国家尺度耕地变化驱动力的定量分析方法  

Quantitative analysis approaches to the driving forces of cultivated land changes on a national scale

作　　者：刘旭华[1] 王劲峰[1] 刘纪远[1] 刘明亮[1] 孟斌[1] 

机构地区：[1]中国科学院地理科学与资源研究所

出　　处：《农业工程学报》2005年第4期56-60,共5页Transactions of the Chinese Society of Agricultural Engineering

基　　金：国家重大基础研究计划(2001CB03-5);国家863计划(2002AA135230-1)

摘　　要：该文使用中国1987～1989年和2000年两期土地利用遥感调查数据,探索了国家尺度土地利用变化驱动力分析的定量研究方法。以全国耕地变化为例,首先通过GIS和遥感技术对全国耕地变化与自然、社会经济状况进行综合分区,然后利用人工神经网络对各类型区的耕地变化的主导因素进行了分析。研究发现,东部沿海地区及四川盆地自1987年到2000年发生的耕地流失严重,城市化导致的城镇人口增长进而导致城镇扩张、市场经济条件下区域经济之间的相互作用、第二、三产业的迅速发展是其主导原因。在变量选取中大样本的神经网络训练得到较好结果,表明神经网络在土地利用变化驱动力研究中具有强大能力,并说明了先分区后筛选变量的研究思路的合理性。该研究方法对其他类型的土地利用变化驱动力研究以及全球变化研究具有借鉴作用。By using digital satellite remote sensing data of China acquired in late 1980s and 2000 respectively, this paper explores the analysis approaches in the study of the driving forces of land use changes in large regions. At first an integrated regionalization was made based on the cultivated land changes, the natural situation and socio-economic changes using GIS and RS techniques. Then using BP neural network, the leading driving forces causing the decrease of gross cultivated land area were found in one class of the above regionalization as a case. In the case study area, the primary contributors are the rapid urbanization, the external forces from around the region namely urban influence, and the rapid development of second and third industry under market economy condition of China. A good result is achieved under 771 samples through BP neural network training, which indicates that neural network technique has a big power in the study of the driving forces of land use change. Meanwhile, it is proved to be feasible and reasonable to use the solution of regionalization first and then filtering variables. The quantitative research methods provide references for the study of other types of land use change in large areas.

关 键 词：耕地利用变化 驱动力 定量研究 

分 类 号：F301.24[经济管理—产业经济]