关键词: 自组织特征映射网络（SOM网络）, 耕地压力分类, 地域差异, 中国

Abstract: Cultivated land is one of the most important natural resources in China. In order to find the way for the sustainable cultivated landuse and to overcome the default of regular cultivated landuse pressure, a logging theological identification technology is introduced and the principle of SOM (selforganizing mapping) network is summarized in this paper. Then a set of cultivated landuse pressure evaluation indicators based on the indicator selection principles are set up according to the cultivated land current condition in China in terms of economy, society and ecology. A program is written, and MATLAB 6.1 software is employed to measure the cluster of regional cultivated land use pressure. The run results of SOM show that 31 provinces (cities) or autonomous regions are classified into various categories on the condition of different steps. The regional cultivated land pressure level disparities are consistent with the economic regional difference. Cultivated land pressure status estimated on SOM can explore distinctively the reason and the results of cultivated landuse pressure changes, which will help administrators to adopt suitable landuse policy and management measures to alleviate cultivated land use pressure and improve cultivated land quality. The evaluation results also indicate that the application of SOM neural network in assessing regional differentiations of cultivated land use pressure level without assuming parametric relationship is convenient, precise and feasible, which can be an alternative approach of assessing regional differentiations of cultivated land use pressure level.

Key words: selforganizing mapping network(SOM network), cultivated landuse pressure cluster, regional difference, China