Remote estimations of oceanic constituents from optical reflectance spectra in coastal waters are challenging because of the complexity of the water composition as well as difficulties in estimation of water leaving radiance in several bands possibly due to inadequacy of current atmospheric correction schemes. This work focuses on development of a multiband inversion algorithm that combines remote sensing reflectance measurements at several wavelengths in the blue, green and red for retrievals of the absorption coefficients of phytoplankton, color dissolved organic matter and nonalgal particulates at 443nm as well as the particulate backscatter coefficient at 443nm. The algorithm was developed, using neural networks (NN), and was designed to use as input measurements on ocean color bands matching those of the Visible Infrared Imaging Radiometer Suite (VIIRS). The NN is trained on a simulated data set generated through a biooptical model for a broad range of typical coastal water parameters. The NN was evaluated using several statistical indicators, initially on the simulated data-set, as well as on field data from the NASA bio-Optical Marine Algorithm Data set, NOMAD, and data from our own field campaigns in the Chesapeake Bay which represent well the range of water optical properties as well as chlorophyll concentrations in coastal regions. The algorithm was also finally applied on a satellite - in situ databases that were assembled for the Chesapeake Bay region using MODIS and VIIRS satellite data. These databases were created using in-situ chlorophyll concentrations routinely measured in different locations throughout Chesapeake Bay and satellite reflectance overpass data that coexist in time with these in-situ measurements. NN application on this data-sets suggests that the blue (412 and 443nm) satellite bands are erroneous. The NN which was assessed for retrievals from VIIRS using only the 486, 551 and 671 bands showed that retrievals that omitted the 671 nm band was the most effective, possibly indicating an inaccuracy in the VIIRS 671 band that needs to be further investigated.
Remote estimation of chlorophyll-a concentration [Chl-a] in the Chesapeake Bay from reflectance spectra is challenging because of the optical complexity and variability of the water composition as well as atmospheric corrections for this area. This work is focused on algorithms for near surface measurements. The performance and tuning of several well established global inversion algorithms that use the NIR and Blue-Green parts of the spectrum are analyzed together with recently proposed algorithm that use the Red-Green part of the spectrum. These algorithms are evaluated and tuned on our field data collected during summer 2013 field campaign in the in the Chesapeake Bay region . These data consist of a full range of water optical properties as well as chlorophyll concentrations and specific absorption spectra from in water samples.
We then compare these algorithms with a multiband retrieval algorithm that was developed using neural networks (NN) and which was trained on simulated data generated through bio-optical modeling typical for a broad range of coastal water parameters, including those known for the Chesapeake Bay. This NN algorithm was then applied to our field measurements and used to retrieve the phytoplankton absorption at 443nm which was then related to [Chl-a]. In this process, special attention was paid to field data consistency in terms of both measured reflectance and [Chl-a] values, to avoid undesirable biases and trends. All algorithm retrievals were finally evaluated by several statistical indicators to arrive at their relative merits and potential for further improvements and application to satellite data.
We present a method for the separation of the non-algal absorption coefficient into its independent components of dissolved species and non-algal particulate absorptions from remote sensing reflectance (Rrs) measurements in the visible part of the spectrum. This separation is problematic due to the similar absorption spectra of these substances. Due to this complication, we approach the problem by constructing a neural network which relates the remote sensing reflectance at the available MODIS visible wavelengths (412, 443, 488, 531, 547 and 667nm) with the ratio of the absorption coefficient of non-algal particulates to the absorption coefficient of dissolved species, thereby permitting analytical separation of the total non-algal absorption into particulate and dissolved components. The resulting synthetically trained algorithm is tested on simulated data as well as independently on the NASA Bio-Optical Marine Algorithm Data set (NOMAD). Very good agreement is obtained, with R2 values of 87% and 78% for the non-algal particulate and dissolved absorption components, respectively for the NOMAD. Finally, we apply the algorithm to MODIS data and present global distributions for these parameters.
