基于模糊聚类与神经网络的风蚀危险性评估研究——以内蒙古自治区为例

Wind Erosion Hazard Assessment Based on Fuzzy Clustering and Neural Network——A Case Study of Inner Mongolia

土地沙化和沙尘暴是中国北方农牧交错带严重的生态环境问题。在地理信息系统的支持下,论文建立风蚀自然因子栅格数据库;选取二连浩特—张家口为典型样带,综合运用模糊聚类技术、专家知识以及神经网络技术,构建区域风蚀危险度评估模型,分析内蒙古自治区风蚀危险性空间分布格局。结果显示:综合应用模糊聚类以及神经网络技术可以有效地对风蚀危险度进行评价,内蒙古自治区风蚀危险性呈现"西高东低、中间过渡"的总体趋势,这与该区域的自然、气候与土地利用背景较吻合,其中,风蚀极险型区域主要分布在阿拉善高原的西北部,面积6.9×104km2;强险型土壤风蚀地区主要位于阴山山脉周围、穿越巴丹吉林沙漠、浑善达克沙地与锡林郭勒盟高原西北部,面积4.3×104km2;危险型风蚀区域大致位于极险和强险之间,面积是1.5×105km2;轻险型土壤风蚀区域主要位于内蒙古的中部,面积为7.5×105km2;风蚀无险型区域主要位于大兴安岭周围,面积1.4×105km2。

One of the most serious environmental problems in Farming-Pastoral Ecotone of Northern China is land desertification and sand storm.With the support of geographic information system,this paper establishes a natural factor database and takes Erenhot-Zhangjiakou as a typical transect to build regional wind erosion hazard assessment model using fuzzy clustering,expert knowledge and neural network technique,and then analyzes the spatial distribution of soil wind erosion hazard in Inner Mongolia;finally,the natural environment of spatial distribution of soil wind erosion hazard and land use background is studied according to land use type of the study area.Result shows： the integrated application of Fuzzy Clustering and Neural Network is an effective way to assess wind erosion hazard;the trend of wind erosion hazard in Inner Mongolia is increasing from east to west,which is consistent with the nature,land use and climate condition in the study area.The zone of severe hazard is mainly in the northwest Alashan with an area of 6.9×104 km2;the zone of intense hazard is mainly located around the Yinshan Mountains,northwest plateau of the Xilin Gol League,through the Badain Jaran Desert and Hunshandake,the area is 4.3×104 km2;the zone of moderate hazard is roughly located between the severe hazard and intense hazard,the area is 1.5×105 km2;the zone of slight hazard is mainly in the central Inner Mongolia,the area is 7.5×105 km2;and the zone of no hazard is located around Daxinganling with an area of 1.4×105 km2.