Remote Sensing of Coastal Waters

1
Remote Sensing of Coastal Waters

• PI Prof. Samir AhmedDirector, Optical Remote
  Sensing
• Presenter
• Dr. Alexander Gilerson
• The City College of the City University of New
  York

Team Prof. S. Ahmed, F. Moshary, B. Gross, Dr.
A. Gilerson, J. Zhou Students R. Fortich, R.
Amin, I. Ioannou, S. Hliang, A. Tonizzo, N.
Steiner, J. Borrero, T. Iijima
2
Main Goals of Lab Measurements, Simulations and
Field Campaigns

• Separation of the overlapping fluorescence and
  elastic scattering applied to algae in sea water
• Parameterization of the fluorescence as a
  function of water parameters for the coastal
  zones
• Testing of the fluorescence height over baseline
  (FLH) algorithms
• Improvement of the NIR Chl retrieval algorithms
  using inversion models and field data in
  Chesapeake Bay, Georgia waters, Long Island Sound
  and Peconic Bay
• Polarization measurements for the separation of
  organic and inorganic water components
• Comparison of measured and simulated reflectances
  using Hydrolight and bio-optical models
• Comparison of in situ tests with satellite data
• BRDF in coastal waters

3
Reflectance spectra for open ocean
SeaWiFS Blue-Green Ratio Algorithm From K.
Carder, et al. ,2003
4
Water composition for open ocean and coastal
waters
Algae CDOM Minerals
Algae
5
Absorption/Backscatter features
1- Chlorophyll absorption can be probed
effectively using 440-570 band ratios 2- In
presence of TSS and CDOM, Blue-Green ratios are
contaminated. 3- Red-NIR algorithms are much
less sensitive to TSS, CDOM. 4- The 670-710
channels effectively probe the ChL absorption
feature and the 730 channel effectively
calculates the backscatter since water abs
dominates
6
Typical Reflectance Spectra for Coastal Waters
MODIS FLH is routinely processed for the open
ocean , it is more complicated for coastal
waters. VIIRS does not have a band near
fluorescence peak
Fluorescence component, elastic and total
reflectance
7
(No Transcript)
8
Field Measurements
Previous trips Chesapeake Bay, Georgia waters,
Long Island Sound, Peconic Bay This year Long
Island Sound, Peconic Bay, Jamaica Bay, NY
Harbor, Hudson River

• Instrumentation
• WET Labs package absorption, attenuation,
  scattering, backscattering, Chl concentrations,
  CDOM fluorescence, temperature, salinity
• Hyperspectral profiler (Satlantic) and GER
  spectroradiometer for water reflectance
  measurements above and below water surface with
  the option of polarization components detection
  (new!)
• Water sampling (Chl, TSS and mineral
  concentrations, CDOM absorption) added this
  year!