Using a dataset consisting of 9000 reflectance spectra simulated using HYDROLIGHT 5 for a broad range of observable natural water conditions, we have developed three neural networks (NNs) working in parallel to model the inverse problem for both oceanic and coastal waters. These NNs are used to relate the water leaving remote sensing reflectance (Rrs) at available MODIS visible wavelengths (412, 443, 488, 531, 547 and 667nm) to the phytoplankton (aph), non-phytoplankton particulate (adm), dissolved (ag) absorption and particulate backscattering (bbp) coefficients at 443nm. These reflectance derived parameters (aph(443), adm(443), ag(443), bbp(443)) are then combined with the measured reflectance values and used as input to a fourth NN, (IOP NN [Chl]), to derive chlorophyll concentration ([Chl]). Unlike NNs previously developed by us that were trained on a synthetic dataset and then tested on the NASA Bio-Optical Marine Algorithm Dataset (NOMAD), the (IOP NN [Chl]) network was both trained and tested solely on NOMAD. Although the inherent optical properties (IOP) can be derived from the optical signal through their direct relation to the Rrs, the relationship of [Chl] to IOP varies with location and season, and is therefore difficult to model globally. In order to demonstrate that the inclusion of derived IOP estimates along with radiance measurements can improve the retrieval of [Chl], we construct a neural network that is trained to derive [Chl] from reflectance measurements only We also compare our [Chl] product to that obtained from the current OC3 algorithm implemented by NASA’s Moderate Resolution Imaging Spectroradiometer (MODIS). Finally, we apply our algorithm to MODIS data and present and analyze the global seasonal variability for all three parameters.
As an integral part of the VIIRS sensor calibration and validation efforts, our group has been continuously monitoring the validity of the Visible Infrared Imager Radiometer Suite (VIIRS)’s Ocean Color (OC) and atmospheric data stream through time series in-situ data acquired at the observatory sites which are part of the AERONET – OC network. This paper addresses the preliminary evaluations of the VIIRS sensor’s performance for retrieving OC data of typical coastal water environments, by carrying out time-series, as well as qualitative and quantitative match-up comparisons analysis between in-situ and satellite retrieved OC data. Initial time-series match-up comparisons carried out for almost a year period (January to December, 2012) show that VIIRS data exhibits strong temporal and statistical agreements with AERONET-OC data demonstrating a potential for enhanced coastal water monitoring from space. VIIRS data processing schemes which apply different vicarious calibration gains are compared and analyzed based on AERONET-OC data as well as OC retrievals of the Moderate Resolution Imaging Spectro-radiometer (MODIS) sensor aboard the Aqua satellite. The underlying cause of the discrepancies observed in VIIRS retrieved normalized water-leaving radiances is also investigated.
Measurements of light intensity, often from space borne sensors, have been used to investigate the optical properties
of the constituents of Earth's ecosystem. In ocean color research, water-leaving radiance can give useful information
about inherent optical properties (IOPs). Additional consideration of polarization of the water-leaving radiance can
lead to a better understanding of the physical and optical characteristics of the water body. Polarization properties
strongly depend on particle microphysics, such as refractive index, effective radius, size distribution, and single
scattering albedo. Using radiative transfer simulations of the polarized light field for various ranges of water
constituents, we were able to develop relationships between the degree of polarization (DOP) and the ratio of
hydrosol absorption to attenuation coefficient. This relationship was then studied for different viewing geometries of
the polarized light and for different sun positions. A Neural Network sensitivity analysis was also performed to
better understand the dependence of DOP on microphysical parameters.
