Classification Of Relief Form Based On The Best Terrain Factor Combination

The earth surface space becomes Landform after long-term evolution. The landform has many characteristics, the long-term continuity in time and the vast difference in space. The geomorphology mainly focused on researching the qualitative of landform type, but the collection of terrain data although using the basic field manual annotation that has big workload and low efficiency. The final result is that it difficult to obtain complete data of a landscape. With the rapid development of computer science, quantitative analysis of landscape has become the main mode of studying current geomorphology. This paper extracted a variety of terrain factor based on the 30 m DEM digital elevation data of study area, then selected the best classification of terrain factors, used classification method of remote sensing image, to synthesis a RGB image while the terrain factors selected is a single band. At last, using the method of unsupervised classification and supervised classification to realize the combination of the landform types in the study area.The main contents and conclusions of this paper are as follows:(1) Based on the DEM data that the resolution is 30 m in the study area, elevating 12 kinds of terrain factors, such as: elevation, slope, slope extraction rate, aspect, aspect rate, the amplitude of landforms, the total curvature, the roughness of ground surface, the cutting depth, the coefficient of variation of elevation and the contour line density. As the amplitude of landforms an example, the neighborhoods were extracted from the window size from 3×3 to 35 × 35 of the terrain data by using ArcGIS analysis method. And then analysis and processing the data by using the average value of method. Finally got the best analysis of optimal window size is 14 ×14 and the best statistical area is 0.1764km2.(2) Doing a qualitative analysis of each factor from the definition and significance, and the quantitative analysis on the terrain factors is by correlation coefficient method and Sheffield entropy method. Ultimately determine 7 indexes to division the types of landforms for the study of area. The 7 indexes are the elevation, the cumulative change rate, sloping extraction rate, DEM shading map, slope and amplitude of landforms.(3) Putting forward a classification system that fit Sichuan Province based on the classification of relief form of Sichuan Province and the actual situation of study area. The topography of the study are divided into 3 species and 6 small class. The species are basin, hilly and mountainous, dam, low hilly, deep hilly, respectively.(4) Proposed the method of classification of relief form based on combination the unsupervised classification and the supervised classification. The geomorphologic shape of the study area is divided, the overall classification accuracy is 74.15%, and the actual surface reference data is consistent with the degree of classification results of 68.37%. Finally, the distribution data of study area is statistical and a monographic geomorphology type map is made.