Evaluation of a system of discrete MEGS-A spectra
The data processing algorithm for the EVE ESP, an instrument which is very similar in
design and operation to SEM, also requires reference spectra to determine absolute irradiance
values. Initially, the EVE MEGS A spectra were used for this purpose, however following a power
supply anomaly on May 26, 2014 (approximately four years into the SDO mission), which has
indefinitely suspended operation of the EVE MEGS-A channel, a new ESP algorithm (Didkovsky,
Wieman, and Woodraska, 2014b) has been implemented which uses 11 different spectra derived
from MEGS-A spectra measured prior to the anomaly such that each representing a different level
of activity.
With this approach, each daily ESP raw data file to be processed is assigned a discrete
activity level (numbered 0 to 10 corresponding to the 11 discrete reference spectra) based on the
daily average effective DN from the zeroth-order channel. The ESP zeroth-order channel is similar
to the SEM zeroth-order channel in that its band pass includes soft X-ray wavelengths as short as
~0.1 nm but includes a C/Ti/C filter which makes the channel nominally insensitive to
wavelengths longer than about 7 nm. The zeroth-order effective DN are the raw DN values with
background (i.e. dark current) DN subtracted and corrections for filter degradation applied. The
87
scale used to assign the activity level/reference spectrum (which spans the range of zeroth-order
effective DN from < 100 DN to ~1936 DN observed over the SDO mission prior to the anomaly)
is shown in Table 8.1. Figure 8.3 shows the total number of days that fall into each of these ranges
over the course of the SDO mission prior to the anomaly.
Table 8.1: Ranges of ESP zeroth-order daily average effective DN corrected for zeroth-order
degradation used to select a reference spectrum based on activity level:
ESP zeroth-order daily average
effective DN range (DN)
Associated Activity
level
Less than 100 0
Between 100 and 336 1
Between 336 and 536 2
Between 536 and 736 3
Between 736 and 936 4
Between 936 and 1136 5
Between 1136 and 1336 6
Between 1336 and 1536 7
Between 1536 and 1736 8
Between 1736 and 1936 9
Greater than 1936 10
88
Figure 8.3 Histogram showing the number of days within each of the 11 activity levels defined
based on ESP daily average zeroth-order effective DN (as shown in Table 8.1). The plot covers the
interval from 30 April 2010 through 26 May 2014. From Didkovsky, Wieman, and Woodraska,
(2014b).
In spite of the differences between the SEM and ESP zeroth-order band passes, the two channels
have shown similar relative variability throughout the SDO mission as shown in Figure 8.4.
Because the SEM zeroth-order channel is sensitive to longer wavelength EUV emissions which do
not exhibit as much solar cycle variability (see e.g. Figure 2.6) as the soft X-ray emissions, it
exhibits a lower relative increase over the rise of solar Cycle 24 than ESP. Nonetheless, a
correlation plot (Figure 8.5) shows that there is a strong quadratic correlation (rcor = 0.964)
89
between the two data sets and that based on a quadratic fit (black line) ESP zeroth-order effective
DN can be reliably mapped to the SEM values.
Figure 8.4 Time series of ESP zeroth-order effective DN (black curve) compared to SEM zeroth-
order effective DN (red curve). Due to differences in band pass and instrument sensitivty the SEM
zeroth-order effective DN values are higher than those from ESP and have been scaled by a factor
of 1:20 for this comparison.
90
Figure 8.5 Correlation plot between SEM and ESP zeroth-order effective DN (blue circles) shows
a strong (rcor = 0.964) quadratic correlation between the data sets
The quadratic relationship between the SEM and ESP zeroth-order effective DN allows the
activity levels used in the ESP data processing algorithm to be determined based on SEM zeroth-
order irradiances. Red lines in figure 8.5 show the mapping of ESP activity level thresholds to
SEM zeroth-order effective DN values. This mapping results in the SEM-based activity levels
shown in Table 8.2
91
Table 8.2: Ranges of SEM zeroth-order daily average effective DN corrected for zeroth-order
degradation used to select a reference spectrum based on activity level:
SEM zeroth-order daily average
effective DN range (DN)
Associated Activity
level
Less than 3067 0
Between 3067 and 6294 1
Between 6294 and 8830 2
Between 8830 and 11184 3
Between 11184 and 13354 4
Between 13354 and 15342 5
Between 15342 and 17147 6
Between 17147 and 18769 7
Between 18769 and 20209 8
Between 20209 and 21466 9
Greater than 21466 10
In Figure 8.6 SEM irradiances determined by adopting the MEGS based reference spectra
used in the ESP data processing algorithm according to the activity levels specified in Table 8.2
are compared to concurrent EVE irradiances. This approach results in irradiances that are in much
better agreement with the EVE measurements than the Version 3.1 irradiances. Average
differences of ~1% and ~4% are obtained relative to ESP and MEGS respectively. With this
approach the adopted reference spectrum is fixed over time intervals for which the SEM zeroth-
order effective DN remain within the same activity level resulting in the discrete steps evident in
the ratio plot with Version 3.1 (Figure 8.6 b).
92
Figure 8.6 a) ESP 26-34 nm irradiances (blue line) determined using a system of 11 reference
spectra established for processing EVE ESP data following the EVE MEGS-A anomaly compared
to SEM Ver 3.1 (dotted line), EVE MEGS (black line), and EVE ESP (gray line). Irradiance ratios
for the FISM based SEM irradiances over SEM Ver 3.1, EVE ESP, and EVE MEGS are shown in
panels b, c, and d respectively with mean ratios shown as dashed lines in each panel.
93
8.4 Spectra based on SOHO CDS measurements and DEM modeling
Though there are very few solar EUV spectral irradaince measurements during the earlier
part of the SOHO mission, particularly around the maximum of solar cycle 23, solar EUV spectra
have been generated based on spectral radiance measurements from the SOHO/CDS (Del Zanna et
al., 2001; Thompson et al., 2002). The Irradiance measurements are obtained using the CDS
Normal Incidence Spectrometer (NIS) channel which includes band passes from 30.8-38.1 nm and
from 51.3-63.3 nm (a complete description of the CDS instrument and its science channels can be
found in Harrison et al., 1995). The longer wavelength band pass provides a measurement of the
He-II 30.4 nm line diffracted in the second-order, and wavelengths shorter than 30.4 can be
determined using a DEM-based technique described briefly below.
When EUV/SXR spectral irradiance measurements are available, but only in a limited
spectral range, they can be used to evaluate DEMs which can in turn be used to calculate
irradiance in a different or broader spectral range. This approach is based on the following
formulation describing the intensity, I(λtr), of emissions for a given transition within a plasma
along a given line of sight (after Warren et al., 1998):
𝐼 (𝜆 𝑡𝑟
) =
1
4𝜋 ∫ 𝑛 𝑢𝑙
𝐴 𝑡𝑟
𝑠 ℎ𝑐 𝜆 𝑡𝑟
𝑑𝑠 (8.1)
where, λtr is the emission wavelength related to the transition, nul is the number density of the
species in the upper energy state of the transition, Atr is the rate at which species in the upper
energy state drop to another state that is lower in energy by the amount of the emitted photon equal
to hc /λtr. The value for nul is dependent both on atomic physics parameters and on density and
94
temperature profiles in the Suns upper atmosphere. The nul term can be expanded however to
express equation 8.1 in a form for which the terms that are related to the quantum mechanical
properties of the emitting species are isolated from those that are related to conditions on the Sun.
For this purpose nul is expressed in terms of relative ion, element, hydrogen, and electron number
densities:
𝑛 𝑢𝑙
=
𝑛 𝑢𝑙
𝑛 𝑖𝑜𝑛 𝑛 𝑖𝑜𝑛 𝑛 𝑒𝑙𝑚𝑡 𝑛 𝑒𝑙𝑚𝑡 𝑛 𝐻 𝑛 𝐻 𝑛 𝑒 𝑛 𝑒 (8.2)
where the first two fractions on the right hand side of the equations are the level population
(nul/nion) and the ionization fraction (nion/nelmt) both of which can be calculated from atomic data
on the relevant ion and transition. The elemental abundance, Ael , relative to hydrogen (nelmt/nH) in
the Sun’s upper atmosphere is available from a number of sources including Feldman et al.,
(2000), and Grevesse et al., (2007). The ratio of hydrogen over electron number densities (nH/ne)
can be approximated based on the relative abundances of He and H assuming that at coronal and
transition region temperatures these elements are completely ionized (and that this ionization is the
main source of electrons with relatively few coming from other less abundant species). Defining a
parameter called the “contribution function,” Gtr(T), which incorporates all of the known atomic
data for the relevant transition as (after Warren et al., 1998):
𝐺 𝑡𝑟
(𝑇 ) ≡ 𝐴 𝑒𝑙𝑚𝑡 𝐴 𝑡𝑟
4𝜋 𝑛 𝐻 𝑛 𝑒 𝑛 𝑢𝑙
𝑛 𝑖𝑜𝑛 𝑛 𝑖𝑜𝑛 𝑛 𝑒𝑙𝑚𝑡 1
𝑛 𝑒 (8.3)
and defining a second parameter, ξ(T), which includes information related to the temperature and
density profiles of the solar atmosphere as:
95
𝜉 (𝑇 ) ≡ 𝑛 𝑒 2
𝑑𝑠
𝑑𝑇
(8.4)
allows equation 8.1 to be expressed as an integral over temperature where the integrand is the
product of these two defined parameters:
𝐼 (𝜆 𝑡𝑟
) = ∫ 𝐺 𝑡𝑟
(𝑇 )𝜉 (𝑇 ) 𝑑𝑇
𝑇 (8.5)
The latter parameter, defined in 8.4, is the quantity known as the DEM. It can be determined from
measurements of emission intensity, I(λtr), by inverting equation 8.5, if the atomic data comprising
the contribution function, Gtr(T), for the relevant ion and transition are known. The atomic data for
transitions relevant to most solar EUV/SXR emissions are available in the CHIANTI atomic
database (Dere et al. 1997, Landi et al. 2013) which is widely used for this purpose.
Using this technique, the spectra established from the CDS measurements cover the entire
range of sensitivity for the SEM first-order channel and can thus be used in the SEM irradiance
calculation. However, CDS irradiance spectra are not available continuously with high time
cadence like the EVE data or even on a daily basis like the FISM and Solar2000 modeled spectra.
The CDS NIS does not have a wide enough field of view to measure the entire solar disk in one
exposure. Irradiance measurements are accomplished using a CDS internal steering mirror to
perform a raster scan over the entire Sun. Scans take about 13.5 hours to complete and are
performed about once per month. Furthermore, the full-disk scans were not a standard part of the
CDS observing program prior to 1998, and for various reasons such as missing exposures within
the scan, or spectral changes due to flare eruption during the scan, not all of the monthly scans
yield a spectrum (Del Zanna and Andretta, 2011).
