The Applicability Of The Gradient Histogram Matching Transformation Analysis

As an important terrain parameter, slope is the quantitative description of ground gradient. However, slopes extracted from DEM will be flattened as the resolution of the DEM becomes coarser. In order to overcome the scale effect of the slope reduction, slope transformation is necessary. Rescaling the coarser slope using histogram matching approach is an effective method. After rescaling, the slope can more accurately reflect the actual ground. At middle or small watershed scale, the slope rescaling is always carried out throughout the whole watershed. As the slope values often widely range and the terrain types usually varied for such a watershed, however, the reasonability of such slope rescaling should be taken into account. In the paper, the research corresponding to this problem was carried out in the Yanhe Watershed and Xiannangou Watershed. Based on the transform approach of slope histogram matching, small and middle scale watersheds were divided into several slope transformation units. By using the slope extracted from the high-resolution Hc-DEM as references, we analyzed the reasonability of this transform approach at small and middle scale watersheds from levels of points, lines and surfaces. The main contents and results of this research are as follows:(1) Dividing the slope transformation units at small and middle scale watersheds based on geomorphologic types as references. For Xiannangou Watershed, it was divided into sloping fields and downlands bounded by Divide line and shoulder line. For Yanhe Watershed, based on the geomorphologic types classification method proposed by David Jupp and the cartography principle as well, we divided the watershed into flat terrain region, small relief loess hilly region and large relief loess hilly region. The slope transformation units help to express the topographic features accurately, and provide accurate dataset for regional slope transformation.(2) Analyze the result of the slope transformation in levels of points, lines and surfaces. Set terrain feature points and feature lines inside the slope maps of middle and small scale watershed, and compare the relationship between the results from the slope transformation and the slope obtained from high resolution datasets. The results show that, the slope values all increase, the slope become steeper after rescaling, and slope reduction is corrected by slope scaling. The higher the resolution of the slope requiring for transforming is, the better the result is. As slope resolutions become coarser, the difference between the entirety transformation and regional transformation becomes larger. The slope values from regional transformation are closer to the reference value. The result from region transformation is better than that from entirety transformation. The analysis in surface level is carried out in two aspects, the entirety slope contrast and the classification contrast. The entirety slope contrast is implemented based on the analysis on slope statistical values, slope frequency curve, determination coefficient and difference maps. The results show that the region transformation is more reasonable than entirety transformation. The results of classification contrast show that the relief characteristics are exaggerated in the gentle slope level; best results were obtained in the middle slope level; in the steep slope level, terrain characteristic requires enhancing much more. The closer the average slope value obtained from the regional transformation is to the reference value, the better the result of the regional transformation is.