科研之友
微信
新浪微博
Facebook
分享链接
摘要
Timely and reliable information on crop acreage is essential for formulating grain production policies and ensuring national food security. The combination of available satellite-based remotely sensed images and traditional sampling methods offers the possibility of improved crop acreage estimation at a regional scale. Due to the administrative convenience, reduced survey cost and workload, two-stage sampling has widely been used for crop acreage survey at the large-scale regions. However, compared with single-stage sampling, the two-stage sampling can introduce larger estimation errors, since it has multiple sampling stages. This study's aim is to optimize the two-stage sampling scheme using satellite-based remotely sensed imagery to improve the accuracy of crop acreage estimation. Taking Mengcheng County, Anhui Province, China, as the study area, this study explored the influence of stratum boundary and sample selection method on the sampling efficiency at the first sampling stage, analysed the impact of sample size on population extrapolation accuracy and then optimized the sample size of the second sampling stage using crop thematic map retrieved by ALOS (Advanced Land Observing Satellite) AVNIR (Advanced Visible light and Near Infrared Radiometer)-2 images in 2009. The results showed that the relative error (RE), coefficient of variation (CV), standard error (SE) of population extrapolation, and sampling fraction (f) using the cumulative square root of frequency (CSRF) method is the minimum among three methods for the stratum boundary determination at the first sampling stage, followed by the equal interval (EI) and equal sample size (ESS) method. Moreover, the RE, CV, and SE of population extrapolation using the ST sampling method is the minimum, compared with simple random (SI) and systematic (SY) sampling method. Therefore, the sampling scheme of the first stage can be optimized by CSRF method for stratum boundary determination and stratified sampling (ST) sampling method for samples selection. At the second sampling stage, RE and CV values of population extrapolation decrease as the sample size increases. Comprehensively considering the accuracy, stability of population extrapolation and sampling cost, the most cost-effective sample size for estimating the winter wheat acreage of the study area is 4. From the perspective of the reasonable selection of sample selection methods, sample size and determination of stratum boundaries, this study provides an important basis for formulating a cost-effective two-stage spatial sampling scheme for crop acreage estimation.
- 出版日期2019
- 单位中国农业科学院农业资源与农业区划研究所
