OPTICAL WATER TYPE BASED ALGORITHMS FOR COASTCOLOUR

摘要

Coastal zones comprise a large variety of water types with different optical properties ranging from clear to very turbid waters with a light penetration depth of only a few centimetres or to waters with high concentrations of organic substances, which makes the water nearly black. For remote sensing of this variety of water types and for determining the water constituents or optical properties from MERIS data we have compared three approaches: a neural network algorithm, which was trained with the reflectances of all water types, a quasi analytical algorithm and a classification of water types combined with a dedicated NN for each water type.