下辽河平原典型区耕地格局变化及其空间要素驱动

为明确粮食主产区耕地格局动态变化特征及其空间驱动因素。该研究以下辽河平原典型区沈阳市为研究区，运用变化轨迹分析法、重心迁移模型和景观格局指数，揭示1980-2000年和2000-2018年耕地格局变化特征，并从空间要素角度出发，运用随机森林算法识别不同区域尺度耕地格局变化空间驱动因素。结果表明：1）1980-2000年耕地格局相对稳定，耕地重心向西南方向迁移；2000-2018年耕地流失严重，耕地重心向西北方向迁移，迁移速度加快。2）2 个研究时段内耕地变化均具有转入转出双向特征，1980-2000年耕地转入与耕地转出速度缓慢，2000-2018年耕地转出面积远高于耕地转入面积，耕地转出区域以城区周边为主。3）不同区域尺度空间驱动因素重要性大小存在差异，空间距离要素、地理空间要素和政策因素对耕地格局变化起主要驱动作用，高程和坡度对耕地格局变化的影响较小。研究证实了随机森林算法能够有效量化耕地格局变化空间驱动因素，研究结果可为构建耕地保护长效机制提供借鉴与参考。

Changes of cultivated land pattern and its spatial driving factors in the typical regions of Lower Liaohe Plain

Abstract: Cultivated land has always been one of the most important fundamentals to maintain food security. This study aims to clarify the dynamic change of the cultivated land pattern and the spatial driving factors in the main grain producing areas in China. The study area was also taken as the Shenyang city, a typical area in the Lower Liaohe Plain. The GIS technology and R programming were utilized to collect the data sources from 1980 to 2000 and 2000 to 2018, such as land use, point of interest (POI), and the delimitation of permanent basic farmland. The cultivated land pattern was then obtained using the change trajectory analysis, the center of gravity migration, and the landscape pattern index. The representative driving factors of cultivated land pattern were extracted from the natural conditions, spatial distance, geographic space, and policy elements. A random forest was used to identify the degree of significance of the spatial driving factors during the variation in the cultivated land pattern at the spatial scale of city, urban and county areas. The results showed that: 1) The pattern of cultivated land was relatively stable from 1980 to 2000, where the center of gravity of cultivated land moved to the southwest by 1.418 0 km. There was a serious loss of cultivated land from 2000 to 2018, where the center of gravity of cultivated land moved 2.32 0 km to the northwest, indicating an accelerated migration speed. 2) The changes of cultivated land from 1980 to 2000 and from 2000 to 2018 both presented the characteristics of transfer-in and transfer-out, indicating a quite different sharpness. Specifically, there was a slow transfer of cultivated land into and out of cultivated land from 1980 to 2000. The transfer-out area of cultivated land was much higher than that of transfer-in from 2000 to 2018, where the transfer-out area of cultivated land was mainly distributed around the urban area. 3) There were some differences in the significance of spatial driving factors at different regional scales. On the whole, the spatial distance, geospatial and policy factors greatly contributed to the change of cultivated land pattern, especially the distance from rural homesteads, the distance from rural roads, and the space of transportation facilities. The natural conditions (elevation, slope) determined the distribution pattern of cultivated land, but indicatied a relatively little influence on the change of cultivated land pattern. Consequently, the random forest can be expected to effectively quantify the spatial driving factors of cultivated land pattern. The finding can provide a strong reference to construct a long-term plan for the protection and sustainable development of cultivated land resources.