Retrieval Of Water Constituents Research Articles

Assessment of Estimated Phycocyanin and Chlorophyll-a Concentration from PRISMA and OLCI in Brazilian Inland Waters: A Comparison between Semi-Analytical and Machine Learning Algorithms

The aim of this work is to test the state-of-the-art of water constituent retrieval algorithms for phycocyanin (PC) and chlorophyll-a (chl-a) concentrations in Brazilian reservoirs from hyperspectral PRISMA images and concurrent in situ data. One near-coincident Sentinel-3 OLCI dataset has also been considered for PC mapping as its high revisit time is a relevant element for mapping cyanobacterial blooms. The testing was first performed on remote sensing reflectance (Rrs), as derived by applying two atmospheric correction methods (6SV, ACOLITE) to Level 1 data and as provided in the corresponding Level 2 products (PRISMA L2C and OLCI L2-WFR). Since PRISMA images were affected by sun glint, the testing of three de-glint models was also performed. The applicability of Semi-Analytical (SA) and Mixture Density Network (MDN) algorithms in enabling PC and chl-a concentration retrieval was then tested over three PRISMA scenes; in the case of PC concentration estimation, a Random Forest (RF) algorithm was further applied. Regarding OLCI, the SA algorithm was tested for PC estimation; notably, only SA was calibrated with site-specific data from the reservoir. The algorithms were applied to the Rrs spectra provided by PRISMA L2C products—and those derived with ACOLITE, in the case of OLCI—as these data showed better agreement with in situ measurements. The SA model provided low median absolute error (MdAE) for PRISMA-derived (MdAE = 3.06 mg.m−3) and OLCI-derived (MdAE = 3.93 mg.m−3) PC concentrations, while it overestimated PRISMA-derived chl-a (MdAE = 42.11 mg.m−3). The RF model for PC applied to PRISMA performed slightly worse than SA (MdAE = 5.21 mg.m−3). The MDN showed a rather different performance, with higher errors for PC (MdAE = 40.94 mg.m−3) and lower error for chl-a (MdAE = 23.21 mg.m−3). The results overall suggest that the model calibrated with site-specific measurements performed better and indicates that SA could be applied to PRISMA and OLCI for remote sensing of PC in Brazilian reservoirs.

Analysis of recurring patchiness in satellite-derived chlorophyll a to aid the selection of representative sites for lake water quality monitoring

Horizontal patchiness of water quality attributes in lakes substantially influences the ability to accurately determine an average condition of a lake from traditional in situ sampling. Therefore, spatial variability has to be accounted for in monitoring programmes which aim at determining the states and trends of ecosystem attributes. We used five years of Sentinel-2 Multispectral Instrument (MSI) data and conducted spatial analyses of surface chlorophyll a (Chl) concentration to map its variability and provide concrete recommendations for resource managers to design in situ sampling programmes. First, we developed a regional calibration of Chl predictions by C2RCC, an openly available processor for atmospheric corrections and water constituent retrieval, using in situ data from eleven temperate lakes in the central North Island of Aotearoa New Zealand. Using 93 match-up samples, we re-fitted C2RCC’s partitioning of constituent absorption coefficients to achieve an improved prediction accuracy for Chl (r2 = 0.79, root mean square error = 5.4 mg m−3). The new relationship was applied to all cloud-free images for thirteen regional lakes for further spatial analysis. We found that the medians calculated within areas of different sizes around in situ sampling locations may increase or decrease, illustrating an unpredictable uncertainty of the representativeness of any in situ sample. We went on to summarise five years of spatial variability by assessing each pixel for its tendency to be near the lake median Chl, higher (near the upper quartile) or lower (near the lower quartile). This spatiotemporal analysis revealed recurring patchiness that we converted to an indication of the representativeness of any location in the lake useful for the selection of more representative sites for future monitoring programmes.

