关键词: 榆林市, 土壤风蚀, T-S模糊神经网络, 危险度评价

Abstract:

：Soli wind erosion, one of the major causes of environmental problems such as farmland degradation and desertification, producing large scale of sandstorm, polluting the environment and harming human beings′ health in arid and semi-arid regions, is the detachment, entrainment, transport deposition of soil particles caused by wind. It is also a serious socio-economic issue which limits the development in those ecologically fragile areas. In the previous studies, methods for accessing wind erosion are based on land use information. In this article, however, another approach that uses GIS and T-S fuzzy neural network model to access soil wind erosion risk is presented. The approach is applied to a case study of the Yulin City located on the wind-sand transition regions of Mu Us sandy land, Inner Mongolian plateau and loess plateau, northwest China, where most serious expansion of desertification is proceeding. Environmental elements of soil wind erosion were selected and their spatial distribution data were extracted using GIS technology and T-S fuzzy neural network model was constructed after training the sample data of different wind erosion risk level. The model was applied to mapping wind erosion risk after inputting the extracted date. The entire jobs were based on the software of ArcGIS9.3 and MATLAB. The results show that: Firstly, the comparison of the risk assessment with land use information in Yulin City demonstrates that the result is in conformity with actual conditions. T-S fuzzy neural network model can reveal the relationship between risk of soil wind erosion and its environment effectively, and provide the basis for predicting wind erosion. Secondly, environmental elements such as wind energy, vegetation coverage, temperature, precipitation, topography control the distribution properties of wind erosion risk in Yulin City. Thirdly, there are 4 divisions of wind erosion risk in Yulin City: the severe risk level, the intense risk level, the moderate risk level and the slight risk level, and the area of different risk level are 4 810 km², 5 824 km², 13 239 km² and 19 234.88 km². Finally, the distribution of soil wind erosion risk is decreasing gradually from southeast to northeast in Yulin City. Moreover, the assessment of wind erosion risk and its distribution has significant meaning to control desertification and develop economic in wind-sand transition regions.

Key words: Yulin City, soil wind erosion, T-S fuzzy neural network model, risk assessment