基于遥感和“两区”划定耕地数据的农作物面积估算分层抽样体系研究

Research on the stratified sampling system for crop area estimation based on remote sensing technology and cultivated land data delineation in the“Two Zones”

关键词:  样方  分层抽样  抽样方法  遥感  “两区”耕地

Key words:  quadrat  stratified sampling  sampling method  remote sensing  cultivated land in“two zones”

摘要:

【目的】 四川气候多云雨天气，通过全覆盖遥感影像解译来掌握主要农作物种植面积、分布非常困难，采用野外实地抽样调查与遥感影像分类相结合是有效可行的方法。【方法】 文章以遥感和GIS为手段，充分利用“两区”划定耕地基础数据，采用两级空间分层抽样方法和实验最优空间抽样体系的设计来建立长期的野外调查样方。采用空间分析技术、传统抽样方法相结合的两级空间分层抽样技术，在不同研究区内，以误差、变异系数和抽样个数为抽样效率的评价指标，实验和确定不同层的样方大小、抽样方法和抽样比例的最优抽样体系；通过确定的最优抽样体系，反推2022年研究区油菜和小麦面积，验证其精度。【结果】 抽样体系中样方大小为0.5 km×0.5 km、抽样方法为分层顺序抽样方法或分层麦士比抽样方法、抽样个数为25个左右，误差和变异系数较小，为最优抽样体系。【结论】 经验证，用最优抽样体系反推的2022年油菜小麦面积精度均在93%以上，建立了适用于四川省不同区域的最优抽样体系，布设了长久的野外调查样方。

Abstract:

【Purpose】 Sichuan has a cloudy and rainy climate，and it is very difficult to understand the planting area and distribution of major crops through full coverage remote sensing image interpretation. The combination of field sampling survey and remote sensing image classification is an effective and feasible method.【Method】 By means of remote sensing and GIS，this paper used the basic data of cultivated land demarcated in "two zones"，a two-level spatial stratified sampling method and the design of an experimental optimal spatial sampling system to establish long-term field survey samples.We adopted the two-level spatial stratified sampling technology combined with spatial analysis technology and traditional sampling methods.We used error，coefficient of variation，and number of samples as the evaluation indicators of sampling efficiency in different research areas，experimented and determined the optimal sampling system for sample size，sampling method，and sampling proportion at different levels. Through the determined optimal sampling system，the area of rape and wheat in the study area in 2022 was inversely estimated to verify its accuracy.【Result】 In the sampling system，the combination of a sample size of 0.5 km×0.5 km，a stratified sequential sampling method or a stratified max ratio sampling method，and a sampling number of about 25 samples had smaller errors and coefficient of variation，making it the optimal sampling system.【Conclusion】 It is verified that the precision of the area of rape and wheat in 2022 by using the optimal combination scheme is more than 93%，the optimal sampling system applicable to different regions of Sichuan Province has been established，and the long-term field survey sample has been set up.