9
Areas of field campaigns
Areas of study Chesapeake Bay
(2005), Georgia waters near Sapelo Island (2006),
Long Island Sound, Peconic Bay in conditions of
strong algae bloom, NY Harbor, Hudson River
10
Variability of water parameters for 2007 field
measurements
11
Hydrolight simulations and field data for the
parameterization of Chl fluorescence as a
function of concentrations of water components
12
Polarization discrimination technique for the
separation of fluorescence and elastic scattering
L lens, FP fiber probe, A aperture, P1, P2
polarizers, C cuvette with algae, WL water
level.
S. Ahmed et al., Opt. Comm., 2004, A. Gilerson et
al., Appl. Opt., 2006
13
CREST Simulated Datasets
4 datasets, 500 reflectances each were simulated
using HYDROLIGHT with 1 nm resolution (instead of
previous 10nm -Lee) for typical conditions of
coastal waters Chl 1-100 mg/m3, Cmin 0 -1
mg/l and 1 100 mg/l, various Chl absorption
spectra Specific Chl absorption for this were
Left - high a around 675nm from Mobley, CCNY
and WASI right - low a around 675nm by
combining pico- and micro-plankton with different
weighting factors)
Datasets are used for analysis of fluorescence
contribution and FLH algorithms, NIR Chl
algorithms, BRDF effects, multispectral and
hyperspectral IOP retrieval algorithms
14
Fluorescence Retrieval from Field measurements in
Chesapeake Bay
--- measured reflectance, --- elastic,
--- with Fl, --- Fl
HYDROLIGHT simulated WET Labs data (using
measured absorption a and attenuation c)
Gaussian fluorescence fit into measured
reflectance a lower Chl, b - higher Chl -
for high Chl fluorescence spectrum is on the
left side of NIR peak
15
Shift of NIR peak with Chl and contribution of
fluorescence to the peak
16
Experimental and simulated shift of NIR max from
685 nm
Shift of NIR maximum with Chl based on
simulated data for ? 0 blue, 1 green, 2
red. Lines approximate boundaries of field data
Shift of NIR maximum with Chl. The line is a
fitted approximation of data from Schalles et al.
Fluorescence quantum yield is about ? 1
17
Chl fluorescence for coastal waters
Comparison of fluorescence maximums calculated
from expressions above with those simulated from
HYDROLIGHT- fluorescence is explicitly related to
water parameters
Comparison of retrieved fluorescence data and
those resulting from HYDROLIGHT simulated dataset
fluorescence is smaller in Case 2 waters
because of attenuation
All measured fluorescence maximums are in the
range of predicted values for coastal waters with
fluorescence quantum efficiency around ?1
18
Contribution of fluorescence magnitude to NIR as
a function of Chl and Cnap
Low Mineral Case
High Mineral Case
Ratio of Fl/Rrs(695) (a) as a function of Chl
and (b) as a function of NAP concentration based
on simulations from our data set (Chl1-100
mg/m3 For low Cnap Fl contributes
10-30, for Cnap gt 10 mg/l impact of fluorescence
on NIR peak is very low
A. Gilerson et al., Optics Express, 2007
19
Testing of Fluorescence Height over Baseline
(FLH) MODIS and MERIS algorithms
20
Fluorescence Height Algorithms
FLH is satisfactory for low Chl and
questionable for high Chl because of impact of
high absorption and scattering
MODIS 667, 678, 748 nm
MERIS 665, 681, 709 nm (753)
21
Simulated performance of MODIS and MERIS FLH
algorithms (low mineral case )
Average Fl for low NAP case (dataset 1)
MODIS
MERIS
a
b
Superimposed FLH
Performance of MODIS (a) and MERIS (b) FLH
algorithms with simulated reflectances. Solid
line superimposed fluorescence magnitude
according to the expression , symbols retrieved
FLH Result strongly depends on Chl
absorption spectra (type of species).
22
Improvement of FLH retrieval by change of central
band
667, 678, 748 nm standard MODIS
667, 685, 748 nm changed central band
Superimposed FLH
Substantial improvement with the center band near
685 nm
23
Simulated performance of MODIS and MERIS FLH
algorithms (more minerals)
Solid line superimposed fluorescence magnitude
according to the expression , symbols retrieved
FLH In presence even small amount of
minerals FLH algorithms fail.
24
FLH retrievals satellite data using new
atmospheric correction algorithm (1240 and 2130
nm bands) and SeaDAS Chesapeake Bay
Chl map for Chesapeake Bay, Delaware Bay,
coast, processed only for low mineral
concentrations (low reflectance at 670 nm)
Satellite retrieval using MODIS FLH algorithm
Correlation exists only for low Chl
25
Comparison of simulated and MODIS FLH retrieval
Simulated FLH using MODIS algorithm and three
values of fluorescence quantum efficiency ?
0.5, 1 and 2 comparison with MODIS FLH
retrieval for Chesapeake Bay area Simulated
values with ? 1 match well with satellite data
A. Gilerson et al., Optics Express (Part II), 2008
26
Improvement of NIR algorithms
27
Peconic Bay Satlantic profiler measurements
Radiance/Irradiance ratio
Lu/Ed
No bloom conditions
Red tide conditions
28
Spectral dependence of scattering for high Chl
Field measurement data from WET Labs AC-S
instrument bc-a. Backscattering should have
similar shape which will significantly affect NIR
peak
b, 1/m
29
Scattering spectra for different field stations
Magnitude of NIR trough on the scattering spectra
Scattering spectra
Scattering can significantly contribute to the
NIR peak
A. Gilerson et al., Sea Technology, 2007
30
Impact of Backscattering on NIR Ratio Algorithm
For The Retrieval Of Chl Concentration
Taking only absorption effects into account
Without correction for variations in bb
With correction
is added in the denominator
31
Polarization measurements for the separation of
organic and inorganic water components
32
Above-water angularly resolved measurements of
the polarization components (preliminary
results)
Specular direction
Anti-specular direction
?vlt0
?vgt0
Sensor
?i
?v
Air
Principal plane
Ocean

