Hyperspectral remote sensing of shallow water environments: a review

Near-shore coastal waters are important for our quality of life, but near-shore environments are under continuous stress due to human activities and natural events. Also, data on some navigational charts are often decades old. Many do not reveal changes due to recent strong storms or coastal evolution. As a result, the doubtful bathymetric data can make coastal navigation dangerous. Methods and techniques are also needed to monitor the properties of near-shore waters as well as the condition of benthic ecosystems such a seagrass beds. Traditional ship-borne surveys are slow and expensive, and there is no need to remeasure all coastal areas as environmental change is in general a slow process. It is more important to find and flag the places where significant changes may have occurred and then focus field programs in those regions. One practical method may be to use satellite sensors, which have been proven very useful for quickly providing important environmental information over large areas. Satellite sensors for ocean studies use the relationship between the spectral signals received at the sensor and the contents below the sea surface to detect oceanic properties such as chlorophyll concentration. The use of passive spectral data for bathymetry was first demonstrated in the late 1960s (Polcyn and Sattinger 1969). Using several different approaches, reasonable results were obtained with limited spectral channels (Polcyn et al. 1970, Lyzenga 1985, Clark et al. 1987). All of these approaches require some important assumptions, however. These assumptions are not always valid. There is a strong need for a method to analytically and simultaneously derive bottom depth and albedo and the optical properties of the water column relying simply on remotely sensed data.