96
8.5 Inter-comparison of FISM, Solar2000, ESP algorithm, and CDS spectra
The irregularity with which the CDS spectra are available limits their utility as reference
spectra for recalculating the entire SEM data set; however, they can be used for inter-comparison
with the other reference spectra discussed in sections 8.1 through 8.3. CDS spectra for two dates,
30 October 2001 near the maximum of solar cycle 23, and 31 May 2010, near the solar cycle 23/24
minimum were selected for this comparison and provided by Del Zanna (2014, personal
communication). Spectra from FISM, Solar2000, and the ESP data processing algorithm are
compared to the CDS based spectra and to SOLERS22 in Figure 8.7 covering the spectral range
affecting irradiance values determined for the SEM 26-34 nm band. EVE MEGS-A was
operational during the latter date and is thus included in the comparison of solar minimum spectra.
Discrete spectra adapted from the EVE ESP data processing algorithm (Section 8.3) include the
highest (level 10) and lowest (level 0) activity spectra based on Table 8.2 and SEM zeroth-order
effective DN of 24882.9 for 30 October 2001 and 2442.1 for 31 May 2010. These spectra are
labeled ESP level 10 and ESP level 0 in the corresponding comparison plots in Figure 8.7.
97
Figure 8.7 – Comparison of spectra from FISM, Solar2000, and the ESP data processing algorithm
with CDS-based spectra and SOLERS for a day near the maximum (MAX) of solar cycle 23 (30
October 2001) and a day near the minimum (MIN) of solar cycles 23 and 24 (31 May 2010). The
comparison includes the spectral range affecting the SEM 26-34 nm irradiance calculations. For
solar MAX, the FISM spectrum provides the closest match to the CDS spectrum. For solar MIN,
FISM, ESP level 0, and EVE MEGS-A spectra are all in good agreement with CDS, however some
advantages in using FISM are identified in the text below.
98
For 30 October 2001 during the solar cycle 23 maximum, FISM (red curve) provides the
closest match to the CDS spectra (black curve) out of the three sources of reference spectra
discussed in the preceding sections based on the plot in the top panel of Figure 8.7. The ESP level
10 fluxes for the solar MAX date are uniformly lower than CDS (and all other spectra with the
exception of SOLERS22), which is likely because this spectrum is based on the highest level of
activity observed by EVE MEGS-A during solar cycle 24 which has been notably weaker than
solar cycle 23. The Solar 2000 fluxes are consistently higher on this day with the exception of the
He-II 30.4 nm peak.
For the solar MIN date compared in the lower panel of Figure 8.7, the FISM, ESP level 0,
and EVE MEGS-A spectra are in good agreement with CDS. As with the solar MAX date, the
Solar 2000 fluxes are characteristically higher across the spectral range shown with the exception
of the He-II 30.4 nm peak. However, as discussed in section 5.3.2, the SEM data processing
algorithm depends on the relative spectral distribution of the reference spectrum, not on the
absolute flux values, so the uniformly higher fluxes associated with the Solar 2000 spectra will not
necessarily result in significantly different SEM irradiance values.
One characteristic of the compared spectra that does significantly influence the SEM
26-34 nm irradiance values is the total flux within the primary 26-34 nm band versus the flux that
is outside of this band yet still contributes to SEM signal, particularly in the range from 15 to 25
nm for which the SEM was found, based on the 2007 NIST calibrations, to have greater sensitivity
than originally expected. In this regard, the SOLERS22 spectrum for which fluxes are concentrated
99
within the 26-34 nm band and relatively low in the 15-25 nm band, deviates from the other spectra,
which explains in part why the SOLERS22-based SEM Version 3.1 irradiance values are
systematically high.
For the two dates included in the comparison SEM irradiances calculated based on each of
the compared reference spectra are shown in Table 8.3. All of the tabulated irradiance values were
determined based on the updated response function from the 2007 NIST calibrations, thus the
SOLERS22-based irradiances are not the same as SEM Version 3.1 irradiances which use the SEM
response function determined before the launch of SOHO. SEM irradiances based on SOLERS22
are notably higher than irradiances based on the other reference spectra, consistent with the
distinguishing differences in the SOLERS22 spectrum mentioned above. Relatively good
agreement is observed among SEM irradiances determined with the other reference spectra, so
while this comparison supports earlier assertions that use of SOLERS22 results in systematically
high irradiances, it does not demonstrate that SEM irradiance values are strongly dependent on
which of the time-dependent reference spectra (i.e., FISM, Solar2000, or ESP algorithm) is used
for processing the SEM data.
100
Table 8.3 SEM 26-34 nm irradiances for a 30 October 2001 (solar MAX) and 31 May 2010 (solar
MIN) determined based on reference spectra from SOHO CDS, FISM, Solar 2000, the ESP data
algorithm, EVE MEGS (for 2010 only), and SOLERS22.
Reference
Spectrum
30 October 2001 (solar MAX) 31 May 2010 (solar MIN)
26-34 nm
Irradiance (W/m
2
)
Ratio (over ESP
algorithm)
26-34 nm
Irradiance (W/m
2
)
Ratio (over ESP
algorithm)
ESP algorithm 1.66E-03 0.98 5.58E-04 1.04
CDS 1.69E-03 1.00 5.38E-04 1.00
FISM 1.72E-03 1.02 5.77E-04 1.07
Solar2000 1.74E-03 1.03 5.66E-04 1.05
EVE MEGS N/A N/A 5.55E-04 1.03
SOLERS22 1.83E-03 1.08 6.27E-04 1.17
Figure 8.8 shows a comparison of the SEM time series throughout the SOHO mission
determined using FISM (solid green line), Solar 2000 (dashed red line), and the ESP algorithm
(dotted blue line) reference spectra. Due to the good agreement of SEM irradiances determined
using the three time-dependent reference spectra, the FISM, Solar 2000, and ESP algorithm curves
are virtually indistinguishable. For a given day the standard deviation of the three SEM values
(determined based on the three different time-dependent reference spectra) are less than about 2%
during solar minimum and less than about 4% during solar maximum. This comparison indicates
the SEM irradiance values do not have a strong dependence on which of the three time-dependent
reference spectra is used, and suggests that similar results will be obtained regardless of which is
selected for reprocessing the pre-SDO SEM data. The time series plotted in figure 8.8 do not yet
provide fully re-calibrated SEM irradiance values (one of the stated objectives of this work) as
they are still based on the degradation model parameters established for Version 3.1 from sounding
101
rocket measurements that were processed with the SEM response functions that were measured
pre-launch, and SOLERS22 reference spectra. While the use of the Version 3.1 degradation model
does not affect the comparison of reference spectra, it may result in a long-term trend in the time
series.
Figure 8.8 – Comparison of SEM 26-34 nm irradiance values determined based on reference
spectra from FISM (solid green line), Solar2000 (dashed red line), and the ESP algorithm (dotted
blue line). All three time series are in good agreement and virtually indistinguishable in the plot
suggesting that the SEM irradiance values do not depend strongly on which of the three time-
dependent reference spectra is used.
102
Although the above comparisons do not show a clear advantage in using one of the time-
dependent reference spectra over the others, they provide some indication that FISM may be the
most reliable of the three. First, the largest discrepancies among daily SEM irradiances determined
with each of the different spectra occur around solar maximum (standard deviations are ~4%
compared to ~2% near solar minimum), and based on the solar maximum comparison of Table 8.3,
the SEM irradiance determined with FISM is the median value of the three. Second, although SEM
irradiances determined using Solar2000 are in reasonable agreement with those from FISM and the
ESP algorithm spectra, the Solar2000 spectral irradiances appear discrepantly high relative to all
other spectra for both the solar minimum and solar maximum days compared in figure 8.7. Third,
as mentioned above, the ESP algorithm reference spectra are based on spectra measured with EVE
MEGS A during solar cycle 24, which has been notably weaker than solar cycle 23, and thus the
included spectra may not accurately represent the full range of activity (particularly the higher
levels of activity) the Sun exhibited during solar cycle 23. For these reasons, the FISM reference
spectra are used in the following chapter to determine SEM irradiances for the entire SOHO
mission.
103
9. Re-processing the SEM data throughout the SOHO Mission
In this chapter the time-dependent FISM reference spectra selected based on the
comparisons of Chapter 8 are used to reprocess the SEM data for the entire SOHO mission. As a
first step, the raw data from the sounding rocket calibration flights must be re-analyzed using
FISM spectra corresponding to the days of the rocket flights. New parameters for the SEM
degradation model will then be determined to minimize differences between the revised sounding
rocket irradiance values and concurrent irradiance values from the on-orbit SEM. For verification
of the assumed mode of SEM degradation (i.e., reduced transmission of the SEM thin-film filter
with time) the SEM degradation model is compared to degradation of the EVE ESP instrument
which includes features for monitoring degradation in flight. SEM raw data throughout the SOHO
mission are re-processed based on FISM reference spectra, the updated first-order response
functions, and the degradation model with updated parameters. The resulting SEM time series is
compared with other available EUV irradiance values from SDO EVE and with solar indices.
9.1 Reprocessing the SEM sounding rocket data and degradation model
Figure 9.1 shows the SEM 26-34 nm Version 3.1 data throughout the SEM mission along
with SEM-clone, RGIC, and EVE ESP ch9 sounding rocket photon flux values. This time series
and corresponding set of sounding rocket flux values is the same as that shown in the bottom curve
of Figure 1.1 but also includes a comparison with fluxes from the in-flight EVE ESP ch9 for the
interval overlapping the SDO mission. Parameters for the Version 3.1 degradation model (i.e.
parameters a, b, and c from equation 5.1) were determined to minimize residual differences
between the sounding rocket fluxes and concurrent fluxes from the on-orbit SEM. Version 3.1, was
104
released in 2008 and thus only includes sounding rocket measurements up to November 2006 in
this minimization. Converting the raw sounding rocket data to photon flux values was done using
the SOLERS22 reference spectrum both for processing the SEM-clone data and for adjusting the
RGIC values from their native 5-57 nm band pass to the 26-34 nm band pass for
comparison/fitting the SEM on orbit data. Thus, as is evident in Figure 9.1, the SEM on orbit
measurements are nominally in good agreement with the SEM-clone and RGIC measurements but
are systematically higher than the EVE/ESP measurements for the reasons identified in Chapter 7.
