Development of a MODIS data based algorithm for retrieving nearshore sea surface salinity along the northern Gulf of Mexico coast

Remote sensing algorithms for retrieval of Sea Surface Salinity (SSS) were generally developed for deep ocean waters while practical applications of SSS are commonly involved in coastal (particularly nearshore) waters. To fill the gap, this paper presents a new SSS algorithm, called Nearshore SSS Algorithm, developed using the Artificial Neural Networks toolbox in the MATLAB Program and 7years of cloud-free Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua data as well as in situ data collected from United States Geological Survey (USGS) stations along the Northern Gulf of Mexico coast. The independent input variables involved in the new algorithm include the water-leaving reflectance at 412nm, 443nm, 488nm, 555nm, 678nm, and 645nm of MODIS Aqua sensor. Results of sensitivity analysis indicate that SSS in nearshore waters is most sensitive to reflectance at 412nm and 488nm, followed by the reflectance at 555nm and 645nm. While reflectance at 678nm is also important, reflectance at 443nm is the least important to SSS in nearshore waters, providing new insights into SSS in coastal waters. MODIS data collected from January 2012 to September 2014 were used for validation of the nearshore SSS algorithm. The linear correlation coefficient of this algorithm for Louisiana coast was 0.7256 and 0.6985 for the development and validation, respectively. A unique feature of this paper is that while the new algorithm is particularly useful to the retrieval of SSS in nearshore waters, it is also expanded to further offshore waters, providing reliable SSS data for various applications of coastal environment and resources management particularly along the northern Gulf of Mexico coast.