Optimal design of spatial sampling schemes for winter wheat sown area estimation

It is essential for grain production, trade and grain security warning that the information on crop sown area is collected timely and accurately at regional and national level. The experiment on optimal design of spatial sampling schemes for estimating winter wheat sown acreage was conducted combining Remote Sensing, Geographic Information Systems with the traditional sampling methods as well as Geostatistics theory, in order to improve the current crop acreage sampling survey systems. Mengcheng County was selected as study area in Anhui Province, China and winter wheat sown area as the study object. 5 kinds of spatial sampling schemes (simple random sampling, spatial random sampling, classical systematic sampling, spatial systematic sampling and stratified sampling) were designed, respectively. Relative error, coefficient of variation (CV) and sampling costs (represented by sample size) were selected as indices to evaluate the efficiency of spatial sampling schemes in the experiment. The experimental results demonstrate that, the efficiency of stratified sampling scheme is the highest among 5 spatial sampling schemes, which the ratio of winter wheat sown area accounting for the area of a square grid in a sampling fundamental unit (that is a square gird with a size of 5´×5´) was selected as the stratification symbol; Sample size, relative error and CV of population total estimators decrease with numbers of stratum increasing, when stratified sampling scheme is used to extrapolate population values and estimate sampling error; Furthermore, when the number of stratum is 5, sample size, relative error and CV of population total estimators are minimum compared with the rest results of stratified sampling schemes at 5 stratification levels.