福建省闽侯县区域农作物种植面积的空间抽样方案

摘要

基于国产GF-1号卫星影像农作物空间分布数据,以正方形网格作为抽样单元形状,综合应用遥感技术、空间抽样方法、模拟退火算法及空间自相关理论,对福建省闽侯县农作物种植面积进行估算.结果表明:根据全局自相关指数Moran′s I及其显著性指标Z Score,选取1500 m×1500 m作为最优单元尺寸;对研究区进行剖分,构建抽样框,对比空间随机抽样、空间系统抽样及空间分层抽样的抽样结果,空间分层抽样方法的样本容量为37,相对误差为3.86%,抽样效率最高;运用模拟退火优化算法规划野外调查的最优路径,确定空间抽样调查方案.%Based on the spatial distribution data of crops of domestic GF-1 satellite image,the square grid was taken as the sampling unit shape,and the remote sensing technology,spatial sampling methods,simulated annealing algorithm and spatial autocorrelation theory were applied to this study for estimating the planting area of crops in Minhou County, Fujian Province. The results showed that according to global autocorrelation index Moran′s I and its significant index Z Score,1 500 m×1 500 m was selected as the opti-mal size unit. And the study area was divided to construct the sampling frame. Compared with the sampling results of spatial random sampling,spatial system sampling and spatial stratified sampling, the sampling efficiency of the spatial stratified sampling method was the highest because its number of samples was 37,the relative error was 3.86%. The space sampling survey scheme was deter-mined because the optimal path for the field survey was found by simulated annealing algorithm.