• Data were collected in the principal scattering
  plane
• A polarizer was attached to a hyperspectral
  radiance sensor (Satlantic)
• The radiometer rotated from -70 to 70 (steps
  of 5) relative to the zenith direction

33
In-situ field sensor to measure angular polarized
spectrum of water radiance (LIS, summer 2007)

• Chl and TSS were also determined from
  water samples collected in situ.

34
Degree of polarization as a function of viewing
angle field measurements Hudson River, Sept. 07

• Angular variations in the degree of polarization
  P show similar features for each wavelength
• P shows a minimum value around the nadir viewing
  angle and peaks around 60 close to Brewster
  angle 53.2 in specular direction

35
Polarized Radiative Transfer Simulation
(code of J. Chowdhary, NASA-GISS)

• The Stokes vectors of radiance were simulated
  through a coupled air-ocean polarized radiative
  transfer code. A four-component water model
  included
• Water
• CDOM
• Phytoplankton-like particles
• Mineral-like particles.
• The Chl and TSS obtained from water sample
  analysis were used to determine the relative
  concentrations of the scattering components

36
Comparison between simulated and measured
polarized components
Measured spectra match simulations after
appropriate corrections. Degree of polarization
from water components is about 0.2
37
Sensor Development (Future work)

• Under-water angularly resolved measurements of
  the polarization components
• Polarizers (0, 90, 45) are attached to three
  hyperspectral radiance sensors (Satlantic)
• ADV simultaneous measurements
• no need to manually switch polarizer
  position
• A step electric motor allows for a full rotation
  of the radiometers to change the angle
  automatically
• ADV faster and convenient measurements
• more accurate determination of the
  angular position

38
Analysis of Satellite Data for Long Island Sound
39
MODIS Aqua time series data for Chl, Western
Sound

• Each region (Western, Central and Eastern) covers
  an area of approximately 1815 sq. km ( 1089 sq.
  mi)
• Flags set when retrieving Chl from satellite
  data files (data not processed for flagged sets)
• (1) Atmospheric Correction Failure
• (2) Pixel is over land
• (4) High sun glint
• (5) Observed radiance very high or saturated
• (12) Turbid water detected

Data show some stability by year
40
Time series data for Chl, Center Sound

• The retrieved Chl data was interpolated to get
  values for days that we had no data for so that
  the time series would span the entire year.
• A sliding mean for 7 days at a time was taken
  to clean up the noise in the time series plot.

41
Time series data for Chl, Eastern Sound

• The results are consistent since we move east and
  closer to the ocean, the values of Chl decrease
  for both years.
• The average Chl was calculated for each region
  for both years and is shown in the box for each
  region.

