Study On Paddy Rice Planting Area Estimation By Spatial Sampling Survey Based On HJ Remote Sensing Images

How to timely and accurately access the crop area has been the focus of the study of the national statistical office. The normal statistical techniques can no longer meet the needs of current agricultural statistical work for efficiency and accuracy due to its inherent limits. The uncertain characteristics of remote sensing information also make it unable to independently fulfill area estimation in the sense of statistics. Which integrate the strengths of both statistics and remote sensing are being widely used. However, the remote sensing technology is not mature enough, and sometimes it is unable to obtain the full coverage of the remote sensing images, which cannot be used to provide the basis for sampling. At this moment, how to take full advantage of the remote sensing images to build a sample survey program is the key of the crop area survey for China Crop Watch System with Remote Sensing.According to a wide range of remote sensing measurements of rice planting area requirements, This paper aims at the mutual limits and affects among the problem of remote sensing data acquisition and the overall accuracy of the sampling method, and takes HJ remote sensing images as auxiliary data for the stratified sampling methods.The spatial sampling plan for rice planting area estimation in the study site was designed, and then applied the more relevant parameters in the design stage and the estimate stage of the spatial sampling. Following are the conclusions drawn from this study:(1) HJ remote sensing images provides a new data source for the crop planting area of remote monitoring. The same resolution and high time resolution can make up for quality with the Landsat-TM data. While the width of its 700km is good for wide range of remote monitoring.(2) The correlations between stratified parameters and survey objects will directly affect the efficiency of stratification. The use of images in the past or the stratified parameters of HJ images with weak correlation other than the most ideal ones will degrade correlations, decrease the efficiency of stratification and need more samples for survey. The results also show that landscape features Edge Density (Li) is better than area directly as a stratified sampling frame.(3) The spatial stratified sampling methods based on the HJ remote sensing images and Field survey plots can effectively improve efficiency and sampling accuracy. Provide timely and accurate real-time forecasting, and decision-making for government departments.