Figure 9.1 SOHO/SEM Version 3.1 photon flux values (black dots) are in nominally good
agreement with sounding rocket irradiance values determined from the SEM-clone (grey squares)
and the RGIC (gray diamonds) raw data using the Version 3.1 SEM response function and
SOLERS22 reference spectrum, but are systematically higher than the EVE/ESP on orbit (green
dots) and sounding rocket (grey circles) flux values.
105
Reprocessing the SEM-clone data using the updated response profile and the FISM
reference spectra presented in chapter 8, should nominally result in little change to the degradation
model, since the shift in calculated irradiance due to these data processing parameters should affect
both the on-orbit and sounding rocket measurements equally, since the two instruments have
similar response functions. However, the RGIC does not have the same response function, and its
measurements will thus be affected differently by the change in reference spectrum. Additionally,
while the change in SEM values is related to updates in both the response function and reference
spectrum, the change in RGIC values is associated with only the updated reference spectrum. Thus
recalculating the sounding rocket measurements does result in some difference in their distribution
with time and will influence the degradation model based upon them.
SEM-clone and RGIC sounding rocket measurements re-processed using FISM reference
spectra are presented (and compared to the Version 3.1 values from table 5.1) in Table 9.1. The re-
processed SEM values are based on the updated first-order response functions from the 2007 NIST
calibration, while the Version 3.1 values are based on the pre-launch response functions.
106
Table 9.1 – SEM clone and RGIC sounding rocket measurements recalculated using the FISM
time dependent reference spectrum (SEM-clone values are calculated using the updated response
function described in section 5.3.1)
Date
Sounding Rocket
Flight number
Calibration
Instrument
26-34 nm flux
reprocessed
(ph/cm2/s)
26-34 nm flux
Version 3.1
(ph/cm2/s)
Ratio:
Reprocessed/
Version 3.1
26 June 1996 36.147
SEM clone 1.13E+10 1.32E+10 0.86
RGIC 9.71E+09 1.15E+10 0.84
11 August
1997
36.164
SEM clone 1.12E+10 1.28E+10 0.87
RGIC 1.17E+10 1.36E+10 0.86
18 August
1999
36.181
SEM clone 1.76E+10 2.09E+10 0.84
RGIC 1.90E+10 2.24E+10 0.85
6 August
2002
36.202
SEM clone 2.25E+10 2.29E+10 0.98
RGIC 2.21E+10 2.43E+10 0.91
5 December
2003
36.211
SEM clone 1.59E+10 1.75E+10 0.91
RGIC 1.50E+10 1.67E+10 0.90
3 August
2005
36.227
SEM clone 1.43E+10 1.53E+10 0.94
RGIC 1.37E+10 1.52E+10 0.90
7 November
2006
36.236
SEM clone 1.11E+10 1.20E+10 0.92
RGIC 1.05E+10 1.22E+10 0.86
24 July 2012 36.263
SEM clone 1.22E+10 1.51E+10 0.87
RGIC 1.36E+10 1.51E+10 0.90
Based on the above reprocessed sounding rocket measurements (as well as measurements
from the ESP clone instrument on the EVE calibration rocket flights which were not used in
determining the Version 3.1 degradation model) degradation model parameters can be recalculated
to minimize residual differences between SEM on-orbit measurements processed using the updated
response function and FISM reference spectra. Minimization of the parameters a, b, and c is
accomplished by finding values that result in a minimum value for the following objective
function:
107
𝑆𝐸𝑀 𝑜𝑏𝑗 (𝑎 , 𝑏 , 𝑐 ) = ∑ [𝑎𝑏𝑠 (𝑆𝑂𝐻𝑂𝑆𝐸𝑀 𝑖 (𝑎 , 𝑏 , 𝑐 ) − 𝑆𝑅𝑚𝑒𝑎𝑠 𝑖 )]
20
𝑖 =1
(9.2)
where SRmeasi, are the sounding rocket irradiance measurements, 20 in all from 12 different
rocket flights (subscript i denotes the measurement number not the flight number), including eight
measurements each from the SEM-clone and the RGIC (shown in table 9.1) and the additional four
measurements from the EVE ESP clone (shown in Table 5.1). SOHOSEMi is the concurrent
irradiance value from the on-orbit SEM which is dependent on the degradation factor defined in
terms of a, b, and c. Coefficients a, b, and c are used in the expression (equation 5.1) for the
thickness of a carbon contaminant layer, while the actual degradation factor (equation 5.2) is based
also on the wavelength dependent absorption cross section of the contaminating material (modeled
as carbon). Because of this complicated dependence of SEMobj on a, b, and c, the objective
function is non-differentiable and an iterative approach is required to obtain the values of a, b, and
c that minimize it. For this purpose the downhill simplex method of Nelder and Mead (1965) is
applied using the IDL “amoeba” function. The resulting carbon thickness model is compared to the
Version 3.1 model in Figure 9.2
108
Figure 9.2 – A comparison of modeled contaminant layer growth based on sounding rocket
measurements processed with the SOLERS22 reference spectrum (red curve) as used in
Version 3.1 of the SEM data and processed using the FISM reference spectrum (blue curve) and
the updated SEM response function (for the SEM clone measurements). The updated model
incorporates an additional sounding rocket flight (36.263 on July 24, 2012) that was not included
in the determination of the Version 3.1 degradation curve.
9.2 Comparison to EVE/ESP measured degradation
For verification of the assumed mode of SEM degradation, the SEM degradation model is
compared to degradation of the EVE ESP instrument determined based on ESP in-flight calibration
measurements. The EVE/ESP instrument has several features for in-flight monitoring of
degradation. In particular it includes a filter wheel with redundant filters, nominally identical to the
primary observing filter, which are exposed to the Sun for only ~1 min per day. Because of this
109
limited exposure there is little opportunity for contaminants to be polymerized on the redundant
filters and they suffer no evident degradation. During daily in-flight calibrations ESP observations
are made with one of the redundant filters and the signal is compared to that obtained with the
primary filter to assess any signal loss due to contamination of the primary filter. This approach
has been quite effective for monitoring the degradation of ESP as verified by EVE sounding rocket
calibration measurements (see for example, Didkovsky and Wieman, 2013) and so the
determination of a carbon growth model like that used for SEM has not been necessary. It is
possible to construct such a model however as a means of comparing the degradation of EVE/ESP
to that of SOHO/SEM since the two instruments are very similar in design and are likely to
degrade in a similar fashion.
The loss of signal due to degradation of the ESP primary observing filter is shown in
Figure 9.3 which shows signal ratios (redundant over primary) for each of the daily calibrations.
ESP observes several narrow wavelength bands as indicated in the figure legend. For a given day,
the wavelength dependent signal loss in the primary filter can be modeled as a transmission loss
through a carbon contaminant layer based on Henke absorption models (Henke et al., 1993).
110
Figure 9.3 – Ratios of signal observed using an ESP redundant filter over that observed with the
primary filter provides a measurement of signal loss due to contamination of the primary filter.
Such measurements are taken as part of the EVE/ESP daily in-flight calibration.
Figure 9.4 shows a comparison of contaminant layer transmission for channels 2, 8, 9 and
QD (at 26, 18, 30, and 2 nm respectively) for the ESP primary filter after 1100 days of operation to
the Henke model for a 33 nm thick C layer. The growth rate of the carbon layer can be determined
111
by fitting the Henke model for given thicknesses to the measured signal loss for corresponding
days throughout the mission. Figure 9.5 shows the modeled C layer thickness as function of time
since April 30, 2010. The function describing carbon layer thickness versus time for the SEM
degradation model (equation 5.1) fits the observed carbon growth rate on the EVE/ESP primary
filter, with the appropriate choice of parameters a, b, and c. It is evident from comparing figures
9.2 and 9.5 that the contaminant deposition on the ESP primary filter is occurring faster than what
has been modeled for SEM based on either the Version 3.1 rocket irradiance values or the updated
values described in the preceding section, but is of the same general order. Because the
contamination source (due to either outgassing from other components on the spacecraft to
deposition of propellant during orbital maneuvers) and degree of exposure to the source is not
likely to be the same for the EVE/ESP and SOHO/SEM, it should not be expected that the rate of
contaminant build up on their respective filters would be the same.
112
Figure 9.4 - Transmission of contaminant layer built up on the ESP primary filter based on filter
wheel degradation measurements (red squares) after 1100 days of operation compared to Henke
model transmission for a 33 nm thick layer of C.
113
Figure 9.5 - ESP filter #3 contaminant layer thickness versus time. The contaminant layer
continues to grow at an exponentially decreasing rate.
9.3 Re-processed SEM irradiance time series for the entire SOHO mission
The SEM data recalculated for the entire SOHO mission using the FISM reference
spectrum and the updated response function and degradation model is shown in Figure 9.6. For this
updated data set, the SEM on-orbit data are in good agreement with the SDO/EVE/ESP on-orbit
and sounding rocket measurements (black and green dots nearly coincide for the period of overlap
with SDO) and with the reprocessed SEM clone and RGIC measurements. The degradation model
0
5
10
15
20
25
30
35
0 200 400 600 800 1000 1200
Equivalent CH layer thickness (nm)
ESP day of operation
Thicknesses from ESP filter wheel based
degradation measurements
exp fit
114
based on the reprocessed sounding rocket data (Figure 9.2) suggests that the accumulated
contaminant layer for the later part of the SEM mission is thicker than that used in the Version 3.1
data. The effect of this difference is evident in the lower inter-minima change in the updated SEM
time series of only about 12% compared to about 15% reported based on Version 3.1 data in
Didkovsky et al. 2010b.
Figure 9.6 – SEM time series recalculated using the FISM reference spectrum and the updated
response function and degradation model. The time series is in good agreement with both the
reprocessed sounding rocket data from the SEM-clone (grey squares) and the RGIC (gray
diamonds), and the EVE/ESP on-orbit (green dots) and sounding rocket measurements (grey
circles) in contrast to the Ver. 3.1 data shown in Figure 9.1. The inter-minima change in photon
flux is slightly lower for the recalculated time series (only ~12% compared to ~15% according to
the Version 3.1 data).
115
The determined relative inter-minima difference of 12% is based on the degradation model
three parameter fit of the SEM data to the sounding rocket measurements. Thus the uncertainty of
this value is determined, in a manner analogous to determining the slope uncertainty, σm, for a
least squares linear fit, based on the standard deviation, σyx, of the sounding rocket flux values
from the SOHO/SEM flux values and the sum of the squares, SSx, of the differences about the
mean of the sounding rocket date according to:
𝜎 𝑚 =
𝜎 𝑦𝑥
√𝑆𝑆
𝑥 =
√
1
𝑛 −3
∑ (𝜙 𝑆𝑂𝐻 𝑂 𝑆𝐸𝑀𝑖 −𝜙 𝑟𝑜𝑐𝑘𝑒𝑡𝑖 )
2
𝑛 𝑖 =1
√∑ (𝐷𝑎𝑡𝑒 𝑆𝑅𝑖 −𝐷𝑎𝑡𝑒 𝑚𝑒𝑎𝑛 )
2
𝑛 𝑖 =1
(9.3)
where n is the number of sounding rocket measurements (n-3 is the least squares degrees
of freedom equal to n data points minus 3 fit parameters), ϕrocketi is the i
th
sounding rocket flux
measurement, ϕSOHOSEMi is the SOHO/SEM flux measured at the time of the i
th
sounding rocket
measurement, DateSRi is the i
th
sounding rocket flight date, and Datemean is the mean of the flight
dates.
