Land Cover Classification Based On Space-borne Full Waveform Laser Altimetry Data

With the increasing impact of human activities on it,the Earth's environment has changed dramatically.One of changes in the Earth's environment is the type of surface cover.It is extremely important to study the classifications of surface cover due to the interaction between the environment and surface cover.Although there are many the researches on the classification of surface cover,the studies using Space-borne laser altimetry data to discover or monitor the alteration of space cover are limited.In this paper,after combing and summarizing the data and characteristics of Space-borne laser altimetry data,the advantages of full waveform Space-borne laser altimetry data in the application of surface cover classification are not only the elevation information of the surface can be obtained,but also the attribute information can be obtained,the height change monitoring of the same coverage type could be carried out,and the monitoring of the change of surface cover type should be extended from simulate surface to three-dimensional spatial information acquisition.Therefore,the study of surface cover classification by using on-board laser altimetry data can better reflect the change of surface mulch in three-dimensional space.The main research contents and contributions of this paper are as follows:(1)Use of ICESat/GLAS(Ice,Cloud and land Elevation Satellite/Geo-sciences Laser Altimeter System)data and CDF(Cumulative Distribution Function)curve matching algorithm has been studied for surface cover classification in Beijing area,the whole study is divided into four parts,and the surface cover types are divided into four categories,including bare land,vegetation,water bodies and buildings.The first part is sample training;The second part studies the classification of the surface coverage of existing algorithms in this area,and the accuracy of the 0.67,Kappa coefficient is 0.415;The third part is to explore the impact of topography on the classification effect,after removing the influence factors of topography,the classification accuracy of the 0.81,Kappa coefficient is 0.58;The last part of the study for the characteristics of classification types to improve the existing algorithm,based on the previous two studies on the basis of the combined reflection energy and wave number of the CDF curve improvement algorithm,the improved algorithm accuracy is 0.9,Kappa coefficient of 0.767,the classification accuracy is significantly improved than the existing algorithm.The reason of error classification of improved algorithm is further analyzed,and it is concluded that some errors classification of the improved algorithm are caused by the inconsistency between the validated data and the experimental data in the time series.(2)The adaptability of this algorithm is analyzed in the area near China Ross station in Antarctica.Due to the unsuitable acquisition of Antarctic classification data and the difference of time coupling,the classification accuracy of this method is not verified.However,by using GLAS data to have more overlapping spot in the Antarctic region,the surface cover classification and elevation change monitoring are studied,the covering type is combined with the overlay elevation information,the change of the feature type in the study area is analyzed yearly,and the analysis results are compared with the research results of other scholars.The potential application of Space-borne laser altimetry data in monitoring neighborhood of surface cover classification is proved.