Remotely sensed estimates of surface salinity in the Chesapeake Bay: A statistical approach

In coastal and estuarine environments, near-surface salinity varies significantly in space and time. As absolute salinity and salinity gradients are central to many physical and ecological processes in these environments, reliable and consistent salinity estimates are a priority for marine research and application communities. Satellite remote sensing has a great potential to meet this need, yet sensors and algorithms designed to monitor open ocean salinity are typically ill-suited for high resolution applications to coastlines and estuaries. Here we present results of multiple statistical models that predict daily, gridded surface salinity at 1 km resolution across Chesapeake Bay as a function of level 2 surface reflectance estimates from the NASA Moderate Resolution Imaging Spectroradiometer (MODIS), onboard the Aqua platform. Eight statistical methods were tested and it was found that sea surface salinity can be accurately predicted via remote sensed products with an accuracy that is more than sufficient for many physical and ecological applications. For the best-performing statistical model, mean absolute error was 1.82 relative to mean Chesapeake Bay salinity of 16.5.