For the sounding rocket flights used to determine the fit parameters for the new degradation
model, σm has a value of about 6.54×10
4
Photons cm
-2
sec
-1
day
-1
or with about 5000 days between
minima, the uncertainty is about 3.27×10
8
Photons cm
-2
sec
-1
or about 4% of the SOHO SEM solar
minimum flux value. Thus an uncertainty of ±4% is specified for the determined 12% decrease in
minimum 26-34 nm flux values for the solar cycles 23/24 minimum compared to the cycles 22/23
minimum.
116
Figure 9.7 Top panel: A comparison of SOHO/SEM updated 26-34 nm irradiances to irradiances
in the same band from the ESP and MEGS channels of the SDO/EVE instrument. Middle panel:
daily ratio of SEM over MEGS (blue curve) plotted with dashed lines showing the mean ratio
(1.0013) and ratio STD (0.048). Bottom panel: daily ratio of SEM over ESP (green curve) plotted
with dashed lines showing the mean ratio (0.9995) and ratio STD (0.0587).
Figure 9.7 shows the SEM comparison with ESP for the period of overlap with SDO (a
closer view of the right hand portion of Figure 9.6) and includes the EVE MEGS spectra integrated
from 26-34 nm. For this comparison the SEM Photon flux values have been converted to
Irradiance values to match the units of the EVE data. The top panel shows SEM to be in good
117
agreement with both EVE MEGS and EVE ESP throughout the SDO mission. Ratios for SEM
over MEGS (blue curve, middle panel) and SEM over ESP (green curve, bottom panel) show that
the daily SEM values match those of both ESP and MEGS within about 5% (one sigma standard
deviation) throughout the SDO mission. Furthermore, the mean ratio values very close to unity
demonstrate excellent agreement in absolute irradiance levels.
9.4 SEM comparison with solar indices
As a means of evaluating the revised degradation model, the SEM time series is compared
to the F10.7 and Mg-II solar activity indices discussed in Section 2.4. Because the F10.7 radio and
Mg-II emissions originate from different source regions within the solar atmosphere and the
indices quantify different solar phenomena (Tapping, 2013), these comparisons are expected to
show some differences. However, the F10.7 measurements are obtained with ground-based
instrumentation which can be more readily calibrated than instruments in space, and the Mg-II
index measures relative change within a spectral absorption feature rather than absolute intensity,
thus the indices are less susceptible to long-term drift related to instrument degradation. Therefore,
if the SEM degradation has not been accurately modeled and corrected, it should show up in these
comparisons as a long-term trend relative to the indices.
118
Figure 9.8 – Comparison of SEM 26-34 nm time series with the F10.7 index (F10.7 daily and 81
day running mean values are averaged and linear fit to the SEM irradiances). The well-documented
lower sensitivity of F10.7 to variations around solar minimum is evident as are differences during
other intervals (e.g. around 2001 and between 2012 and 2013), however with about the same
signals for the peaks between 2011 and 2012. These time intervals are isolated and not indicative
of a steady long-term relative trend between time series
In figure 9.8 the SEM daily irradiance values are compared to the average of the daily and
81-day running mean F10.7 values (as used in similar comparisons by Viereck, et al., 2001). A
linear fit has been applied to the F10.7 values to match the scale of the SEM data. Although the
F10.7 values closely match SEM throughout the declining phase of solar cycle 23 and the rising
phase of solar cycle 24, they level off and remain above the SEM values which continue to vary
through the solar minimum itself, highlighting the well-documented (e.g., Tapping and DeTracey,
119
1990; Woods et al., 2000; Tobiska, 1988; Viereck, et al., 2001; Solomon et al., 2013) lower
variability of F10.7 emissions compared to EUV emissions around solar minimum. Apart from
other shorter-term differences over isolated intervals (e.g. around 1998, 2001, and 2012) the two
data sets are in good agreement throughout the SOHO mission. This agreement, particularly at the
beginning and the end of the time series, suggests there is no significant trend in the SEM
irradiances relative to the F10.7 values and thus that SEM degradation has been appropriately
corrected. This plot thus supports the assertion that lower EUV emissions for the solar cycle 23/24
minimum relative to the cycle 22/23 minimum observed by SEM are real, not instrumental, and
that the known deterioration of the correlation between EUV and 10.7 cm radio emissions around
solar minimum can explain why a similar inter-minima difference is not evident in the F10.7 data.
120
Figure 9.9 – Comparisons of SEM 26-34 nm time series with the daily Mg-II core to wing ratio
linear fit to the SEM data. The Mg-II index shows larger solar rotation variability and differences
are evident at isolated intervals, but the good agreement over much of the time series (in the
beginning and end in particular) does not suggest a steady long-term relative trend. The Mg-II
index shows signal levels similar to SEM for the two latest solar minima (e.g. about 10% decrease
for the 23/24 minimum as reported by Solomon, S.C. and Qian, L.: 2011) and similar variability
following the 2008 minimum, e.g. for the peaks between 2011 and 2012.
In figure 9.9 SEM daily irradiances are compared to the Mg-II core-to-wing ratio values
(from Space Environment Technologies www.spacewx.com) with a linear fit applied to match the
scale of the SEM data. Like the comparison with F10.7, the SEM data are, with the exception of a
few isolated intervals, in good agreement with the Mg-II, particularly at the beginning and end of
the time series, providing further validation of the SEM degradation model. Unlike, F10.7, the Mg-
II index remains in good agreement with SEM throughout the solar cycle 23/24 minimum and
121
shows a similar reduction compared to the previous minimum (e.g. about 10% decrease for the
23/24 minimum reported by Solomon, S.C. and Qian, L.: 2011), indicating further that the
observed inter-minima differences are real.
122
10. Calculating SEM equivalent irradiances based on EVE
measurements
SEM Ver 3.1 irradiance values based on the pre-flight response function and SOLERS22
reference spectrum have been in use for a number of applications including as a solar irradiance
index in thermospheric density modeling (Bowman et al., 2008; Tobiska et al., 2006), and for
inter-comparison with other EUB instrumentation (Haberreiter et al., 2014, Machol et al., 2013). It
is desirable to continue providing the SEM such applications from the standpoint that (particularly
for space weather operations) continuity of a data set is sometimes more important than absolute
accuracy (Tobiska, 2012). Based on the improved understanding of the SEM degradation,
sensitivity to reference spectrum, and spectral response, that have been gained in part from this
work it is possible to reconstruct SEM Version 3.1 irradiance values based on ESP channel 9
measurements. Such a reconstruction would be of value to continue the SEM Version 3.1 data set
using the newer ESP instrumentation if the SEM (in its 20th year at the time of this writing) were
to be decommissioned.
The reconstruction is based on the good agreement (Figure 9.4) between the updated SEM
irradiances, ΦʹSEM1, determined as described in chapter 9, and 26-34 nm irradiance values
determined from EVE ESP, Φ ESPch9 , according to the following formulation extending from
equation 5.3:
123
Φʹ
𝑆𝐸𝑀 1
= 𝑘 𝐹𝐼𝑆𝑀 𝐷𝑁
𝑆𝐸𝑀𝑐 ℎ1
−𝐼 𝑏𝑘𝑔𝑟𝑑 𝐴 ∙∫ 𝜂 ′
1
∙𝜙 𝐹𝐼𝑆𝑀 ∙𝑓 ′
𝑑𝑒𝑔𝑟𝑎𝑑 ∙𝑑𝜆 ∙𝑓 1𝐴𝑈
𝜆 2
𝜆 1
∫ 𝜙 𝐹𝐼𝑆𝑀 𝑑𝜆
𝜆 2
𝜆 1
≈ Φ
𝐸𝑆𝑃𝑐 ℎ9
(10.1)
where ϕFISM is the daily FISM spectrum, kFISM, is the correction for the SEM sensitivity outside of
26-34 nm based on the daily FISM reference spectrum, η ʹ1 is the SEM first-order channel
efficiency from the 2007 NIST calibration, and f ʹdegrad is the time dependent degradation factor
based on the revised degradation model described in section 9.1. The form of equation (5.3) for
determining the Version 3.1 correction factor, k1, based on the SOLERS22 reference spectrum can
be applied to the FISM reference spectrum to determine kFISM:
𝑘 𝐹𝐼𝑆𝑀 =
∫ 𝜙 𝐹𝐼𝑆𝑀 𝑑𝜆
34𝑛𝑚
26𝑛𝑚
∫ 𝜙 𝐹𝐼𝑆𝑀 𝜆 2
𝜆 1
𝑑𝜆
(10.2)
Substituting the right side of (10.2) into (10.1) and solving for the SEM effective DN
values, DNSEMch1 – Ibkgrd, in terms of Φ ESPch9 results in:
DN
SEMch1
-I
bkgrd
≈
Φ
ESPch9
∙A∙ ∫ η'
1
∙ ϕ
FISM
∙ f '
degrad
∙ dλ ∙f
1AU
λ2
λ1
∫ ϕ
FISM
dλ
34
26
(10.3)
Substituting the right side of the equation for DNSEMch1 – Ibkgrd, into Version 3.1 irradiance
equation (5.5) and simplifying gives an approximated (i.e. reconstructed) Version 3.1 irradiance
value that is based on ESP ch9 irradiance rather than the DN value from SEM:
124
Φ
𝑆𝐸𝑀 1
≈
Φ
ESPch9
∙∫ 𝜙 𝑆 22
𝑑𝜆 ∙
34
26
∫ η'
1
∙ ϕ
FISM
∙ f '
degrad
∙ dλ
λ2
λ1
∫ ϕ
FISM
dλ
34
26
∙∫ 𝜂 1
∙𝜙 𝑆 22
∙𝑓 𝑑𝑒𝑔𝑟𝑎𝑑 ∙𝑑𝜆
𝜆 2
𝜆 1
(10.4)
The SEM Version 3.1 time series is compared to the reconstructed time series based on
equation (10.4) in figure 10.1. The reconstruction is in very good agreement with the true SEM
Version 3.1 values, comparable to the level of agreement between the EVE ESP channel and the
revised SEM values compared in Figure 9.4. The mean ratio of the reconstruction over the actual
Version 3.1 values throughout the time series is 1.02 with a ratio standard deviation of 0.03.