Estimating the Stokes vector components of the polarized water radiance from above water measurements is a
challenging task, mainly because of their small magnitude and the strong contamination by the sky light reflected on
the sea surface. Consequently, in most applications the Stokes vector components are considered equal to zero
except of I, the total reflectance. In this study, both below and above water measurements are used to assess the
feasibility of such retrievals and their use to determine the water composition. In-water inherent optical properties
(IOPs) were measured with commercially available instrumentation. In addition, in-water polarization characteristics
were measured by our multi-angular hyperspectral sensor which provided the Stokes components for a scattering
angles range of the 0-180° and a full spectral range between 400 and 750 nm. Second, a customized HyperSAS
(Satlantic) instrument is used from the coastal platform in Long Island Sound, NY (LISCO) acquiring above water
measurements. That instrumentation includes, in the standard configuration, two hyperspectral radiance sensors for
measuring upwelling and sky radiances and one irradiance sensor for measuring downwelling irradiance. In our
installation, HyperSAS capabilities were augmented by adding two radiance sensors having two polarizers oriented
at 0 and 45°, with respect to a reference axis ("HyperSAS-POL"). An ad hoc procedure, which included measurements and radiative transfer computations, has been developed enabling to estimate the contribution of the sky glint and subtract it from the signal directly measured by HyperSAS-POL. As a result, the retrieved spectral shape of the underwater degree of polarization is consistent with what obtained from in situ underwater measurements and depends on the IOPs of the ocean itself. In addition, the demonstrated correctness of this polarized measurements from LISCO site enable us to provide continuous time series from the beginning of June 2010.
The Long Island Sound Coastal Observational platform (LISCO) near Northport, New York, has been recently
established to support satellite data validation. LISCO is equipped with both multispectral SeaPRISM and
hyperspectral HyperSAS radiometers for ocean color measurements. LISCO substantially expands observational
capabilities for the continuous monitoring and assessment of ocean color satellite data quality. This offers the
potential for improving the calibration and validation activities of current and future Ocean Color satellite missions,
as well as for satellite intercomparisons and spectral characterization of coastal waters. Results of measurements
made by both the multi and hyperspectral instruments, in operation since October 2009, are presented, evaluated and
compared with ocean color satellite data. The comparisons with the normalized water-leaving radiance derived from
SeaPRISM with that from MERIS, MODIS and SeaWiFS showed satisfactory correlations (r > 0.9 at 550nm) and
consistencies (APD < 15% at 550nm). Similar and equivalent results are obtained when the hyperspectral
HYPERSAS data are compared with the same satellite datasets. The results confirm that the LISCO site is
appropriate for use in calibration/validation of the ocean color satellites in coastal waters and as a key element of the
AERONET-OC network. This makes it possible to foresee a wider use of the LISCO site to monitor current and
future ocean color multispectral (NPOESS, Sentinel) and hyperspectral (HICO) satellite missions.
Our previous studies showed that the Fluorescence Line Height (FLH) product, which uses 3 NIR bands at 667, 678, and
746 nm on the MODerate-resolution Imaging Spectroradiometer (MODIS) sensor, and similar bands on MERIS sensor,
is not reliable in coastal waters because of a peak in the elastic reflectance spectra which occurs due to the confluence of
chlorophyll and water absorption spectra and which overlaps spectrally the chlorophyll fluorescence. This combination
of two overlapping peaks makes fluorescence signal retrieval inaccurate. As a consequence, the present FLH algorithm
significantly underestimates fluorescence magnitudes in coastal waters. To overcome this problem, we introduce a new
and more accurate approach for the retrieval of FLH in turbid waters by the MODIS sensor, which exploits the
correlation between the blue-green and red bands reflectance ratios. We show that by making use of the combined
remote sensing reflectance's (Rrs) at 488nm, 547nm, 667nm and 678nm we can retrieve fluorescence accurately in case
2 waters even for low fluorescence quantum yield when fluorescence magnitudes are low. The derivation and validation
of our algorithm was performed using extensive synthetic datasets which cover a large variability of parameters typical
of coastal waters: with CDOM absorption at 400nm 0-2 m-1, mineral concentration 0-5g/m3 and chlorophyll
concentration of 0.5-100 mg/m3. In addition, we applied this proposed algorithm to MODIS satellite data and compared
it with the traditional FLH algorithm.
