DEM Based Research On Terrain Texture Of The Loess Plateau

The geomorphic characteristics of "thousands of gullies" in the Loess Plateau show significant self similarity in multi-scale space,and have obvious texture characteristics of local-irregular and macro-regular.The existing texture analysis for the Loess Plateau is based on the statistical texture characteristics.The structural characteristics and quantitative research on the typical landform of the Loess Plateau are relatively lacking.In this study,the Loess Plateau is taken as the research area and two terrain textures are proposed.One texture is macroscopic texture composed of geomorphic units characteristic.The other texture is microscopic texture composed of slope surface and slope variability.In terms of the method,the calculation of multidirectional and normalized regularity and main period based on the accumulated distance matching function(SDMF)is proposed.This method can be used to extract structural features of terrain texture in different scales.In the application,this paper uses statistical and structural texture features to extract and identify erosion gullies and terraces in the typical small watershed of the Loess Plateau combined with the object-oriented analysis framework on the micro scale.On the macro scale,this paper analyzes and identifies the structural features of the yuan,liang and mao represented by positive and negative topography.This study tries to develop new methods of digital terrain analysis and promote the development of GIS.The main contents and research results of this paper are as follows:(1)The concept framework of the terrain texture of the Loess Plateau is proposed.And the basic concepts of terrain texture and Loess Plateau terrain texture are also proposed.Under the existing classification framework system,the macroscopic texture composed of geomorphic units(loess yuan unit,loess liang unit,and loess mao unit)characteristic and the microscopic texture composed of slope surface and slope variability are proposed.The combination characteristics of natural texture(random characteristics)and approximate regular texture(structural characteristics)are analyzed.This paper expounds the basic characteristics,division system,data expression and formation process of the terrain texture of the Loess Plateau,and sums up scientific problems.(2)In terms of methods,this paper mainly proposes the calculation method of regular degree based on SDMF,as well as the method of extracting the first main period and multi-directional and normalized regular degree.In feature space,SDMF features can be calculated by significant peaks.By calculating the spatial size and the regular degree of spatial distribution of texture elements in a certain scale,the regular degree can be used as the structural quantitative characteristics of the terrain texture of the Loess Plateau.Combined with GLCM,the analysis framework of terrain texture of Loess Plateau is formed.(3)Based on the texture analysis framework,this paper analyzes and identifies the terrain texture of typical small watershed on the micro scale.The erosion gully area of typical small watershed of loess has significant random texture characteristics,while terrace has significant approximate regular texture characteristics.In this paper,GLCM method combined with SDMF regular degree can effectively represent the erosion ditch and terrace area.The extracted boundary of terraces is optimized by the superposition analysis of chessboard segmentation and multi-scale segmentation.In the extraction experiment,three sample areas,Ansai,Changwu and Suide,were tested.The extraction average accuracy of erosion ditch was 85.78%,and that of terrace extraction was 86.36%.(4)Based on the texture analysis framework,the typical sample areas of the Loess Plateau are quantified and identified on the macro scale.Based on GLCM,this paper analyzes the influence of gray level and direction parameters on the terrain texture of DEM data.At the same time,this paper uses the geometric center expansion method to analyze the appropriate window of texture statistical features.GLCM contrast has been proved to be the most distinguishing and effective feature.64 level gray level is a stable and significant quantitative value of different categories.In addition,based on SDMF method,this paper makes quantitative analysis on hilly and gully areas.This method uses the normalized SDMF features of positive and negative topography to calculate the regularity,which can quantify and identify the Loess Hilly and gully area.The average regularity(0.64)of the liang hilly and gully area is larger than that of the mao hilly and gully area(0.48).Based on the above quantitative analysis,a three-layer quantitative recognition model of global characteristics,local morphology and local structure is constructed.The model is used to identify 7 types of Typical Landforms in the Loess Plateau with an average recognition rate of 81.83%.In this paper,the concept model of the terrain texture of the Loess Plateau is put forward,and the analysis framework of the structure and statistical characteristics of the terrain texture of the Loess Plateau is preliminarily explored.The analysis framework can be applied to the recognition and classification of multi-scale loess terrain texture,and promote the method of digital terrain analysis of Loess Plateau.