Combined Effect of Reduced Band Number and Increased Bandwidth on Shallow Water Remote Sensing: The Case of WorldView 2

WorldView 2 (WV2), launched in September 2009, is a satellite with hyperspatial resolution ( $\sim$ 0.5–2 m) capability for Earth surface observation. It has eight spectral bands with enhanced signal-to-noise ratio to cover the visible-to-near-infrared (V–NIR) domain, thus providing a great potential for remote sensing of coastal ecosystem, in particular, the aquatic environments with shallow bottoms (e.g., coral reefs and seagrass beds). Traditionally, it requires $\sim$ 15 spectral bands in the V–NIR domain for reliable analytical retrieval of bottom properties (e.g., bathymetry) from remotely observed radiance spectrum. Data from WV2, however, have eight bands, and the width of each band is quite wide ( $\sim$ 50 nm or more). Thus, the spectral configuration of WV2 is far from optimal for spectral remote sensing of various complex shallow environments, and it is important and necessary to know how such a band setting affects the reliability of remote-sensing retrievals. Here, we applied a hyperspectral optimization scheme [hyperspectral optimization processing exemplar (HOPE)] to a simulated shallow-bottom data set (sandy bottom) and compared retrievals from both hyperspectral and WV2 spectral settings. Retrieved results suggest that, for bottom contribution making up 40% or more of the measured signal, the depths derived from both hyperspectral and WV2 settings are generally consistent for waters shallower than 5 m. However, depths derived with WV2 setting have greater uncertainty and, in general, are shallower than those derived from the hyperspectral setting, particularly for waters deeper than 10 m. Options to produce higher confident properties from such band settings are discussed.