基于SOM神经网络模型的耕地非农化压力区域差异研究——以湖北省为例

Regional Differences of Farmland Conversion based on SOM Neural Network Model——A Case Study in Hubei Province

基于2006-2014年湖北省16市的面板数据,从经济因素、社会因素和资源禀赋三方面构建湖北省耕地非农化压力评价指标体系,运用SOM神经网络模型对湖北省的耕地非农化压力区域差异进行实证分析。结果表明:(1)湖北省16个主要城市的耕地非农化压力存在一定的区域差异。2006年以西南、东北地区往省内中部地区递减;2008年其区域差异变化不大;2010年则从西南-东北走向往边缘化发展;2012年呈现以武汉市、宜昌市为耕地非农化高压中心并辐射周边城市,形成了两条共同发展的平行线;直至2014年,武汉市耕地非农化压力趋于平缓,武汉市周边小城市的耕地非农化压力有小幅度上升。(2)SOM神经网络模型具有较强的聚类和容错能力,适用于开展耕地非农化压力的区域差异研究。研究结果反映了湖北省耕地非农化压力的空间异质性,为土地资源区际优化配置提供了基础。

This paper constructs the evaluation index system of farmland conversion pressure in Hubei province from economic factors,social factors and resource endowments,and uses SOM(Self-organizing feature Map)to gauge the pressure of farmland conversion in 16cities of Hubei province from 2006to 2014.The result shows that firstly,there are some regional differences of the farmland conversion pressure in 16cities of Hubei province.In 2006,this pressure showed decreasing trend from southwest,northeast to the central regions of this province.In 2008,there was little change in regional differences.In 2010,it showed the marginalized pattern.In 2012,it showed that Wuhan city,Yichang city are the centers,the pressure was decreasing to the surrounding cities.Until 2014,the pressure on farmland conversion in Wuhan tends to be gentle and some small cities around Wuhan have a slight rise.Secondly,the SOM neural network model with a strong ability of clustering and fault-tolerant can be used to classify farmland conversion pressure.The results reflect the spatial heterogeneity of the farmland conversion pressure in 16cities of Hubei province,which provides a scientific reference for the optimal allocation of land resources.