Classification of lidar data into water and land points in coastal areas

Over the last years lidar has become one of the major techniques to obtain spatial data in coastal areas. Due to the fact that lidar systems can provide several 3D points per square meter and high height accuracy, lidar data is suitable for several applications in the field of coastmonitoring and coastprotection. Generally, a digital terrain model (DTM) is used as basic spatial information for applications like morphologic change detection and hydrological modelling. In order to generate a DTM in coastal areas from lidar data, a classification process has to be performed to separate the lidar points into water and land points. Only land points, representing the coastsurface, are used to calculate the DTM. In this paper, we present a new method to classify lidar data in water points and land points. The original points of each flight strip are classified scan line by scan line. Several parameters which are directly related to each point as well as the point distribution within one scan line are used for the classification method. A fuzzy logic concept is applied to determine a membership value for every point belonging to the class water. Then, a threshold method is employed to classify the points of every scan line. Afterwards, classification discrepancies are detected and corrected by comparing height differences between neighboured water and non-water points. In order to achieve a more realistic classification result small isolated point groups of a certain class are removed. To illustrate the ability of the algorithm two examples with different characteristics (lidar scanner system, point density, point distribution etc.) are presented. The results are promising and constitute a proof-of-concept for the suggested method.