基于信息熵分层抽样的冬小麦区域种植面积估算

摘要

Based on the concept of information entropy and the posterior probability of maximum likelihood classification,indicators of information entropy and information content were designed as auxiliary variables for remote sensing sampling of wheat. Through a comparative analysis between such indicators and the traditional indicator of area scale,this research verified effectiveness of the methodology proposed. The results showed that compared with the traditional indicator of area scale,the indicators of information entropy and information content could effectively increase accuracy of extrapolation estimation via sampling in different-sized sampling units.In Tangyin county with single planting structure and regular plot,the two indicators mentioned above,in the sampling unit between(80 m ×80 m) and(400 m ×400 m),could increase the relative accuracy of extrapolated results to varying degrees; while in Zhongmu county with complex planting structure and relatively fragmentized plot,the two indicators mentioned above,in the sampling unit smaller than 192 m×192 m,could effectively increase the estimation accuracy.%基于信息熵概念和最大似然分类后验概率设计了信息熵和信息量指标作为冬小麦遥感抽样的分层标志,与传统面积规模指标进行对比分析,验证了该方法的有效性.试验结果表明,信息熵和信息量指标相较传统面积规模指标,在不同尺寸抽样单元下,能够有效提高抽样反推的估算精度.在种植结构单一、地块规整的汤阴县,上述2个指标在(80 m×80 m) ~(400 m×400 m)抽样单元下,均能不同程度地提高反推结果的相对精度;在种植结构复杂、地块较破碎的中牟,上述2个指标能够在192 m×192 m以下抽样单元下有效提高估算精度.