Remote sensing retrieval of chlorophyll-α in inland waters based on ensemble modeling: a case study on Panjiakou and Daheiting reservoirs

Accurate remote sensing retrieval of chlorophyll-α (Chl-α) concentrations in inland waters raises a challenge due to the optical complexity of water constituents. Five Chl-α retrieval models, including single-band, band ratio, three-band, four-band, and partial least square models, were established with the measured spectra and Chl-α concentrations were measured at 36 stations in Panjiakou and Daheiting Reservoirs. To improve the Chl-α retrieval accuracy, three ensemble models, namely, entropy weight-based ensemble model (EW-EM), set pair analysis-based ensemble model (SPA-EM), and Bayesian model averaging-based ensemble model (BMA-EM), were developed for Chl-α retrieval with the weighted average of the five Chl-α retrieval models. All models were evaluated based on random calibration and validation samples. Ensemble modeling improved the Chl-α retrieval accuracy through integrating multiple Chl-α retrieval models. Compared to EW-EM and SPA-EM, BMA-EM could not only improve the Chl-α retrieval accuracy but also provide reliable confidence intervals for Chl-α retrieval. Ensemble modeling has application prospects in remote sensing retrieval of water constituents in inland waters.

Retrieval of Water Constituents from Hyperspectral In-Situ Measurements under Variable Cloud Cover—A Case Study at Lake Stechlin (Germany)

Remote sensing and field spectroscopy of natural waters is typically performed under clear skies, low wind speeds and low solar zenith angles. Such measurements can also be made, in principle, under clouds and mixed skies using airborne or in-situ measurements; however, variable illumination conditions pose a challenge to data analysis. In the present case study, we evaluated the inversion of hyperspectral in-situ measurements for water constituent retrieval acquired under variable cloud cover. First, we studied the retrieval of Chlorophyll-a (Chl-a) concentration and colored dissolved organic matter (CDOM) absorption from in-water irradiance measurements. Then, we evaluated the errors in the retrievals of the concentration of total suspended matter (TSM), Chl-a and the absorption coefficient of CDOM from above-water reflectance measurements due to highly variable reflections at the water surface. In order to approximate cloud reflections, we extended a recent three-component surface reflectance model for cloudless atmospheres by a constant offset and compared different surface reflectance correction procedures. Our findings suggest that in-water irradiance measurements may be used for the analysis of absorbing compounds even under highly variable weather conditions. The extended surface reflectance model proved to contribute to the analysis of above-water reflectance measurements with respect to Chl-a and TSM. Results indicate the potential of this approach for all-weather monitoring.

SNR (Signal-To-Noise Ratio) Impact on Water Constituent Retrieval from Simulated Images of Optically Complex Amazon Lakes

Uncertainties in the estimates of water constituents are among the main issues concerning the orbital remote sensing of inland waters. Those uncertainties result from sensor design, atmosphere correction, model equations, and in situ conditions (cloud cover, lake size/shape, and adjacency effects). In the Amazon floodplain lakes, such uncertainties are amplified due to their seasonal dynamic. Therefore, it is imperative to understand the suitability of a sensor to cope with them and assess their impact on the algorithms for the retrieval of constituents. The objective of this paper is to assess the impact of the SNR on the Chl-a and TSS algorithms in four lakes located at Mamirauá Sustainable Development Reserve (Amazonia, Brazil). Two data sets were simulated (noisy and noiseless spectra) based on in situ measurements and on sensor design (MSI/Sentinel-2, OLCI/Sentinel-3, and OLI/Landsat 8). The dataset was tested using three and four algorithms for TSS and Chl-a, respectively. The results showed that the impact of the SNR on each algorithm displayed similar patterns for both constituents. For additive and single band algorithms, the error amplitude is constant for the entire concentration range. However, for multiplicative algorithms, the error changes according to the model equation and the Rrs magnitude. Lastly, for the exponential algorithm, the retrieval amplitude is higher for a low concentration. The OLCI sensor has the best retrieval performance (error of up to 2 μg/L for Chl-a and 3 mg/L for TSS). For MSI, the error of the additive and single band algorithms for TSS and Chl-a are low (up to 5 mg/L and 1 μg/L, respectively); but for the multiplicative algorithm, the errors were above 10 μg/L. The OLI simulation resulted in errors below 3 mg/L for TSS. However, the number and position of OLI bands restrict Chl-a retrieval. Sensor and algorithm selection need a comprehensive analysis of key factors such as sensor design, in situ conditions, water brightness (Rrs), and model equations before being applied for inland water studies.