As a result of rich nutrient and terrestrial influence dissolved organic matter plays an important role in determining
the optical properties of coastal water. Despite the fact that the features of its fluorescence spectroscopy depend on
its complicated chemical components, which are source specific, estimation of solar stimulated CDOM fluorescence
is usually based on a fixed Gaussian spectral shape or a modeled fluorescence transfer function obtained from
extracted fuvic and humic acid. The excitation emission matrix (EEM) and absorption spectrum of CDOM extracted
from seawater in various coastal zones are measured using Perkin Elmer fluorescence spectrometer. Unlike previous
research, the obtained EEM spectroscopy data after calibration are then entered into the Hydrolight radiative transfer
program together with the data of inherent optical properties such as absorption and attenuation simultaneously
collected by Wetlabs in-situ instrument package to evaluate realistically the contribution of solar stimulated CDOM
fluorescence to the total reflectance in the Hudson River and New York City area. In addition, the CDOM
contribution to the total reflectance is evaluated with the presence of other water components based on a depth
integrated fluorescence model and a semi-analytical reflectance model. The simulated remote sensing reflectance
with the CDOM fluorescence on and off are compared with the field recorded spectra, through which its impact on
the closure of ocean color data as well as on the accuracy of estimating backscattering ratio and chlorophyll
fluorescence is also assessed.
The efficient monitoring of coastal, or Case 2 waters by optical remote sensing has always been a challenging task. This
study develops a neural network (NN) model to examine the possibility of accurate retrieval in waters where semianalytical
and empirical algorithms do not perform satisfactory due to the large variability in the coexistence of
particulates, dissolved matter and phytoplankton species. A multi-layer forward neural network was constructed to
estimate the total absorption, phytoplankton absorption, total suspended matter and color dissolved organic matter
absorption and total backscattering at the same time from in-situ measured water surface reflectance spectra. The neural
network was trained using 60% of the 1000 reflectance spectra from a synthetic datasets that were generated using
Hydrolight for water properties typical to coastal regions. Then, the neural network model was tested with the remaining
40% of the simulated reflectance spectra and applied to field data. Primarily the NN was trained and tested with the input
of traditional visible channels. Thereafter one more channel was added from the UV region and the NN was again
trained and tested. The retrievals with the addition of UV improve both in the simulated and field data..
Recent studies show that the contribution of the chlorophyll fluorescence component to the NIR reflectance is significant
only for relatively low concentrations of minerals; furthermore, Fluorescence Line Height (FLH) algorithms do not work
properly for [Chl] > 5-10 mg/m3 because of the deviation of elastic reflectance from the baseline. But even for relatively
low chlorophyll concentration [Chl] values the relationship between FLH and [Chl] is affected by several factors which
should be taken into account. The sensitivity of this relationship to atmospheric correction schemes, incident and
viewing angles and chlorophyll retrieval algorithms are analyzed, and a special correlation procedure is developed to
minimize these effects. Effective fluorescence quantum yield distributions retrieved from field measurements and
satellite imagery are also compared with the previously indirectly estimated values based on the analysis of the shift of
the observed NIR reflectance peak from the fluorescence peak at 685 nm.
Fluorescence Line Height (FLH) algorithms are effective for fluorescence retrieval in the open ocean where elastic
reflectance in the fluorescence zone does not deviate much from the baseline. In coastal waters, FLH algorithms are
significantly complicated by the overlap of the fluorescence and elastic reflectance peaks. To test accuracy of MODIS,
MERIS and other FLH algorithms, we compared numerical simulations using an extensive synthetic database suitable
for case II waters, with results of extensive field measurements of reflectance, absorption and attenuation spectra by us
in the Chesapeake Bay, as well as satellite FLH data from several areas that typically show low correlation between
[Chl] and FLH. Our synthetic datasets were created using the HYDROLIGHT radiative transfer code with IOP's
connected to parameterized microphysical models in accordance with procedures used to generate the IOCCG dataset,
but with some added improvements. These included higher (1 nm) spectral resolution, a wider range of parameters
typical for coastal waters, including chlorophyll specific absorptions with significant variations in spectral shapes and
magnitude. HYDROLIGHT simulations of elastic reflectance using measured attenuation/extinction spectra followed by
subtraction from measured reflectance, permitted retrieval of the fluorescence contribution to the latter, for comparisons
with the data set simulations. We find relatively small fluorescence contributions to surface reflectance for mineral
concentrations > 5 mg/l because of strong attenuation in the excitation zone and enhanced elastic reflectance making
fluorescence detection unrealistic. For lower mineral concentrations, we find that some combinations of NIR observation
bands permit reasonably good FLH retrievals in conditions where specific absorption spectral variation is not very high,
and that application of multi-spectral algorithms can be more efficient for the retrieval of fluorescence contributions in
coastal areas.
