| Agriculture is the foundation of our national economy.Accurate and timely investigation of crop acreage are the premise for the state to formulate grain policy and ensure food security.With the continuous improvement of application system of crop acreage by remote sensing,remote sensing,remote sensing technology has been widely applied in the survey of crop acreage.At present,most researches on crop acreage based on remote sensing were mainly concentrated in the northern plains of China.However,there are not many related researches in the hilly area of Fujian.At the same time,the related researches mainly used medium and low resolution remote sensing data such as MODIS and Landsat,while the application of new domestic satellite data such as GF-1 and GF-2 were relatively few.This paper selected Minhou County of Fujian Province as the study area and used GF-1 remote sensing data as the main data source.Remote sensing technology,sampling technology,simulated annealing algorithm and spatial autocorrelation theory were used to sample the crop acreage in the study area.The spatial sampling plan which was suitable for the crop acreage in the study area was determined in order to provide a scientific reference for the spatial sampling survey of the crop acreage in Fujian hilly areas.The main conclusions were as follows:(1)Object-oriented classification methods were used to extract crop acreage information from GF-1 remote sensing data.The overall classification accuracy was as high as 92.70%,and the Kappa coefficient was 0.9180.It indicated that the image object information extraction based on the object-oriented classification method can effectively obtain remote sensing identification data of crop acreage in the study area.(2)Taking the square grid as the basic shape of the sampling unit,8 kinds of remote sensing sampling frames of 500m×500m,800m×800m,1000m×1000m,1200m× 1200m,1500m× 1500m,2000m×2000m,2500m×2500m,3000m×3000m were constructed.And 1500m×1500m was selected as the optimal sampling unit size by comparing the spatial autocorrelation indices and variation coefficient of crop acreage of 8 kinds of remote sensing sampling frames.(3)Traditional sampling methods and spatial sampling methods were used to sample crop acreage in the study area.Compared with other sampling methods,the sampling efficiency of the spatial stratified sampling method was the highest because its relative error was 3.86%,the coefficient of variation was 6.03%,and the sampling cost was 6.03.(4)Combined with the vector map of the traffic network in the study area,the estimation accuracy,road accessibility,and sample survey costs are taken into consideration.Taking the shortest field survey path as the objective function,the simulated annealing algorithm was used to plan the optimal routes for field surveys at each sampling points.This will save survey time and reducing survey costs when conducting field sampling surveys.(5)On the basis of spatial stratified sampling,the revised crop acreage was estimated and the crop acreage in the study area was 174,142 hectares.Comparing the revised crop acreage with the statistical results of the statistics of Fuzhou Bureau of Statistics in 2015,the accuracy of its estimation reached 94.78%.It meant that the spatially stratified sampling method had higher accuracy in the sampling estimation of crop acreage in the study area. |
PDF Full Text Request