Remote Estimation of Chlorophyll-a in Coastal Waters Using Red and near Infrared Spectral Regions

Recent advances in the development of the atmospheric correction models made the retrieval of surface reflectance spectra of coastal waters from the top of atmosphere signals more accurate and inspired the further development of the coastal retrieval algorithms. This includes algorithms which employ the red and NIR bands and which are less sensitive to the absorption of the colored dissolved organic matter (CDOM) and scattering of mineral particles than traditional blue-green ratio algorithms. Simple Red-NIR band ratio algorithms are known for a long time [1,2] and successfully used for the estimation of chlorophyll a [Chl] concentrations from reflectance spectra in coastal and inland waters [3], as [Chl] is above 5 mg/m when peak around 700 nm is becoming pronounced. Generally this peak includes two components: i) elastic one which is due to the local minimum of absorption coefficient as a result of the confluence of the phytoplankton and water absorption spectra, and ii) chlorophyll fluorescence with the maximum near 685 nm [2,4,5]. For [Chl]<5mg/m elastic component is small and the magnitude of the peak and [Chl] are mostly related to the fluorescence signal. For [Chl]>5mg/m both components contribute to the signal with the peak of elastic component and combined reflectance moving to the red with increasing [Chl]. Advanced version of Red-NIR algorithms includes 3 bands instead of 2 [6], which assumes better separation of the phytoplankton absorption related to [Chl] from other absorption and scattering components in the red and NIR part of the spectrum. Both 2 and 3 bands algorithms have been tested in the multiple water environments [3,7,8] but the origins of the uncertainties were difficult to trace because multiple unknown parameters involved were not measured directly in the experiments. Optimization of the band positions was done on the synthetic datasets simulated using semi-analytical model [9], however, sensitivity of the retrievals to the CDOM absorption, concentration of minerals, phytoplankton absorption spectra were not considered. In this work we tested 2 and 3 bands algorithms with the bands available on MODIS and MERIS satellite sensors using comprehensive synthetic datasets of reflectance spectra and inherent optical properties (IOP) related to a wide range of water parameters as well as a new field dataset.