Mehhods For Rice Identification And Sown Area Measure Using ASAR Data--A Case Study In Baoying County Of Jiang Su

Governments of different levels have always paid great attention to the techniques to accurately investigating rice yield and sown area. Owing to its objectivity, timelines and accurateness of the technology of remote sensing, it had been applied to rice inspection for a long time. However, rice grows in the south mostly abounded with rainfall, resulting in application limitation of optical remote sensing for rice inspection, since it is very difficult to get a good optical image (such as NOAA, SPOT and LANDSAT) during the whole rice growing period. Therefore, the active ware of microwave communication-SAR (Synthetic Aperture Radar), which is timely, independent of illumination and high rate repeating coverage, has been considered as one of the most important information sources of agricultural inspection in these areas.This paper applied the multi-temporal and multi-polarization Envisat-ASAR data in the rice growing period to study the methods of basic pretreatment on image and the backscatter coefficient of typical nature object, especially the rice, using VV polarization and VH polarization. Based on the above methods, we chose the color image to distinct rice plants and its sown area. We also tested the identification precision with GIS and GPS technology. Our results are as follows:1, Integrating the ERDAS, CORELDRAW and other software, we rectified the ASAR images according to topography pictures. At the same time, we developed a method to accurately integrating the optical and ASAR images.2, Basing on the analysis of the spot noise on the ASAR image, After comparing the filter effects between several general noise wave filters, strain have made a kind of relative school after wave effect flexible vague is subordinate to degree since suit to strain wave algorithm, have reached the needs of actual application.3, The backscatter coefficient in appearance of VV polarization and VH polarization has drawn samples to compare, such as rice, water, dry crop and resident. And it has determined three ways of fusing color imagines to rice distinction and area measurement, including the VH polarization of August 31 and of June 22, the VV polarization of August 31 polarization and the VH polarization of June 22.4. Having compared the precision rate between the supervised classification and nerve network classification, the author found that the supervised classification method was better with a higher precision rate.5. The author selected 8 pieces of sample lands to test the distinction precision arte using DGPS measurement. Without any after-treatment, the precision rates of classification and distinction were 90.13% and 95.16%, respectively. After deleting the small groups, these rates changed. The inspection precision rate of sown area was 92.62% through comparing the inspected area with the survey data reported by the local governments.