Correction of ozone influence on TOA radiance

An accurate atmospheric correction (AC) of Earth remote-sensing data in the spectral region 450–800 nm has to account for the ozone gas absorption influence. Usual operational AC codes employ a fixed ozone concentration corresponding to a climatologic average for a certain region and season, e.g. the mid-latitude summer atmosphere of the Moderate Resolution Atmospheric Transmission (MODTRAN) code. The reasons for a fixed ozone column are that ozone does not vary rapidly on a spatial and temporal scale, and additionally, the look-up table (LUT) size for AC is already big. This means that another degree of freedom for the ozone parameter would dramatically increase the size of the LUT database and the time required for LUT interpolation. In order to account for this effect, we use already existing LUTs that were calculated for a certain ozone reference level, e.g. an ozone column of g = 330 Dobson Units (DU) for MODTRAN’s mid-latitude summer atmosphere. Then the deviation of the top-of-atmosphere (TOA) radiance ΔL(g) from the reference level L(g = 330) is calculated as a function of solar and view geometries. The calculation is performed for a set of 36 wavelengths in the ozone-sensitive spectrum (450–800 nm) and five ozone columns. The last step computes the linear regression coefficients for each wavelength and geometry. The results are stored in a small table (11 kB). It is shown that the ozone influence is accurately accounted for by multiplying the modelled radiance L(g = 330) with a factor depending on g and wavelength yielding TOA radiance relative errors smaller than 0.5% for a wide range of ozone concentrations between 180 and 500 DU. Selected examples of a sensitivity study of the ozone effect on the retrieval of water constituents demonstrate the need to account for ozone in the AC step.

Optimizing a remotely sensed proxy for plankton biomass in Lake Kivu

Many regions of the world, including inland lakes, present with suboptimal conditions for the remotely sensed retrieval of optical signals, thus challenging the limits of available satellite data-processing tools, such as atmospheric correction models (ACM) and water constituent-retrieval (WCR) algorithms. Working in such regions, however, can improve our understanding of remote-sensing tools and their applicability in new contexts, in addition to potentially offering useful information about aquatic ecology. Here, we assess and compare 32 combinations of two ACMs, two WCRs, and three binary categories of data quality standards to optimize a remotely sensed proxy of plankton biomass in Lake Kivu. Each parameter set is compared against the available ground-truth match-ups using Spearman’s right-tailed ρ. Focusing on the best sets from each ACM–WCR combination, their performances are discussed with regard to data distribution, sample size, spatial completeness, and seasonality. The results of this study may be of interest both for ecological studies on Lake Kivu and for epidemiological studies of disease, such as cholera, the dynamics of which has been associated with plankton biomass in other regions of the world.

Leveraging EO-1 to Evaluate Capability of New Generation of Landsat Sensors for Coastal/Inland Water Studies

Monitoring coastal and inland waters, recognized as case II waters, using the existing Landsat technology is somewhat restricted because of its low signal-to-noise ratio (SNR) and its relatively poor radiometric resolution. The new generation of Landsat, Landsat Data Continuity Mission (LDCM) carrying the Operational Land Imager (OLI), has enhanced features allowing for a more lucid characterization of water constituents in either coastal or inland waters with respect to Landsat-7 (ETM+). This paper applies a physics-based approach to fully examine the potential of OLI in terms of its enhanced features in a water constituent retrieval framework. An EO-1 dataset, including Hyperion and the Advanced Land Imager (ALI), together with nearly coincident ETM+ imagery were atmospherically corrected using a data-driven approach. An in-water radiative transfer model, i.e., Hydrolight, was applied to generate a Look-Up-Table (LUT) of simulated surface reflectances for various combinations of water constituents. Using the Hyperion-derived concentration maps as validation sources, it was found that the simulated OLI imagery is superior to ETM+ on the order of 40%, 20%, and 28% when retrieving the concentrations of chlorophyll-a and total suspended solids (TSS), as well as the absorption of the colored dissolved organic matter (CDOM), respectively. It was also demonstrated that the simulated OLI imagery outperforms the simulated ALI and the recorded ALI datasets in the retrieval of chlorophyll-a and CDOM absorption. It is concluded that the new generation of Landsat enables mapping and monitoring of case II waters with accuracies not achieved with the previous Landsat satellite series.

