基于自组织特征映射(SOFM)网络的农牧交错带景观分区——以内蒙古自治区为例

Landscape regionalization of agro-pastoral transitional zone using self-organizing feature maps(SOFM) network:A case study of Inner Mongolia Autonomous Region,China.

以我国北方景观分异最突出的内蒙古自治区农牧交错带为研究区域,运用自组织特征映射(SOFM)网络,基于归一化植被指数(NDVI)、土地利用程度综合指数和干燥度构建景观空间分异指数,定量识别研究区景观空间分异特征.结果表明:研究区可划分为森林景观区、林草过渡景观区、耕地-草地混合景观区、典型草原景观区、草原-荒漠过渡景观区和荒漠景观区6大景观区;与传统自然地理区划方案相比,该方法能够快速地对较大尺度景观空间分异进行定量划分,分区结果基本符合研究区自然景观特征及其空间分异规律.

Taking the agro-pastoral transitional zone of Inner Mongolia Autonomous Region, which has the most extrusive landscape differentiation in Northern China, as the object, a landscape spa- tial differentiation index was established based on the NDVI, integrated index of land use degree, and aridity index, and the characteristics of the landscape spatial differentiation were quantitatively identified by using SOFM network. The study area could be divided into six landscape zones, i. e. , forest zone, forest-steppe transitional zone, agro-steppe mixed zone, typical steppe zone, steppe- desert transitional zone, and desert zone. As compared with the traditional schemes of physical geo- graphical regionalization, our landscape regionalization could quickly regionalize the large scale landscape spatial differentiation in a quantitative way, and the regionalization results were basically in accordance with the original landscape characteristics and the landscape spatial differentiation pattern of the study area.