Amazon Basin Water Quality Monitoring Using Meris and Modis Data

The use of satellite data to monitor inland water quality in the Amazon basin is investigated. Both floodplain and river main channels are considered using the medium resolution sensors MODIS and MERIS A large database from the HYBAM project gives the water main properties for a large floodplain, 700 km upstream the Amazon river mouth. Parameters measured include the total suspended sediment (TSS), chlorophyll-a (Chla) and organic matter (OM) in particular. The analysis of 7 MERIS images acquired near 4 field measurement campaigns realized in 2003 and 2004 allows to understand the variability of the MERIS spectral profiles. We show that TSS and Chl-a affect the reflectance spectra and that the organic matter is causing strong absorption in the blue-green domain as well as over black waters. A reduced but significant temporal variability is observed which may be caused by the difference in viewing geometry. We propose a correction to reduce the geometry influence, and we develop a simple TSS retrieval algorithm based on linear regression segments to retrieve TSS for any flood stage. Accuracy assessment shows that the retrieval RMSE is of 50 mg/l when the TSS concentrations range between 0 and 800 mg/l. We demonstrate the potential of MERIS data for chlorophyll (Chl) retrieval and we present retrieval maps that are discussed in relation to the functioning of the floodplain lakes. Finally, we use the MERIS 15-band spectral to predict the MODIS profiles over the floodplain and the Amazon river. We demonstrate that MODIS 500-meter mode is as much as efficient for TSS retrieval that MERIS over nonproductive waters. Over waters presenting chlorophyll activity with reduced TSS load (inferior to 50 mg/l), we find that only MERIS performs efficiently. Using TSS measurements collected every 10 days in the Amazon river, we show that MODIS allow to the sedimentary fluxes at the river surface with an error of 30 mg / l.