Impact of computational methods and spectral models on the retrieval of optical properties via spectral optimization

Spectral optimization algorithm (SOA) is a well-accepted scheme for the retrieval of water constituents from the measurement of ocean color radiometry. It defines an error function between the input and output remote sensing reflectance spectrum, with the latter modeled with a few variables that represent the optically active properties, while the variables are solved numerically by minimizing the error function. In this paper, with data from numerical simulations and field measurements as input, we evaluate four computational methods for minimization (optimization) for their efficiency and accuracy on solutions, and illustrate impact of bio-optical models on the retrievals. The four optimization routines are the Levenberg-Marquardt (LM), the Generalized Reduced Gradient (GRG), the Downhill Simplex Method (Amoeba), and the Simulated Annealing-Downhill Simplex (i.e. SA + Amoeba, hereafter abbreviated as SAA). The Garver-Siegel-Maritorena SOA model is used as a base to test these computational methods. It is observed that 1) LM is the fastest method, but SAA has the largest number of valid retrievals; 2) the quality of final solutions are strongly influenced by the forms of spectral models (or eigen functions); and 3) dynamically-varying eigen functions are necessary to obtain smaller errors for both reflectance spectrum and retrievals. Results of this study provide helpful guidance for the selection of a computational method and spectral models if an SOA scheme is to be used to process ocean color images.

Integrating Landsat-7 Imagery with Physics-based Models for Quantitative Mapping of Coastal Waters near River Discharges

Remote sensing has traditionally been used to retrieve water constituents by establishing a relationship between in- situ measured quantities and image-derived products. Motivated by the dramatically improved potential of the Landsat Data Continuity Mission (LDCM), this paper describes a different approach for water constituent retrieval where both thermal and visible spectral bands of the Enhanced Thematic Mapper Plus (ETM+) instrument on board Landsat-7 are utilized. In this effort, Landsat data is integrated with a 3D hydrodynamic model to obtain profiles of particles and dissolved matter in the near shore zone in the vicinity of two river discharges. The procedure is based upon performing many hydrodynamic simulations by adjusting input environmental/physical variables and generating Look-Up-Tables (LUTs). This is conducted in two phases, namely the model calibration and the constituent retrieval. In the calibration phase, the best model output is determined by searching the LUT for the optimal surface temperature map compared to the Landsat-derived surface temperature map. The profiles of particles and dissolved matter are retrieved in the second step by comparing several modeled surface reflectance maps with atmospherically compensated Landsat-7 imagery. Various case scenarios of simulated water constituent profiles drive an in-water radiative transfer code, i.e. Hydrolight, which simulates water-leaving reflectance ( d r ). The best match, obtained via optimization, demonstrated an average root-mean-squared-error (RMSE) of 0.68%, i.e., 0.0068 reflectance units, calculated over the two river plumes. It is concluded that calibrating a physics-based model using the Landsat-7 imagery can provide a more lucid insight into the dynamics of spatially non-uniform waters. Ongoing efforts show that, due to its enhanced radiometric fidelity, the LDCM should significantly improve our proposed approach for the retrieval of water constituents.

