Rice Mornitoring And Mapping With ASAR Data

Rice monitoring and yield estimate is of importance in China, where rice is one of staple food crops. It is difficult to obtain optical images in the right time and place, for most of rice cultivaliton lies in rainy and cloudy regions. Besides, traditional methods on the basis of local statistics for rice monitoring are always inaccurate. Spaceborne SAR (Synthetic Aperture Radar) instruments, having all weather capability, present a good alternative to those traditional methods for rice monitoring and yield estimate.A major objective of this study is to assess the use of ASAR data for rice field mapping and rice growth monitoring. The approach includes preprocessing of ASAR data, the analysis of the temporal variation of the radar backscatter of 4 classes of rice fields, the analysis of correlation between the radar backscatter and rice parameters (wet biomass, LAI, NDVI), and development of methodologies for rice field mapping. At last the identification accuracy was tested by GIS and GPS technology. Results are as follows:1 , Speckle suppression of ASAR data. In order to reduce the speckle noise that degrades SAR data, a pre-processing chain was designed. It comprises 4 steps: (a) calibration, (b) extraction of test area, (c) co-registration, (d) multi-channel filtering. Steps (a), (b), (c) were implemented by some software, and a filter programmed in C was used for step (d). This multi-channel filter reduces speckle without damaging thin structures, by linearly combining M input images on a pixel-to-pixel basis, thus creating M output images with reduced speckle. It has been confirmed that the multi-channel filter has better performance than other optical filters.2 , Backscatter signature of rice fields. Temporal variation of backscattering coefficients of 4 categories of rice fields were derived from ASAR images of the test region in 2004. Ground data were collected to calculate wet biomass and LAI of rice,NOAA images were utilized to derive NDVI, and the correlations between the backscattering coefficient and those rice parameters were analyzed. Based on the backscatter behaviors of rice, a method for rice fields mapping using image ratio techniques has been developed.3> Classification method of rice fields. First, both methods based on image ratio techniques and supervised classification were applied to separate rice fields from other land with ASAR/APP data, including using one date image and multi-data images, and different types of rice were separated with ratio techniques;Second, a regional mapping with Wide Swath ASAR data has been finished. Since WS data have only one single polarization, HH or W, the mapping approach consists in using the backscatter temporal change;Third, the area changes of rice fields between 2004 and 2005 was detected using the classification method of ratio techniques.4> Validation of rice fields classification. In order to obtain reference maps for assessing accuracy of mapping methods, Differential Global Positioning System (DGPS) has been used for accurate in situ mapping of several samples of about 1 km2, comprising rice fields and other land use classes. According to comparison, the best threshold was confirmed, so as to the fact that the ratio techniques has a better accuracy than supervised method when doing rice mapping.