Full Waveform Analysis: Icesat Laser Data for Land Cover Classification

Analysis of the full waveform return pulse of laser altimeter systems is expected to increase the possibilities and accuracy in well-known applications of laser altimetry like DTM generation, forestry and earth surface analysis. NASA’s ICESat Geoscience Laser Altimeter System (GLAS) was launched in 2003 and acquires full waveform data along profiles covering the entire earth. In this study, the possibility of using ICESat data for land-cover classification is investigated over the Netherlands using data obtained from September to November 2003. The method, that automatically analyzes raw waveforms, divides the footprints of the laser pulses in four categories: high vegetation (high trees or forest), urban, water, and bare land/low vegetation. First, Gaussian components are fitted to and then several parameters derived from the Gaussian decomposition are used for discriminating the land-cover layers. Finally, the accuracy of the land-cover classes is validated by computing a confusion matrix based on the CORINE land cover database 2000 (CLC2000) covering the same study area. Using our method we get 73% accuracy of classification.