Supervised classification of water regions from lidar data in the Wadden Sea using a fuzzy logic concept

The Wadden Sea is an almost untouched area with a size of about 7300 km along the German, Dutch and Danish coast. Because of tide the area is flooded two times a day, creating a very special and sensitive ecosystem. In order to protect the Wadden Sea up-todate Digital Terrain Models (DTM) of high accuracy are needed to detect morphological changes. Lidar is an adequate method to obtain an accurate DTM. However Lidar is not able to penetrate water regions. Thus, raw Lidar data contain several water points, which do not belong to the terrain surface, leading to a wrong DTM. In this paper we present a supervised classification method to detect water regions from Lidar data using a fuzzy logic concept. Starting with raw data points of one strip, the points are grouped into scan lines. Based on training areas for the classes water and mudflat the features height, intensity and 2D point density are analysed. The significance level of the assumption that each feature differs for both classes is determined. Then, individual weights are derived from this significance level for every feature taking into account systematic feature changes depending on the angle of incidence of each laser pulse. A fuzzy logic classification is used to distinguish all points into water and mudflat points. Several additional steps are performed in order to refine and improve the classification result. Two meaningful examples are presented, which show the capability of this supervised fuzzy classification.