Measuring Three Dimensional Landscape Pattern Using DEM And LiDAR Data

Most of landscape patterns and ecological processes take place in three-dimensional surface,however,the traditional landscape analysis are conducted in two-dimensional space without the consideration of the third information of landscape such as topography undulation or the height of landscape components.Modern technological development of LiDAR,SAR or Aerial Oblique Photogrammetry have made the acquisition of the three dimensional information of human or natural landscape rapidly and conveniently,including DEM,DSM,3D Map or three dimensional vegetation structure.The tendency to three-dimensional realization of landscape has obliged the development of landscape pattern analysis to three-dimensional space.This proposal systematically studies the methods of landscape pattern analysis in 3D space based on the theoretical summary of 3D landscape pattern analysis,which will focus on 3D landscape metrics and 3D spatial statistics methods.The 3D landscape metrics include the derivation of 2D landscape pattern metrics to 3D space and the 3D landscape surface roughness metrics.The 3D spatial statistical methods mainly include the autocorrelation spectrum and Fourier power spectrum.1)The derivation algorithm of the 2D landscape metrics and 3D spatial statistical methods can be inferred,and the 3D landscape roughness metrics are introduced based on the principle deduction;2)The effectivity and adaptability of 3D landscape metrics and spatial statistical methods can be analysed using 3D landscape pattern metrics,autocorrelation spectrum,power spectrum and angular spectrum method based on the simulated landscape data.3)The features and adaptability of 3D landscape pattern methods in measuring natural landscape are analysed using the DEM and land use data in Jilin City,while the effectivity and adaptability of 3D methods in studying the dynamics of landscape pattern are analysed using the DEM,DSM and the vegetation height information extracted from the LiDAR data in Baicheng City.4)By the principal deduction and sample calculation,the characteristics and adaptabilities of the proposed methods are analysed and compared,and the principles,methods and algorithms of landscape pattern analysis in 3D space are determined with the release of programming codes.In the simulated landscape,the 3D landscape roughness metrics can extract the texture characteristics of terrain,and the surface texture aspect ratio index can be applied to determine the direction of the surface(isotropy or anisotropy).As for the simulated anisotropy terrain,no matter in landscape or class level,the variation characteristics of the 3D landscape pattern metrics are significant with the terrain fluctuation changing.Besides,the results of statistical methods indicate the texture aspect 45 degree in all the anisotropy terrain,and the autocorrelation spectrum reveals the dependency of land use on the terrain.In the Jilin landscapes,the results of the derivation of the 2D landscape metrics in landscape level indicate a stronger terrain undulation in landscape 1 compared with the landscape 2,while the results of 3D roughness metrics show an opponent trend in landscape level.Because the derivation algorithm takes the single patch as the calculation unit with the roughness algorithm single pixel,and the latter can be easily influenced by the extreme terrain undulation.The results of spatial statistical methods indicate the texture aspect 90 degree both in landscape 1 and 2,and the autocorrelation index can be applied to judge the texture characteristic as a supplementary parameter.In the Baicheng landscape,the results of the 3D roughness metrics indicate that the selected metrics are sensitive to describe the differences between the DEM and DSM,while the derivation metrics not.Because the landscape pattern metrics emphasize the homogeneity of patch,while the roughness metrics prefer the contribution of single pixel to the total landscape.Even the DSM adds the vegetation information compared with the DEM,the variation of the same patch type is similar.In the total,the results in simulated and natural landscapes show an effectivity and adaptability of the 3D landscape pattern measurement including the landscape pattern metrics and spatial statistical methods,which can be useful to quantify the landscape heterogeneity and study the dynamic characteristics of vegetation,land use and urbanization.