Improved remote sensing retrievals of the chlorophyll fluorescence component in coastal water reflectance can
significantly help environmental impact assessments. While retrieval of chlorophyll fluorescence from satellite
observations of open ocean reflectance using Fluorescence Line Height (FLH) algorithms is now routine, it is much
more complicated in coastal waters where the fluorescence overlaps with a NIR elastic scattering peak arising from the
combination of photosynthetic pigment and particulate scattering and absorption, and rapidly increasing water
absorption. To examine retrieval accuracies attainable in coastal waters by MODIS and other FLH algorithms, we
compared the results of extensive numerical simulations with those of our field measurements in the Chesapeake Bay.
The relationship between the contribution of fluorescence in the reflectance spectra and [Chl] and other water
constituents was analyzed by simulations of more than 1000 reflectances using the HYDROLIGHT radiative transfer
program. For these, IOP were related to parameterized microphysical models, following the same procedures used to
generate the IOCCG dataset, but with higher (1 nm) spectral resolution, and wider range of parameters including
chlorophyll specific absorption more typical of coastal waters. Results of simulations and field measurements show that
the variability of retrieved fluorescence can be attributed largely to its attenuation in the water by algae, CDOM and
mineral particles, and much less to the variation of the fluorescence quantum yield. Our systematic parametric study of
fluorescence as a function of the other water components is then used to define the range of water parameters where
fluorescence contributes significantly to the NIR peak reflectance, and where it is almost undetectable.
A prototype for an active backscattering probe for continuous backscattering spectrum measurement in sea water was designed and tested in both lab and field experiments. For the laboratory experiments, at various algae concentrations, the impact of overlapping chlorophyll fluorescence was effectively eliminated by using a long pass filter on the excitation light source. In the field tests with an active light source, and under the conditions of strong algal scattering (eg bloom conditions) it was shown that the impact of ambient background day light can be accounted for by successive measurements with the lamp illumination periodically on-off. The results show that the spectral shape of backscattering of algae cells is highly structured and consistent with Mie calculations which take into account the anomalous dispersion due to the strong absorption features of chlorophyll. The backscatter ratio, however, is found to be stable (within 10%) throughout the entire spectral region outside the chlorophyll fluorescence band, even when scattering is dominated by that from algae cells.
With the increasing recognition of the need for using the NIR bands for chlorophyll retrieval in coastal waters it is necessary to account not only for the spectral modulation of the total elastic backscatter by the chlorophyll absorption spectra, as it is normally done, but to also take into account the spectral signature of the backscatter itself, whether from mineral or organic particulates, including algae, and to assess how these factors effect retrieval algorithms. Based on our recent field measurements in coastal waters, we have undertaken a study to examine the spectral behavior of the backscatter to total scattering ratio as a function of suspended solids and chlorophyll loadings. The total scattering spectra is obtained using the WET Labs AC-S instrument which provides hyperspectral measurements of absorption and attenuation, in conjunction with the bb9 instrument which provides direct measurement of backscatter, as well fluorescence measurement of chlorophyll concentration [Chl]. The relevant WET Labs absorption and attenuation data were then used as input into Hydrolight radiative transfer simulations to obtain the backscattering ratio spectral distributions. Preliminary NIR algorithms, which were evolved for high [Chl] coastal waters and which focus on the contribution of spectral changes due to chlorophyll backscattering in the NIR, are presented. It is expected that these algorithms will ultimately prove to be less dependent on regional tuning.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.