42
Impact of the atmospheric correction type (SWIR
or NIR) algorithms on Chl retrieval
SWIR (1240, 2130 nm) Chl standard MODIS
(blue-green)
NIR(748, 865 nm) Chl standard MODIS
Chl values for NIR atm correction 2 times
higher
43
Comparison of the field measurements and MODIS
Aqua satellite data in Long Island Sound (SWIR
atm correction)
June 25, 2007
GER measured reflectance
44
Bidirectional reflection distribution function
(BRDF) in coastal waters
45
Bidirectionality of the Oceans Reflectance
Zenith
Satellite
LW
Bidirectional reflectance distribution function
(BRDF) describes directional distribution of
light radiance emerging from oceanic waters.BRDF
is a function of bio-optical properties,
wavelength and depth.
?O
?
?F
Lu
?'
bb is a backscattering, a -absorption
Nadir
46
Main objectives of study

• Analyze spectral dependence of Q parameter for
  various coastal water conditions through
  simulations and field measurements
• Develop second order model for remote sensing
  reflectance Rrsfunction(?) which takes into
  account sun and viewing angles
• Compare this model with other models and field
  data

47
Instruments for field measurements
Satlantic Hyperspectral Profiler 350 800 nm,
137 channels
Irradiance sensor to measure upwelling irradiance
Eu(?) was used in nadir position Q(?) Eu(?) /
Lu(?) Third Ed(?) sensor stays on the boat

• Instrument is designed to support
  experiments in the case 2 waters

48
Model comparison
rrs (0.0949? 0.0794(?2) for Case 1 Gordon
et al., 1988
rrs (0.0895? 0.1247(?2) Lee et al., 2002
These models (second order) do not take into
account ?0, ?
rrs 0.0512(14.6659? -7.8387?2 5.4571
?3) ? (10.1098/cos ?0) (10.4021/cos
?) - WASI, Gege, 2005
Rrs (?g1 (?2)g2(?3)g3(?4)g4) gi are
tabulated as a function of ?0, ? and azimuth
angle ?f Park, Ruddick, 2005
These models (fourth order) are inconvenient in
the retrieval
Proposed model
It is a second order approximation for Hydrolight
simulations
This model (second order) takes into account
effects of ?0, ? by combining second order and
WASI models
49
Results of model comparison
T030, T 0
Low minerals
High minerals
Minerals Cnap 0 1 mg/l
low ?bb/(abb)
Minerals Cnap 1 100 mg/l high ?
Our second order (black) model fits the
Hydrolight data well over the whole range of ?
and is only a second order relationship (simpler
and more convenient in retrieval)
50
Comparison of the model with field data (results)
for one of the stations in Georgia waters
Comparison of model and field data ?0 30, ?
0 Maximum ? is relatively low need test for
higher values
Fitting of simulated reflectance into measured one
51
Publications and presentations 2007-08