125
Figure 10.1 Reconstruction of SEM Version 3.1 irradiance values based on EVE ESP channel 9
which has been found to be in good agreement with the updated SEM irradiance values determined
based on this work.
126
11. Conclusions
A revised calibration of the SOHO SEM 26-34 nm solar irradiance data spanning nearly
two decades has been described. The revised calibration improves upon earlier calibrations of the
SEM data in several ways. First, it incorporates new data related to the SEM instrument response
function and solar spectral distribution within the SEM spectral range of sensitivity, as well as
additional sounding rocket calibration measurements, not available at the time of the earlier
calibrations. Second, it results in SEM irradiance values that are in much better agreement with
more recent reliable irradiance measurements from the EVE instrument suite aboard SDO. Third, it
provides a new estimate of differences in solar irradiance values between the solar cycles 23/24
minimum compared to those of the solar cycles 22/23 solar minimum, and provides further
confirmation that the lower irradiance values during the latter minimum are real, and not an artifact
of uncorrected instrument degradation. Finally, it demonstrates that in the event that SEM is
decommissioned prior to EVE ESP, the long-term 26-34 nm data set established with SEM can be
reliable continued with measurements from the newer ESP instrument without significant
discontinuities due to differences in instrument sensitivity.
Applying the revised, more accurate response function, and time-dependent reference
spectra to the processing of the SEM data reduces 26-34 nm irradiance values by about 20%,
compared to the most recent Version 3.1 release of the SEM data. These lower irradiance values
are shown to be in much better agreement in comparisons with 26-34 nm irradiances based on
measurements from SDO EVE MEGS and SDO EVE ESP. Applying these revised parameters (for
127
times when high time-cadence reference spectra are available from MEGS A) also improves the
correlation between 15sec-cadence SOHO/SEM measurements and simultaneous measurements
from SDO EVE MEGS during large solar flares. Additionally, comparisons of irradiance in the 0.1
– 7 nm band extracted from SOHO/SEM zeroth-order measurements agree with measurements
from SDO EVE channels within about 15 % on average when the revised parameters are used
compared with about 40 % when the Version 3.1 parameters are used. These results are relevant to
efforts to determine accurate irradiances through inter-comparisons among the various EUV
instruments currently in operation.
The SEM daily average 26-34 nm data for the entire SOHO mission dating back to
1 January 1996 have been reprocessed using time-dependent spectra from FISM. While FISM was
selected for this purpose, comparisons show that similar SEM irradiance values are obtained based
on other time-dependent reference spectra (i.e., from the Solar2000 spectral model, the EVE
MEGS system of spectra used for processing ESP irradiances, and discrete spectra based on SOHO
CDS measurements) in contrast to the discrepant values obtained using the fixed SOLERS22
spectrum. Complete reprocessing of the SEM data set required that sounding rocket measurements
also be reprocessed with FISM in place of SOLERS22 and with the updated response function
applied to the SEM-clone measurements. Updated degradation model parameters based on these
revised sounding rocket measurements indicate that SEM has experienced loss of sensitivity/filter
transmission beyond that assumed for Version 3.1.
Accounting for this further degradation in the updated 26-34 nm time series has resulted in
a revised estimate of differences in solar irradiance values between the solar cycles 23/24
128
minimum compared to those of the solar cycles 22/23 solar minimum. Based on the revised time
series, irradiances for the latter minimum were lower by about 12±4% versus 15±6% based on the
Version 3.1 time series. Good absolute agreement with EVE measurements throughout the SDO
mission and good relative agreement with the F10.7 and Mg-II indices throughout the SOHO
mission suggest that SEM degradation has been appropriately corrected. The comparison with
F10.7 further illustrates that, considering the known differences in the variability of EUV
emissions compared to 10.7 cm radio emissions around solar minimum, the lower SEM irradiance
levels for the latter minimum are not inconsistent with the similar levels observed in F10.7. This
result is relevant to current discussions over this inter-minima change which is evident based on
some indicators (e.g., thermospheric density, Mg-II index) but not others (e.g., global mode
ionosphere total electron content, F10.7 index).
In addition to the good agreement achieved between the updated SEM irradiance values
and those from EVE, it has been shown that the SEM Version 3.1 values can be reliably
reproduced based on EVE ESP measurements, which is of potential value in current applications
of the SEM data that require consistency over absolute accuracy. These results demonstrate how
the improved understanding of the SEM degradation, spectral response and dependence on
reference spectra gained from this work will benefit efforts to continue the longstanding EUV
record established by the SEM using newer EUV instrumentation.
129
References
Acton, L. W., Weston, D. C., Bruner, M. E.: 1999, Deriving Solar X Ray Irradiance from Yohkoh
Observations. J. Geophys. Res. 104, A7, 14827.
Allen, C. W.: 1948, Critical frequencies, sunspots, and the Sun's ultra-violet radiation, Terr. Magn.
Atmos. Electr., 53(4), 433
Afraimovich, E. L., E. I. Astafyeva, A. V. Oinats, Yu. V. Yasukevich, and I. V. Zhivetiev: 2008,
Global electron content: a new conception to track solar activity. Annales Geophysicae, 26,
335
Anderson, D., and Fuller-Rowell, T., 1999, The Ionosphere: Space Environment Topics SE-14,
NOAA, Space Weather Prediction Center, http://www.swpc.noaa.gov/info/Iono.pdf
Bailey, S. M., T. N. Woods, C. A. Barth, S. C. Solomon, L. R. Canfield, and R. Korde (2000),
Measurements of the solar soft X-ray irradiance from the Student Nitric Oxide Explorer: first
analysis and underflight calibrations, J. Geophys. Res., 105, 27,179
Bates, D.R.: 1959, Some Problems concerning the Terrestrial Atmosphere above about the 100 km
Level, Proc. R. Soc. Lond., 253, 451
BenMoussa, A., S. Gissot, U. Schühle, G. Del Zanna, F. Auchère, S. Mekaoui, A.R. Jones, D.
Walton, C.J. Eyles, G. Thuillier, D. Seaton, I. E. Dammasch, G. Cessateur, M. Meftah, V.
Andretta, D. Berghmans, D. Bewsher, D. Bolsée, L. Bradley, D. S. Brown, P. C. Chamberlin,
S. Dewitte, L. V. Didkovsky, M. Dominique, F. G. Eparvier, T. Foujols, D. Gillotay, B.
Giordanengo, J.-P. Halain, R. A. Hock, A. Irbah, C. Jeppesen, D. L. Judge, M. Kretzschmar,
D. R. McMullin, B. Nicula, W. Schmutz, G. Ucker, S. Wieman, D. Woodraska, T. N.
Woods, 2013, On-orbit degradation of Solar Instruments, Solar Phys, 288(1):389-434. DOI:
10.1007/s11207-013-0290-z
Boerner, P., Christopher Edwards, James Lemen, Adam Rausch, Carolus Schrijver, Richard Shine,
Lawrence Shing, Robert Stern, Theodore Tarbell, et al.: 2010, Initial Calibration of the
Atmospheric Imaging Assembly (AIA) on the Solar Dynamics Observatory (SDO) , Sol.
Phys., 275(2), 41
Bouwer, D., Jones, A., and Tobiska, W. K.: 2013, personal communication
Bowman, B. R. and Tobiska, W. K.:2008, A New Empirical Thermospheric Density Model
JB2008 Using New Solar and Geomagnetic Indices. In: AIAA 2008-6438, AIAA/AAS
Astrodynamics Specialist Conference and Exhibit, Honolulu, Hawaii.
Breit, G. and Tuve, M.A.: 1926, A Test of the Existence of the Conducting Layer. Physical Review
28 (3): 554–575
130
Carlson, R. W., Ogawa, H. S., Phillips, E. , and Judge, D. L.,: 1984, Absolute measurement of the
extreme UV solar flux. Applied Optics, 28(14), 2327.
Chamberlain, J. W. and Hunten, D. M.: 1987, Theory of Planetary Atmospheres An Introduction to
Their Physics and Chemistry. Copyright © 1987 by Acadmeic Press, Inc.
Chamberlin, Phillip C., Thomas N. Woods, and Francis G. Eparvier,: 2007, Flare Irradiance
Spectral Model (FISM): Daily component algorithms and results. Space Weather J., 5,
S07005.
Chapman, S.: 1931, The absorption and dissociative or ionizing effect of monochromatic radiation
in an atmosphere on a rotating earth, Proceedings of the Physical Society, 43, 26
Del Zanna, G., and Andretta, V., 2011, The EUV spectrum of the Sun:SOHO CDS NIS irradiances
from 1998 until 2010, Astronomy and Astrophysics, 528, A139.
Del Zanna, G., B. J. I. Bromage, E. Landi and M. Landini,: 2001, Solar EUV spectroscopic
observations with SOHO/CDS, I. An in-flight calibration study. Astronomy and Astrophysics
379(2), 708.
Dere, K. P., E. Landi, H.E. Mason, B. C. Monsignori Fossi, P. R. Young,: 2007, CHIANTI - An
Atomic Database For Emission Lines Paper I: Wavelengths greater than 50 Angstroms,
Astron. Astrophys. Suppl. Ser., 125(1), 149
Didkovsky, L. V., Judge, D. L., Jones, A. R., Wieman, S., Tsurutani, B. T., and Mcmullin, D..:
2007, Correction of SOHO CELIAS/SEM EUV Measurements Saturated by Extreme Solar
Flare Events. Astron. Nachr. 328, 36.
Didkovsky, L. V., Judge, D. L., Wieman, S. R. and McMullin, D.: 2010a, Minima of Solar Cycles
22/23 and 23/24 as Seen in SOHO/CELIAS/SEM Absolute Solar EUV Flux. In Cranmer, S.,
Hoeksema, T., Kohl, J. (eds.) SOHO-23: Understanding a Peculiar Solar Minimum, ASP
Conference Series 428, 73.
Didkovsky, L., Judge, D., Wieman, S., Woods, T., Eparvier, F., Jones, A.: 2010b Comparison of
the First Solar EUV Irradiance Measurements from the SDO/EVE EUV Spectrophotometer
(ESP) with the SOHO/CELIAS Solar EUV Monitor (SEM) Absolute Solar EUV Irradiance.
COSPAR 2010, Bremen, Germany.
Didkovsky, L., Judge, Wieman, S., Woods., T., Jones, A.: 2012, EUV SpectroPhotometer (ESP) in
Extreme Ultraviolet Variability Experiment (EVE): Algorithms and Calibrations. Solar Phys.
