基于改进空间模拟退火算法的耕地质量监测点优化布设  

OPTIMAL LAYOUT OF CULTIVATED LAND QUALITY MONITORING POINTS BASED ON IMPROVED SPATIAL SIMULATED ANNEALING ALGORITHM

作　　者：杨文浩 刘振华[1,3] 杨颢 喻红刚 胡月明 Yang Wenhao;Liu Zhenhua;Yang Hao;Yu Honggang;Hu Yueming(College of Natural Resources and Environment,South China Agricultural University,Guangzhou 510642,Guangdong,China;Hunan Branch of Guangdong Youyuan Land Information Engineering Co.,Ltd,Changsha 410004,Hunan,China;Guangdong Land Information Engineering Technology Research Center,Guangzhou 510642,Guangdong,China;Hainan University,Haikou 570228,Hainan,China;South China Academy of Natural Resources Science and Technology,Guangzhou 510610,Guangdong,China)

机构地区：[1]华南农业大学资源环境学院,广东广州510642 [2]广东友元国土信息工程有限公司湖南分公司,长沙410004 [3]广东省土地信息工程技术研究中心,广州510642 [4]海南大学,海口570228 [5]广州市华南自然资源科学技术研究院,广东广州510610

出　　处：《中国农业资源与区划》2023年第2期110-118,共9页Chinese Journal of Agricultural Resources and Regional Planning

基　　金：国家自然科学基金“赤红壤区耕地质量演变机理与提升机制”(U1901601)。

摘　　要：[目的]科学合理地设置耕地质量监测点是加强我国耕地保护、提升耕地质量的重要依据,探讨监测点设立的准确性与提高耕地质量监测效率及成本具有重要的现实意义。[方法]文章以广州市从化区为例,综合考虑监测点的代表性、调查成本、道路可达性、监测点的适宜性等因素,以Kriging预测均方根误差、坡度、道路可达性为指标建立县域选取耕地质量监测样点指标体系,并利用改进的空间模拟退火算法对监测点进行优化,并与改进前的和不同尺度的网格法进行了对比分析。[结果]基于空间分层抽样模型最终确定了74个监测点。在满足一定精度要求下,与网格法相比,改进的空间模拟退火算法能大幅度缩小监测点数。在同等监测点数下,改进后的空间模拟退火算法布设的监测点在耕地质量预测方面的精度远高于网格法,略低于改进前。与改进前和网格法相比,改进后的监测点多设立于靠近道路和地形平坦的地方,能有效地避免设立在深山野林及高坡度的地方。[结论]在不损失过多精度的前提下,该方法不仅能有效满足预测县域耕地质量的变化情况的需求,同时提高了县域耕地质量监测效率和节约了监测成本。The scientific and reasonable setting of cultivated land quality monitoring points is an important basis for strengthening the protection and improving the quality of cultivated land in China,and it is of great practical significance to explore the accuracy of monitoring point establishment and improve the efficiency and cost of cultivated land quality monitoring.Taking Conghua district of Guangzhou city as an example,the article considered the representativeness of monitoring points,survey cost,road accessibility and suitability of monitoring points,established an index system for selecting cultivated land quality monitoring sample points in the county with Kriging prediction root mean square error,slope and road accessibility as indicators,and optimized the monitoring points with the improved spatial simulated annealing algorithm,and compared it with the pre-improved and different scales of grid method.The results were listed as follows.Based on the spatial stratified sampling model,74 monitoring points were finally determined.The improved spatial simulated annealing algorithm could significantly reduce the number of monitoring points compared with the grid method under certain accuracy requirements.Under the same number of monitoring points,the accuracy of monitoring points deployed by the improved spatial simulated annealing algorithm was much higher than that of the grid method and slightly lower than that of the pre-improvement method in predicting the quality of cultivated land.Compared with the pre-improvement and grid methods,the improved monitoring points were more often set up near roads and flat terrain,which could effectively avoid setting up in deep forests and high slope areas.In summary,without losing too much accuracy,the method not only effectively meets the demand of predicting the changes of the county cultivated land quality,but also improves the efficiency and saves the cost of monitoring the county cultivated land quality.

关 键 词：耕地质量 监测 样点布设 优化 模拟退火算法 

分 类 号：F301.2[经济管理—产业经济]