• Journal Publications
• A. Gilerson, J. Zhou, S. Hlaing, I. Ioannou, B.
  Gross, F. Moshary, and S. Ahmed, Fluorescence
  Component in the Reflectance Spectra from Coastal
  Waters. II. Performance of retrieval algorithms,
  Optics Express, 16, 2446-2460, 2008.
• A. Gilerson, J. Zhou, S. Hlaing, I. Ioannou, J.
  Schalles, B. Gross, F. Moshary, S. Ahmed,
  Fluorescence component in the reflectance
  spectra from coastal waters. Dependence on water
  composition, Optics Express, 15, 15702-15721,
  2007.
• A. Gilerson, J. Zhou, M. Vargas, B. Gross, F.
  Moshary and S. Ahmed. Impact of particulate
  scattering in coastal waters on reflectance
  spectra simulations and Chesapeake Bay
  measurements. Sea Technology, 48, No. 9, 2007.
• Conference Proceedings and Presentations
• S. Ahmed, A. Gilerson, J. Zhou, I. Ioannou, S.
  Hlaing, B. Gross, F. Moshary Impact of Apparent
  Fluorescence Shift on Retrieval Algorithms for
  Coastal Waters, Proceedings of Ocean Optics,
  XVIII, Montreal, October, 9-13, 2006.
• S. Ahmed, A. Gilerson, J. Zhou, J. Chowdhary, I.
  Ioannou, R. Amin, B. Gross, F. Moshary.
  Evaluation of the impact of backscatter spectral
  characteristics on Chl retrievals in coastal
  waters. Proc. SPIE Vol. 6406, 64060A, Goa, India,
  November, 2006.
• S. Ahmed, A. Gilerson, J. Zhou, I. Ioannou, S.
  Hlaing, B. Gross, and F. Moshary. The Effect of
  Reabsorption of Chlorophyll Fluorescence and
  Elastic Scattering in Coastal Waters on the
  Efficacy of Retrieval Algorithms. Proc. of 87th
  AMS Annual Meeting. San Antonio, TX, Jan. 2007.
• A. Gilerson, J. Zhou, I. Ioannou, S. Hlaing, W.
  Jerez, B. Gross, F. Moshary, S. Ahmed
  Characterization of the Fluorescence Reflectance
  Components and Performance of Retrieval FLH
  Algorithms in Coastal Waters. NASA Ocean Color
  Meeting, Seattle, April, 2007
• M. Vargas , M.M. Oo , B. Gross , F.Moshary , S.
  Ahmed. Improved robustness of atmospheric
  correction for highly productive coastal waters
  using the SWIR retrieval algorithm together with
  water leaving reflectance constraints at 412nm.
  NASA Ocean Color Meeting, Seattle, April, 2007
• A. Gilerson, J. Zhou, S. Hlaing, I. Ioannou, B.
  Gross, F. Moshary and S. Ahmed Fluorescence
  contribution to reflectance spectra for a variety
  of coastal waters, SPIE 6680, San Diego, CA,
  August, 2007
• S. Ahmed, A. Gilerson, J. Zhou, S. Hlaing, I.
  Ioannou, W. Jerez, B. Gross, F. Moshary. Impact
  of scattering and absorption of photosynthetic
  pigments on fluorescence retrieval algorithms for
  coastal waters, SPIE 6743, Florence, Italy, Sept.
  2007.
• A. Gilerson, J. Zhou, R. Fortich, I. Ioannou, S.
  Hlaing, B.Gross, F. Moshary and S. Ahmed.
  Spectral dependence of the bidirectional
  reflectance function in coastal waters and its
  impact on retrieval algorithms. Proceedings of
  IGARSS-07, Barcelona, Spain, July, 2007.

52
Student activities

• Fourth Education and Science Forum Science,
  Stewardship and Sustainability October 30, 31 and
  November 1, 2006 Florida AM University NOAA
  Environmental Cooperative Science Center,
  Tallahassee, Florida
• Attendees Profs. S. Ahmed, F. Moshary, B. Gross,
  Students R. Fortich, R. Amin
• NASA SeaDAS Workshop UMBC, January 8-10 2007,
  UMBC.
• Attendees Drs. A. Gilerson, J. Zhou, Students W.
  Jerez, I. Ioannou.
• NOAA CoRP Symposium, Maryland, June 19-20, 2007.
• Attendees R. Fortich, R. Amin, S. Hliang
• Presentations
• I. Ioannou, S. Hlaing, W. Jerez, A. Gilerson, J.
  Zhou, B. Gross, F. Moshary, S. Ahmed, Evaluation
  of MODIS and MERIS Algorithms for Chlorophyll
  Fluorescence Retrieval from the Reflectance
  Spectra in Coastal Waters oral presentation.
• R. Fortich, S. Hliang, J. Zhou, A. Gilerson, B.
  Gross, F. Moshary and S. Ahmed, Analysis of the
  bidirectional reflectance function under the
  variable conditions of coastal waters
• R. Amin, J. Zhou, A. Gilerson, B. Gross, F.
  Moshary and S. Ahmed, Evaluation of IOPs
  accuracy measurements and its impact on simulated
  reflectances
• S. Hlaing, I. Ioannou, W. Jerez, A. Gilerson, J.
  Zhou, B. Gross, F. Moshary, S. Ahmed,
  Fluorescence Contribution to Water Leaving
  Radiances and Its Dependence on Water
  Constituents in Coastal Waters

53
Leveraged Projects
A proposal entitled Exploring Techniques for
Improving Retrievals of Bio-Optical Properties of
Coastal Waters was prepared and submitted to the
Office of Naval Research for research
collaborations with the Naval Research
Laboratory, Stennis Space Center for 900,000
over 3 years. Approximately half the work in this
proposal leverages CREST efforts.