275, 179.
Didkovsky, L., and Wieman, S., 2013, ESP preliminary results from sounding rocket flight 36.290,
SDO/EVE Monthly Report, October 2013
131
Didkovsky, L., and Wieman, S.: 2014a, Ionospheric total electron contents (TECs) as indicators of
solar EUV changes during the last two solar minima, JGR, 119, (2014), doi: 10.1002-
2014JA019977
Didkovsky, L., Wieman, S., and Woodraska, D.: 2014b, Some Results From a New Algorithm for
Using Available SDO/EVE/MEGS-A Data as Reference Solar Spectra for the EVE/ESP
Absolute Solar Irradiance Calculations, a poster presented at the 2014 Living With A Star
Science Meeting, Portland, OR
Dominique, M., J.-F. Hochedez, W. Schmutz, I. E. Dammasch, A. I. Shapiro, M. Kretzschmar, A.
N. Zhukov, D. Gillotay, Y. Stockman, A. BenMoussa 2013: The LYRA Instrument Onboard
PROBA2: Description and In-Flight Performance. Solar Phys. 286, 21.
Edlén, B., Die Deutung der Emissionslinien im Spektrum der Sonnenkorona. Mit 6 Abbildungen
Zeitschrift für Astrophysik, 22, p.30
Emmert, J. T., J. M. Picone, J. L. Lean, and S. H. Knowles: 2004, Global change in the
thermosphere: Compelling evidence of a secular decrease in density, J. Geophys. Res., 109,
A02301
Emmert, J. T., J. M. Picone, and R. R. Meier: 2008, Thermospheric global average density trends,
1967–2007, derived from orbits of 5000 near‐Earth objects, Geophys. Res. Lett., 35, L05101
Emmert, J. T., J. L. Lean, J. M. Picone: 2010, Record‐low thermospheric density during the 2008
solar minimum, Geophys. Res. Lett., 37, L12102.
Evans, J.S., D.J. Strickland. W.K. Woo, D.R. McMullin, S.P. Plunkett, R.A. Viereck.: 2010, Early
Observations by the GOES-13 Solar Extreme Ultraviolet Sensor (EUVS). Solar Phys. 26, 71.
Feldman, U and J M Laming: 2000, Element Abundances in the Upper Atmospheres of the Sun
and Stars: Update of Observational Results, Phys. Scr., 61 222
Foukal, P.: 1998, Extension of the F10.7 Index to 1905 using Mt. Wilson Ca K Spectroheliograms,
Geophys. Res. Lett., 25, 15
Funsten, H. O., Ritzau, S.M., Harper, R. W., Korde, R.:2004, Fundamental limits to detection of
low-energy ions using silicon solid-state detectors, Appl. Phys. Lett. 84, 3552.
Furst, M.L., Graves, R.M., Madden, R.P.: 1993, Synchrotron ultraviolet radiation facility (SURF
II) radiometric instrumentation calibration facility. Opt. Eng. 32, 2930.
Gary, G. A.: 2001, Plasma Beta Above a Solar Active Region: Rethinking the Paradigm, Solar
Physics, 203, 71.
Grevesse, N., Asplund, M., & Sauval, A. J.: 2007, The Solar Chemical Composition, Space Sci.
Rev., 130, 105
132
Grotrian, W.: 1939, Zur Frage der Deutung der Linien im Spektrum der Sonnenkorona,
Naturwissenschaften 27, 214
Gruntman, M.A.: 1995, EUV radiation filtering by free-standing transmission gratings,
AppliedOptics, v.34, N.25, 5732-5737
Haberreiter, M., Delouille, V., Mampaey, B., Verbeeck, C., Del Zanna, G., and Wieman, S.: 2014,
Reconstruction of the solar EUV irradiance from 1996 to 2010 based on SOHO/EIT images,
J. Space Weather Space Clim., 4, A30, DOI:10.1051/swsc/2014027
Haberreiter, M,: 2011 Towards the reconstruction of the EUV irradiance for solar Cycle 23. In:
Proceedings of the International Astronomical Union 7, 97.
Hall, L. A., J. E. Higgins, C. W. Chagnon, and H. E. Hinteregger: 1969, Solar-cycle variation of
extreme ultraviolet radiation, J. Geophys. Res., 74(16), 4181
Hall, L. A., and H. E. Hinteregger: 1970, Solar radiation in the extreme ultraviolet and its variation
with solar rotation, J. Geophys. Res., 75, 6959
Harrison R.A., Sawyer E.C., Carter M.K., et al., , 1995, The Coronal Diagnostic Spectrometer for
the solar and heliospheric observatory. Sol. Phys. 162, 233.
Hedin, A. E., et al.: 1977, A global thermospheric model based on mass spectrometer and
incoherent scatter data MSIS, 1. N2 density and temperature, J. Geophys. Res., 82,16, 2139–
2147
Henke, B.L., Gullikson, E.M., Davis, J.C.: 1993, X-ray interactions: photoabsorption, scattering,
transmission, and reflection at E = 50 – 30 000 eV, Z = 1 – 92. At. Data Nucl. Data Tables
54(2), 181.
Hernández-Pajares, M., Juan, J.M., Sanz, J.: 1999, New approaches in global ionospheric
determination using ground GPS data, Journal of Atmospheric and Solar-Terrestrial Physics,
61, 1237.
Hernández‐Pajares, M., J. M. Juan, J. Sanz, R. Orus, A. Garcia‐Rigo, J. Feltens, A. Komjathy, S.
C. Schaer, and A. Krankowski: 2009, The IGS VTEC maps: A reliable source of ionospheric
information since 1998, J. Geod., 83, 263
Hock, R. A.: 2012, The Role of Solar Flares in the Variability of the Extreme Ultraviolet, PhD
thesis, University of Colorado, Boulder
Hock, R.A., Chamberlin, P.C., Woods, T.N., Crotser, D., Eparvier, F.G., Woodraska, D.L.,
Woods, E.C.: 2012, Extreme Ultraviolet Variability Experiment (EVE) Multiple EUV
Grating Spectrographs (MEGS): Radiometric Calibrations and Results. Solar Phys. 275, 145.
Hovestadt, D., Hilchenbach, M., Bürgi, A., Klecker, B., Laeverenz, P., Scholer, M., Grünwaldt, H.,
Axford, W.I., Livi, S., Marsch, E.,Wilken, B.,Winterhoff, H.P., Ipavich, F.M., Bedini, P.,
133
Coplan, M.A., Galvin, A.B., Gloeckler, G., Bochsler, P., Balsiger, H., Fischer, J., Geiss, J.,
Kallenbach, R., Wurz, P., Reiche,K.-U., Gliem, F., Judge, D.L., Ogawa, H.S., Hsieh, K.C.,
Möbius, E., Lee, M.A., Managadze, G.G.,Verigin, M.I., Neugebauer, M.: 1995, CELIAS –
charge, element and isotope analysis system for SOHO. Solar Phys. 162, 441.
Hurlburt, N., and DeRosa, M.: 2008, On The Stability Of Active Regions And Sunspots, Ap. J.,
684, L123–L126
Jacchia, L.G.: 1965, Static Diffusion Models of the Upper Atmosphere with Empirical
Temperature Profiles, Smithsonian Contributions to Astrophysics, 8, p.215.
Jacchia, L. G.: 1971, Revised static models of the thermosphere and exosphere with empirical
temperature profiles, Spec Report 332, Smithsonian Astrophysical Observatory, Cambridge,
Mass.
Jensen, Anna B.O., and Cathryn Mitchell.: 2011, GNSS and the ionosphere: what's in store for the
next solar maximum? GPS World, 22.2, 40
Judge, D.L., McMullin, D.R., Ogawa, H.S., Hovestadt, D., Klecker, B., Hilchenbach, M., Möbius,
E., Canfield, L.R., Vest, R.E., Watts, R., Tarrio, C., Kuehne, M., Wurz, P.: 1998, First solar
EUV irradiances from SOHO by the CELIAS/SEM. Solar Phys. 177, 161.
Judge, D., Ogawa, H.S., McMullin, D.R., Gangopadhyay, P., and Pap, J.M.: 2002, The SOHO
CELIAS/SEM EUV Database From SC23 Minimum to the Present. Adv. Space Res. 29,
1963.
Kelley, Michael C.: 1989, The Earth’s Ionosphere: Plasma Physics and Electrodynamics,
Copyright © 1989, Academic Press
Klimchuk, J. A.: 2006, On Solving the Coronal Heating Problem, Solar Physics 234, 41–77
Krumrey, M., Herrmann, C., Müller, P., Ulm, G.: 2000, Synchrotron-radiation-based cryogenic
radiometry in the X-ray range. Metrologia 37, 361.
Landi, E., P. R. Young, K. P. Dere, G. Del Zanna, and H. E. Mason, 2013: CHIANTI—An Atomic
Database For Emission Lines. XIII. Soft X-Ray Improvements and Other Changes, ApJ 763
86
Lanyi, G. E., and Roth, T.: 1988, A comparison of mapped and measured total ionospheric
electron content using global positioning system and beacon satellite observations, Radio
Sci., 23, 483.
Lean, J. L., J. T. Emmert, J. M. Picone, and R. R. Meier,: 2011a, Global and regional trends in
ionospheric total electron content. J. Geophys. Res. 116, A00H04.
Lean, J. L., J. T. Emmert, J. M. Picone, and R. R. Meier,: 2011b, Ionospheric total electron
content: Global and hemispheric climatology. J. Geophys. Res. 116, A10318.
134
Lean, J. L., J. M. Picone, and J. T. Emmert: 2009, Quantitative forecasting of near‐term solar
activity and upper atmospheric density, J. Geophys. Res., 114, A07301.
Liang, M.-C., K.-F. Li, R.-L. Shia, and Y. L. Yung (2008), Short-period solar cycle signals in the
ionosphere observed by FORMOSAT-3/COSMIC, Geophys. Res. Lett., 35, L15818
Machol, J., Viereck, R.: 2013, GOES Preliminary Calibrations, a talk presented at The Inter-
Calibration and Degradation Workshop, Belgium Solar-Terrestrial Center of Excellence, 15 –
18 April
Mannucci, A. J., B. D. Wilson, D. N. Yuan, C. H. Ho, U. J. Lindqwister, and Runge, T. F.: 1998,
A global mapping technique for GPS-derived ionospheric total electron content
measurements. Radio Science, 33(3), 565.
McMullin, D. R., D. L. Judge, M. Hilchenbach, F. Ipavich, P. Bochsler, P. Wurz, A. Bürgi, W. T.
