气候变化领域集成服务门户

Optical data collected in coastal waters off South Florida and in the Caribbean Sea between January 2009 and December 2010 were used to evaluate products derived with three bio-optical inversion algorithms applied to MODIS/Aqua, MODIS/Terra, and SeaWiFS satellite observations. The products included the diffuse attenuation coefficient at 490 nm (K-d_490) and for the visible range (K-d_PAR), and euphotic depth (Z(eu), corresponding to 1% of the surface incident photosynthetically available radiation or PAR). Above-water hyperspectral reflectance data collected over optically shallow waters of the Florida Keys between June 1997 and August 2011 were used to help understand algorithm performance over optically shallow waters. The in situ data covered a variety of water types in South Florida and the Caribbean Sea, ranging from deep clear waters, turbid coastal waters, and optically shallow waters (K-d_490 range of similar to 0.03-1.29 m(-1)). An algorithm based on Inherent Optical Properties (IOPs) showed the best performance (RMSD<13% and R-2 similar to 1.0 for MODIS/Aqua and SeaWiFS). Two algorithms based on empirical regressions performed well for offshore clear waters, but underestimated K-d_490 and K-d_PAR in coastal waters due to high turbidity or shallow bottom contamination. Similar results were obtained when only in situ data were used to evaluate algorithm performance. The excellent agreement between satellite-derived remote sensing reflectance (R-rs) and in situ R-rs suggested that the different product uncertainties resulted primarily from algorithm inversion as opposed to atmospheric correction. A simple empirical model was developed to derive Z(eu), from K-d_490 for satellite measurements of nearshore waters. MODIS/Aqua gave the best results in general relative to in situ observations. Our findings lay the basis for synoptic time-series studies of water quality in coastal ecosystems, yet more work is required to minimize the bottom interference in the Florida Keys optically shallow waters. (c) 2012 Elsevier Inc. All rights reserved.