Algorithms for Remote Sensing Estimation of Chlorophyll-a in Turbid Productive Waters using Red and Near Infrared Bands
Gitelson, Anatoly1; Gurlin, Daniela1; Moses, Wesley1; Gilerson, Alex2; Yacobi, Yosef3
1University of Nebraska, UNITED STATES; 2CUNY, UNITED STATES; 3IOLR, ISRAEL
Advances in the development of the atmospheric correction models have made the retrieval of surface reflectance spectra of coastal waters from the top of atmosphere signals more accurate and inspired the development of advanced retrieval algorithms for estimating chlorophyll-a (Chl-a) concentration in coastal and inland waters. This includes algorithms that employ the red and NIR spectral bands, which are less sensitive than the traditionally used blue-green ratio algorithms to the absorption by colored dissolved organic matter (CDOM) and scattering by mineral particles. We tested such algorithms using comprehensive synthetic datasets of reflectance spectra and inherent optical properties (IOP) related to various water parameters and compared the results with those obtained from field measurements, MERIS and MODIS satellite imagery.
Over 2000 reflectance spectra were simulated using the radiative transfer program HYDROLIGHT, with 1 nm resolution for conditions that are typical for inland and coastal waters. Simulations were based on the findings of many authors for IOP characteristics, and were similar to the assumptions used in the construction of the IOCCG datasets. Field measurements were collected from several coastal and inland water bodies in the USA (lakes in Nebraska, the Chesapeake Bay, the Long Island Sound, lakes in the vicinity of the New York City, and the Hudson/Raritan Estuary), Sea of Galilee (Israel), and Azov Sea (Russia). Field spectrometers were used to collect at-surface reflectance spectra. Water samples were taken and analyzed in the laboratory to retrieve the concentrations of Chl-a and mineral particles through analytical procedures and the IOPs of water (absorption and attenuation) through spectrophotometric procedures. Two-band NIR-red models based on the spectral channels of MODIS (R748/R667) and MERIS (R753/R665 and R708/R665), and a three-band NIR-red algorithm based on the spectral channels of MERIS (centered at 665, 708 and 753 nm) were tested. The MERIS-based two-band NIR-red model R708/R665 was found to be very robust in estimating Chl-a concentrations for a wide range of water conditions, with a minimal sensitivity to CDOM concentration and fluorescence quantum yield. Algorithms that include NIR bands near 750 nm were significantly less accurate. Algorithms established using proximally sensed data were applied to aircraft and satellite data. The algorithms yielded high accuracies in estimating Chl-a concentrations in diverse water bodies and did not require regional re-parameterization. The results illustrate the tremendous potential of the NIR-Red models based on the spectral channels of MERIS to estimate Chl-a concentration in turbid productive waters, and provide an indication of the excellent results expected from the future OLCI sensor, which contains all the spectral channels of MERIS.