Thompson, J. S. Newmark: 2002, In-flight comparisons of solar EUV irradiance
measurements provided by the CELIAS/SEM on SOHO. In: Pauluhn, A., Huber, M.C.E.,
von Steiger, R. (eds.) The Radiometric Calibration of SOHO, ISSI Scientific Report SR-002,
135
Mitra, A. P.: 1959, Time and height variations in the daytime processes in the ionosphere: Part I: A
noontime model of the ionosphere loss coefficient from 60 to 600 km over middle latitudes.
J. Geophys. Res. 64, 733
Nelder, J. A. and R. Mead: 1965, A Simplex Method for Function Minimization, The Computer
Journal, 7 (4): 308
Nikutowski B, Brunner R, Erhardt C, Knecht S, Schmidtke G.,: 2011, Distinct EUV minimum of
the solar irradiance (16-40 nm) observed by SolACES spectrometers onboard the
International Space Station (ISS) in August/September 2009. Adv. Space Res. 48, 899.
Ogawa, H. S., Judge, D. L.: 1986, Absolute Solar Flux Measurement Shortward of 575Å. J.
Geophys. Res., 91, 7089.
Picone, J. M., A. E. Hedin, D. P. Drob, and A. C. Aikin: 2002, NRLMSISE-00 empirical model of
the atmosphere: Statistical comparisons and scientific issues, J. Geophys. Res., 107(A12),
1468
Qian, Liying, Alan G. Burns, Phillip C. Chamberlin, and Stanley C. Solomon: 2011, Variability of
thermosphere and ionosphere responses to solar flares, J. Geophys. Res., 116, A10309
Richards, P. G.: 2012, Re-evaluation of thermosphere heating by solar extreme ultraviolet and
ultraviolet radiation, Can. J. Phys. 90, 759–767
Rishbeth, H., and R. G. Roble: 1992, Cooling of the upper atmosphere by enhanced greenhouse
gases: Modeling of thermospheric and ionospheric effects, Planet. Space Sci., 40, 1011
135
Roble, R. G., and R. E. Dickinson: 1989, How will changes in carbon dioxide and methane modify
the mean structure of the mesosphere and thermosphere?, Geophys. Res. Lett., 16, 1441
Scime, E.E., Anderson, E.H., McComas, D.J., Schattenburg, M.L.: 1995, Extreme-ultraviolet
polarization and filtering with gold transmission gratings. Appl. Opt. 34, 648.
Schattenburg, M.L., Anderson, E.H.: 1990, X-ray/VUV transmission gratings for astrophysical and
laboratory applications. Phys Scr. 41, 13.
Schmidtke, G., Thomas N. Woods, John Worden, Gary J. Rottman, Harry Doll, Claus Wita and
Stanley C. Solomon: 1992, Solar EUV irradiance from the San Marco Assi: A reference
spectrum, Geophys. Res. Lett., 19, 2175
Solomon, S. C., T. N. Woods, L. V. Didkovsky, J. T. Emmert, and L. Qian: 2010, Anomalously
low solar extreme-ultraviolet irradiance and thermospheric density during solar minimum,
Geophys. Res. Lett., 37, L16103
Solomon, S.C. and Qian, L.: 2011, Model Simulations of Thermosphere and Ionosphere Density
during Solar Minimum, CCSCM Workshop, Boulder, Colorado, 17 May 2011
Solomon, S.C., L. Qian, L.V. Didkovsky, R.A. Viereck, and T.N. Woods: 2011 Causes of low
thermospheric density during the 2007-2009 solar minimum. J. Geophys. Res. 116, A00H07.
Solomon, S. C., Qian, L., and Burns A. G.: 2013, The anomalous ionosphere between solar cycles
23 and 24, Journal Of Geophysical Research: Space Physics, 118, 6524–6535,
doi:10.1002/jgra.50561,
Stolarski, R. S., Hays, P. B., Roble, R. G.: 1975, Atmospheric Heating by Solar EUV Radiation.
J. Geophys. Res. 80, 2266.
Storz, M. F., Bruce R. Bowman, and Major James I. Branson: 2002, High Accuracy Satellite Drag
Model (HASDM), AIAA/AAS Astrodynamics Specialist Conference and Exhibit, Monterey,
CA, AIAA 2002-4886
Tapping, K. F.: 2013, The 10.7 cm solar radio flux (F10.7), Space Weather, 11, 394–406,
doi:10.1002/swe.20064, 2013
Tapping, K. F., DeTracy, B.: 1990, The Origin of the F10.7cm Flux, Solar Physics, 127: 321-332
Tapping, K. F.: 1987, Recent solar radio astronomy at centimeter wavelengths: The temporal
variability of the 10.7-cm flux, J. Geophys. Res., 92(D1), 829
Tascione, T. F.: 2010, Introduction to the Space Environment: Second Edition, Copyright © 2010
by the Krieger Publishing Company
136
Thompson, W. T., McMullin, D. R., Newmark, J. S.: 2002, Comparison of CDS Irradiance
Measurements with SEM and EIT. In: Pauluhn, A., Huber, M.C.E., von Steiger, R. (eds.) The
Radiometric Calibration of SOHO, ISSI Scientific Report SR-002, 211.
Tobiska, W. K., 2012, personal communication
Tobiska W. K., Carlson, H. C., Schunk, R. W. (eds.): 2011, American National Standard - Guide
to Reference and Standard Ionosphere Models, Copyright © 2011 American Institute of
Aeronautics and Astronautics
Tobiska, W.K., Bouwer, S.D., Bowman, B.R.: 2006, The development of new solar indices for use
in thermospheric density modeling. In: AIAA-2006-6165, AIAA/AAS Astrodynamics
Specialists Conference and Exhibit; Keystone, CO.
Tobiska, W. K., Tom Woods, Frank Eparvier, Rodney Viereck, Linton Floyd, Dave Bouwer, Gary
Rottman, O.R. White: 2000, The SOLAR2000 empirical solar irradiance model and forecast
tool, Journal of Atmospheric and Solar-Terrestrial Physics, 62, 14, 1233-1250
Tobiska, W. K.: 1988, A Solar Extreme Ultraviolet Flux Model, Doctoral Thesis, University of
Colorado, Boulder
Tsurutani, B. T. D. L. Judge, F. L. Guarnieri, P. Gangopadhyay, A. R. Jones, J. Nuttall, G. A.
Zambon, L. Didkovsky, A. J. Mannucci, B. Iijima, R. R. Meier, T. J. Immel, T. N. Woods, S.
Prasad, L. Floyd, J. Huba, S. C. Solomon, P. Straus, and R. Viereck: 2005, The October 28,
2003 extreme EUV solar flare and resultant extreme ionospheric effects: Comparison to other
Halloween events and the Bastille Day event, J. Geophys. Res., 32, L03S09
Vest, R., Canfield , L. R., Furst, M.L., Graves, R.M., Hamilton, A. D., Hughey, L. R., Lucatorto,
T. B., Madden, R. P.: 1999, NIST Programs for Calibrations in the Far Ultraviolet Spectral
Region. In: Carruthers, G. R., and Dymond, K. F. (eds.), SPIE Proc. 3818, 15; Ultraviolet
Atmospheric and Space Remote Sensing: Methods and Instrumentation II.
Viereck, R., L. Puga, D. McMullin, D. Judge, M. Weber, and Tobiska, W. K.: 2001, The Mg II
Index: A Proxy for Solar EUV, Geophys. Res. Lett., 28, 1343–1346.
Viereck, R.: 2004a, Monitoring of EUV, UV, and Total Solar Irradiance from NOAA Satellites,
Living With a Star Workshop, March 2004, Boulder, CO
Viereck, R. A., L. E. Floyd, P. C. Crane, T. N. Woods, B. G. Knapp, G. Rottman, M. Weber, L. C.
Puga, and M. T. DeLand: 2004b, A composite Mg II index spanning from 1978 to 2003,
Space Weather, 2, S10005.
Viereck, R. A., M. Snow, M. T. DeLand, M. Weber, L. Puga, and D. Bouwer: 2010, Trends in
solar UV and EUV irradiance: An update to the MgII Index and a comparison of proxies and
data to evaluate trends of the last 11 year solar cycle, Abstract GC21B‐0877 presented at
2010 Fall Meeting, AGU, San Francisco, Calif., 13–17 Dec.
137
Warren, H. P., Mariska, D., and Lean, J.: 1998, A new reference spectrum for the EUV irradiance
of the quiet Sun 1 - Emission measure formulation, J. Geophys. Res., 103, A6, 12,077.
Wende, C. D., The normalization of solar x-ray data from many experiments, Sol. Phys., 22, 492,
1972.
Wieman, S.R., L.V. Didkovsky, and D.L. Judge: 2014, Resolving differences in absolute
irradiance measurements between the SOHO/CELIAS/SEM and the SDO/EVE, Sol. Phys.,
289, 2907–2925, DOI: 10.1007/s11207-014-0519-5.
Wieman, S. R., Judge, D. L., Didkovsky, L. V.:2011, Solar EUV Monitor (SEM) absolute
irradiance measurements and how they are affected by choice of reference spectrum. In:
Fineschi, S., Fennelly, J. (eds.), SPIE Proc. 8148, 84180G; Solar Physics and Space Weather
Instrumentation IV. DOI: 10.1117/12.893163
Wieman,S., Didkovsky,L., Judge.D., Jones.A., Harmon,M., A filter free dual transmission grating
spectrometer for the extreme-ultraviolet, In: Fineschi, S., Viereck, R., SPIE Proc. 6689,
Solar Physics and Space Weather Instrumentation II, 66890R, DOI:10.1117/12.732953
Woods, T. N., and G. J. Rottman: 1990, Solar EUV irradiance derived from a Sounding Rocket
Experiment on November 10, 1988, J. Geophys. Res., 95(A5), 6227
Woods, T.N., Ogawa, H., Tobiska, K., Farnik, F.: 1998, SOLERS-22 WG-4 and WG-5 report for
the 1996 SOLERS-22 workshop. In: Pap, J.M., Fröhlich, C., Ulrich, R.K. (eds.) Solar
Electromagnetic Radiation Study for Solar Cycle 22, Kluwer, Dordrecht, 511.
Woods, T. N., Tobiska, W. K., Rottman, G. J., Worden, J. R.: 2000, Improved solar Lyman-α
irradiance modeling from 1947 through 1999 based on UARS observations, Journal of
Geophysical Review, 105 (A12), 27,195-27,215
Woods, T. N., Eparvier, F. G., Bailey, S. M., Chamberlin, P. C., Lean, J., Rottman, G. J., Solomon,
S. C., Tobiska, W. K., and Woodraska, D. L.,: 2005, Solar EUV Experiment (SEE): Mission
overview and first results. J. Geophys. Res. 110, A01312.