Characterizing the relative calibration of Landsat-7 (ETM+) visible bands with Terra (MODIS) over clear waters: The implications for monitoring water resources

Since its launch, the Enhanced Thematic Mapper plus (ETM+) onboard Landsat-7 has been continuously monitored via different calibration techniques to ensure it maintains the science requirements for demanding application areas. The majority of its applications, including agriculture and forestry, require a robust calibration for medium to high reflective targets. However, when imaging water resources, then the question becomes whether the calibration coefficients are valid for the dark end of the sensor's responsivity curve. Motivated by the Landsat Data Continuity Mission (LDCM) and its potential for providing long-term, robust water studies, in this effort, the calibration status of ETM+ visible bands are examined using a cross-calibration technique. The well-calibrated Terra-MODIS scenes (collection 5) of the past decade over relatively optically stable waters were chosen to evaluate ETM+ stability. The cross-comparison showed that the calibration instability of ETM+ reflective bands lie well within its radiometric uncertainty. The slight calibration differences were found to be less than 1.6%, 0.93%, and 3.2% for the blue, green and the red bands obtained for the period when the MODIS had been radiometrically stable. The NIR band of ETM+, however, exhibits, on average, 4.8% higher signals than those of MODIS. The error budget analysis for the calibration differences showed that 1.2%, 1.6%, 3.5%, and 6.5% errors are associated with the Visible-Near-Infrared (VNIR) bands, respectively. Using the results from the calibration study combined with simulations, it was demonstrated that ETM+ underestimates the retrieved diffuse surface reflectance (Rd) in the blue and the green bands as much as 12.2% and 4.4%, respectively, while the red band is overestimated 37%, on average, when studying slightly/moderately turbid. The uncertainties in the retrieval of Rd were applied in a water constituent retrieval framework where a physics-based model was utilized to obtain concentrations of chlorophyll-a (CHL) and total suspended solids (TSS) from a Landsat-7 dataset in slightly/moderately turbid waters. It was found that the calibration-induced errors in the retrieval of Rd yield, on average, 20% error in retrieval of these components. The results indicate that although ETM+ is well calibrated, its calibration status should be quantified rigorously for water studies when physics-based methods are employed for the removal of atmospheric effects. The over-water characterization of Landsat satellites will become more crucial when the LDCM becomes operational due to its increased capabilities for water resource studies.

Correction of sun glint effect on MIVIS data of the Sicily campaign in July 2000

To assess the suspended and dissolved matter in water in the visible and near infrared spectral regions it is necessary to estimate with adequate accuracy the water leaving radiance. Consequently radiance measured by a remote sensor has to be corrected from the atmospheric and the sea surface effects consisting in the path radiance and the sun and sky glitter radiance contributions. This paper describes the application of the sun glint correction scheme on to airborne hyperspectral MIVIS measurements acquired on the area of the Straits of Messina during the campaign in July 2000. In the Messina case study data have been corrected for the atmospheric effects and for the sun-glitter contribution evaluated following the method proposed by Cox and Munk (1954, 1956). Comparison between glitter contaminated and glitter free data has been made taking into account the radiance profiles relevant to selected scan lines and the spectra of different pixels belonging to the same scan line and located out and inside the sun glitter area. The results show that spectra after correction have the same profile as the contaminated ones, although, at this stage, free glint data have not yet been used in water constituent retrieval and consequently the reliability of such correction cannot be completely evaluated.

Mapping of water constituents in Lake Constance using multispectral airborne scanner data and a physically based processing scheme

A physically based method to determine water constituents in case II waters from imaging spectrometer data is presented. The processing scheme includes modules for sunglitter correction, aerosol retrieval, atmospheric correction, and the retrieval of water constituents from subsurface reflectance spectra in case II waters. For the aerosol retrieval, the sun glitter, and the atmospheric correction modules, an azimuthally resolved radiative transfer model for a multilayer atmosphere-ocean system with a flat water surface is used. The retrieval of water constituents is achieved by an iterative fitting algorithm to adjust modeled and measured underwater reflectances. The processing scheme is applied to airborne multispectral scanner Daedalus AADS1268 in ATM (airborne thematic mapper) modus. Owing to the sensor properties (reduced radiometric sensitivity of the blue channel), it was not possible to work completely without additional data; Gelbstoff absorption could not be retrieved. It had to be inputted and was kept constant over the mission area. The processing system has been tested for multitemporal mapping of water constituents at Lake Constance. The results show the system as an efficient and reliable tool for the inversion of remote sensing data from natural waters.