| Crop acreage information is an important basis for the country to formulate food policies and economic plans.Accurate and timely access to crop acreage information is of great significance for scientifically adjusting the scale structure of crop production,strengthening crop production management,ensuring effective supply of agricultural products and national food security.With the development of Earth observation technology,the spatial sampling method combined with satellite remote sensing and traditional sampling theory has fully utilized the respective advantages of remote sensing and sampling statistics,and has been widely used in many countries including the United States,the European Union and China.In the large-area crop area statistical survey business,the accuracy and timeliness of crop acreage information acquisition have been effectively improved.However,in the past,the spatial sampling research and practice of crop area all assumed that the sampling unit satisfies the independent unrelated principle of classical statistical requirements,it only uses the traditional sampling method for sample selection and overall extrapolation,and does not consider the spatial correlation and variability of crop cultivation due to natural conditions,socio-economic factors,etc,resulting in significant deficiencies in the rationality of sampling plan design and the accuracy of overall inference.For above problems,this paper selects Fengtai County of Anhui Province,Dehui City of Jilin Province,and Shuyang County of Henan Province as the study areas.Using GF-1 remote sensing image data to extract three kinds of crops,using spatial analysis technology,geostatistics theory and traditional sampling methods to quantitatively evaluate the spatial autocorrelation and spatial variability of various crops,and analyze the characteristics and laws of spatial autocorrelation and spatial variability with the change of sampling unit scale.Design a variety of sampling schemes,select the sample size(n),the sample extrapolation total relative error(Re)and the coefficient of variation(CV)of the total population estimator as the evaluation index of sampling efficiency,analyze the effects of spatial autocorrelation and spatial variability of crops on sampling efficiency,compare the sampling efficiency of traditional sampling with spatial sampling considering spatial autocorrelation and variation,and the optimization of spatial autocorrelation and spatial variability crop area sampling schemes is realized from three aspects: sampling unit scale,sample size and sample layout.The main conclusions are as follows.(1)The spatial autocorrelation of winter wheat area in Fengtai County was evaluated by the global spatial autocorrelation index Moran's I.It was found that the spatial autocorrelation of winter wheat area in the sampling unit decreased with the increase of sampling unit scale.The extrapolation estimation results of the three spatial sampling schemes indicate that the effective sampling unit size of the spatial autocorrelation of crop area affecting the sampling efficiency ranges from 500 m to 2000 m.Within the scale of this sampling unit,the spatial sampling scheme based on 5% sampling ratio is the optimal sampling scheme suitable for spatial autocorrelation crops.The sample space layout results show that the spatial autocorrelation can improve the sampling accuracy by the way of sample unit layout without considering the hierarchical mark.(2)The global spatial autocorrelation index Moran's I,the local spatial autocorrelation index Moran's I and the local autocorrelation LISA map are selected as the evaluation indicators of the spatial autocorrelation of crop area.The local LISA aggregation map of crop area indicates that the local autocorrelation of winter wheat area in the sampling unit shows the transitional characteristics of multitype aggregation to single type aggregation.When the sampling unit scale is 3500 m×3500 m,the crop area within the sampling unit has significant global spatial autocorrelation,and the local spatial autocorrelation is the strongest.The traditional sampling method and spatial sampling method are used to extrapolate the area of winter wheat under the scale of 3500 m×3500 m sampling unit and compare the sampling efficiency of various sampling methods,it is found that under the condition of 10% expected error,the stratified sampling with the ratio of winter wheat area in the sampling unit as the stratified marker has the highest sampling efficiency.(3)When the variogram is used as the evaluation index of crop spatial variability,the abutment value tends to decrease gradually with the increase of the sampling unit scale.The nugget coefficient has a maximum value at 500 m and 2500 m,and the crop area in the sampling unit is The spatial autocorrelation at 2500 m is the smallest and the spatial variability is the largest.When the variogram is used as the evaluation index of crop spatial variability,the abutment value tends to decrease gradually with the increase of the sampling unit scale,and the nugget coefficient has a maximum value at 500 m and 2500 m.When the sampling unit scale is 2500 m,the spatial autocorrelation of the crop area in the sampling unit is the smallest and the spatial variability is the largest.The extrapolation estimates of spatially variable crop areas by the three sampling schemes indicate that the effective scale of the spatial variability of crop area in the sampling unit to the sampling efficiency ranges from 500 m to 3500 m,sampling schemes with 5% sampling ratios within the scale of this sampling unit are the most efficient spatially variable crop sampling schemes.(4)According to the spatial variability of crop area in the sample unit,three kinds of spatial sampling methods were used to evaluate and compare the estimation accuracy of crop area extrapolation by Kriging method.The results of cross-validation and independent data set verification showed that under the condition of 10% expected error,For the optimal estimation of the crop area with relatively strong total variability in the system by spatial local interpolation(general Kriging interpolation),the system space space system sampling plan is a sample point selection scheme that makes the error within the region small and the precision is higher. |
PDF Full Text Request