Woods, T. N., and G. Rottman,: 2005, The XUV photometer system (XPS): Solar variations
during the SORCE mission. Sol. Phys. 230, 345.
Woods, T. N., Francis G. Eparvier: 2006, Solar ultraviolet variability during the TIMED mission,
Adv. Space Res., 37, 219.
Woods, T. N., Chamberlin, P. C., Harder, J. W., et al. 2009, Geophys. Res. Lett., 36, 1101
Woods, T.N., Chamberlin, P., Eparvier, F.G., Hock, R., Jones, A., Woodraska, D., Judge, D.,
Didkovsky, L., Lean, J., Mariska, J., et al.,: 2012, Extreme ultraviolet variability experiment
(EVE) on the solar dynamics observatory (SDO): Overview of science objectives, instrument
design, data products, and model developments. Solar Phys. 275, 115.
138
Yang, S. H., J. Zhang, C. L. Jin, L. P. Li, and H. Y. Duan: 2009, Response of the solar atmosphere
to magnetic field evolution in a coronal hole region, A&A 501, 745–753
139
Appendix: List of Related Publications
Papers
Wieman, S.R., L.V. Didkovsky, and D.L. Judge: 2014, Resolving differences in absolute
irradiance measurements between the SOHO/CELIAS/SEM and the SDO/EVE, Sol. Phys., 289,
2907–2925, DOI: 10.1007/s11207-014-0519-5.
Wieman, S. R., Judge, D. L., Didkovsky, L. V.:2011, Solar EUV Monitor (SEM) absolute
irradiance measurements and how they are affected by choice of reference spectrum. In: Fineschi,
S., Fennelly, J. (eds.), SPIE Proc. 8148, 84180G; Solar Physics and Space Weather
Instrumentation IV. DOI: 10.1117/12.893163
Wieman,S., Didkovsky,L., Judge.D., Jones.A., Harmon,M.:2007, A filter free dual transmission
grating spectrometer for the extreme-ultraviolet, In: Fineschi, S., Viereck, R., SPIE Proc. 6689,
Solar Physics and Space Weather Instrumentation II, 66890R, DOI:10.1117/12.732953
Didkovsky, L., and Wieman, S.: 2014, Ionospheric total electron contents (TECs) as indicators of
solar EUV changes during the last two solar minima, JGR, 119, (2014), doi: 10.1002-
2014JA019977
Haberreiter, M., Delouille, V., Mampaey, B., Verbeeck, C., Del Zanna, G., and Wieman, S.: 2014,
Reconstruction of the solar EUV irradiance from 1996 to 2010 based on SOHO/EIT images, J.
Space Weather Space Clim., 4, A30, DOI:10.1051/swsc/2014027
Conference and Workshop Talks
Wieman, S. R., Didkovsky, L. V., Judge, D. L.,: 2010, SOHO/CELIAS Solar EUV Monitor
(SEM) Absolute Solar EUV Irradiance Measurements Spanning Two Solar Minima (Invited),
GC33C-05 presented at 2010 Fall Meeting, AGU, San Francisco, Calif., 13-17 Dec
Didkovsky, L., Judge, D. L., Wieman, S., Jones, A., Eparvier, F., Woods, T., 2010: ”A
Comparison of the First Light He II (30.4nm) Irradiance Measurements from SDO/EVE/ESP and
SOHO/CELIAS/SEM”, 38th COSPAR, Bremen, Germany, July 21, 2010
Wieman, S., Didkovsky, L., Judge, D.: 2011, Solar EUV Monitor (SEM) Instrument Overview,
Background, and Calibration, presented at the Solar EUV Inter-calibration Workshop, LASP,
University of Colorado, Boulder, October 25-27, 2011
Wieman, S. R., Didkovsky, L. V., Judge, D. L.,: 2012, SOHO SEM Estimates for Lower EUV
Irradiance in 2008-2009, EVE Calibration Workshop, Yosemite, CA, Nov. 1, 2012
140
Wieman, S., Didkovsky, L., Judge, D.: 2013, SOHO/Solar EUV Monitor (SEM) and
SDO/EVE/EUV SpectroPhotometer (ESP) Calibration, Degradation and Comparisons (Invited)
Solar Terrestrial Center for Excellence Solar EUV Irradiance Working Group, Royal Observatory
of Belgium, April 15-17, 2013
Wieman, S., Machol, J., Jones, A.: 2013, Comparison of EUV and soft x-ray measurements, EVE
Science Working Group, Boulder, CO, September 24, 2013
Abstract (if available)
Abstract
Solar irradiance measurements in the extreme ultraviolet (EUV) and soft X‐ray (SXR) spectral ranges are important to studies of solar variability and its impact on the geospace environment and to operations affected by space weather. While the availability of solar EUV measurements has increased over the last two decades, data from the Solar EUV Monitor (SEM), part of the Charge, Element, and Isotope Analysis System (CELIAS) on board the Solar and Heliospheric Observatory (SOHO) remain unique in that they are continuous with high time cadence over a long time period which includes two solar minima, and their accuracy has been maintained based on a long series of sounding rocket calibration underflights, resulting in their use as the basis for a solar activity index for space weather operations. The purpose of this work is to further improve the absolute calibration of the SEM EUV irradiance measurements using data unavailable during earlier SEM calibrations. Solar EUV variability occurs over a range of timescales, including periodic changes associated with the 11‐year solar sunspot cycle (i.e. the 22‐year solar magnetic cycle). Thus, long‐term stable irradiance measurements are important in order to understand this variability and its relation to long‐term changes in the geospace environment. Maintaining the absolute accuracy of irradiance measurements is required to avoid data inconsistencies and the misinterpretation of instrument-related bias when compiling long‐term data sets comprised of measurements from more than one instrument. ❧ Although analyses of the SEM absolute calibration have been ongoing since the launch of SOHO and have incorporated measurements from a series of sounding rocket underflights, in recent years additional data have become available which have created opportunities and motivation for further refinement of the SEM measurements. Firstly, converting the SEM raw data into irradiance values depends on the instrument response function and on the spectral distribution of solar irradiance (i.e. reference spectrum) within the SEM sensitivity range, and both of these parameters have been determined with greater accuracy than those used in SEM data processing to date. Secondly, reliable EUV irradiance measurements in a spectral range overlapping that of SEM are available from the Solar Dynamics Observatory’s EUV Variability Experiment (SDO/EVE) which includes provisions for in‐flight calibration and degradation monitoring, and these measurements differ from concurrent SEM values (obtained using the original response function and reference spectrum) by an amount that is consistent throughout the SDO and SOHO mission overlap suggesting there may be a systematic offset in the SEM irradiance values. Thirdly, the SEM measurements show lower (by ~15%) irradiance values for the minimum of Solar Cycles 22 and 23 compared to Solar Cycles 23 and 24—this inter‐minima change is consistent with the response of the earth’s upper atmosphere over the same period with regard to some solar EUV‐driven processes (i.e. thermospheric neutral density) but inconsistent with others (i.e. global mode ionospheric total electron content). Further investigation is thus required to determine whether the lower EUV irradiance measured by SEM is real or an artifact of long‐term instrument degradation. ❧ Resolving these three issues concerning the SEM data set is the objective of this dissertation, and the included work has several key results. It demonstrates that the differences between SOHO/SEM and SDO/EVE EUV irradiance measurements are resolved by reprocessing the SEM raw data using an updated SEM instrument response function (introduced as part of this work) and time‐dependent solar reference spectra. Additionally, it provides an updated time and wavelength dependent SEM instrument degradation function which is necessary to refine the estimate of inter‐minima change in EUV irradiance based on the SEM data. Finally, this work provides a procedure for producing a SEM‐equivalent EUV irradiance index based on SDO/EVE measurements. The motivation for this final effort is that SOHO/SEM irradiance values based on the previous response function and reference spectrum have already been adopted as a solar irradiance index used for modeling thermospheric density, and because such space weather operations are concerned with long‐term consistency, it is desirable to continue providing an irradiance index equivalent to that provided thus far by SEM using newer EUV instrumentation after the SOHO mission has ended.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Trajectory mission design and navigation for a space weather forecast
PDF
Three-dimensional kinetic simulations of whistler turbulence in solar wind on parallel supercomputers
PDF
Particle-in-cell simulations of kinetic-scale turbulence in the solar wind
PDF
Numerical and experimental investigations of dust-plasma-asteroid interactions
PDF
Experimental and numerical investigations of charging interactions of a dusty surface in space plasma
PDF
Experimental investigation on dusty surface charging in plasma
PDF
The space environment near the Sun and its effects on Parker Solar Probe spacecraft and FIELDS instrumentation
PDF
Interactions of planetary surfaces with space environments and their effects on volatile formation and transport: atomic scale simulations
PDF
Laboratory investigations of the near surface plasma field and charging at the lunar terminator
PDF
Increased fidelity space weather data collection using a non-linear CubeSat network
PDF
A new hot-spot temperature model and methodology to determine the loss of insulation life in distribution transformers
PDF
Reduction of large set data transmission using algorithmically corrected model-based techniques for bandwidth efficiency
PDF
High temperature latent heat thermal energy storage to augment solar thermal propulsion for microsatellites
PDF
Techniques for analysis and design of temporary capture and resonant motion in astrodynamics
PDF
An analytical and experimental study of evolving 3D deformation fields using vision-based approaches
PDF
A framework for automated administration of post stroke assessment test
PDF
The development of a hydraulic-control wave-maker (HCW) for the study of non-linear surface waves
PDF
Hydrogen peroxide vapor for small satellite propulsion
PDF
Mechanical behavior of materials in extreme conditions: a focus on creep plasticity
PDF
Performance prediction, state estimation and production optimization of a landfill
Asset Metadata
Creator
Wieman, Seth R.
(author)
Core Title
Revised calibration of a long-term solar extreme ultraviolet irradiance data set
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Astronautical Engineering
Publication Date
04/10/2015
Defense Date
02/24/2015
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
EUV instrumentation,OAI-PMH Harvest,solar EUV irradiance,solar EUV monitor,solar soft X-ray irradiance
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Gruntman, Michael (
committee chair
), Didkovsky, Leonid (
committee member
), Rhodes, Edward J., Jr. (
committee member
), Wang, Joseph (
committee member
)
Creator Email
seth.wieman@usc.edu,wieman@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-545984
Unique identifier
UC11297877
Identifier
etd-WiemanSeth-3273.pdf (filename),usctheses-c3-545984 (legacy record id)
Legacy Identifier
etd-WiemanSeth-3273.pdf
Dmrecord
545984
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Wieman, Seth R.
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
EUV instrumentation
solar EUV irradiance
solar EUV monitor
solar soft X-ray irradiance