介绍了光学和微波遥感影像相结合进行泰国水稻种植面积监测方法。泰国雨季雨量充足 ,气温适合 ,同一时间耕地上水稻的物候多样 ,每月水稻种植面积都发生变化。利用旱季的TM影像 ,获得耕地信息。同时利用TM影像覆盖的雷达区域进行非耕地去除 ,进行非监督分类 ,提取反映水稻种植不同生长期的雷达影像后向反射系数特征 ,建立各种类型的分类模型 ,采用监督分类的方法对全景雷达数据进行水稻种植情况调查 ,并分别予以识别和统计 ,反映研究区水稻月种植情况。分类结果通过类别检验和面积量算检验进行精度评价和分析。

Thailand is the largest exporter of rice in the word grain market with a reputation of high grain quality.One-half of the rice land was located in the Northeast region.Planted area changes monthly in tropical agriculture,unlike that of agriculture in temperate zones.This paper estimated the area planted with rice using remote sensing data in center-northeast of Thailand.We propose a method which can be used to estimated the rice planted area in tropical regions by RADAR data.The arable land area was measured using Landsat Thematic Mapper (TM) data acquired in the dry season and identified minthly rice planted fields using Synthertic Aperture Radar(SAR) data acquired in the rainy seasin(planting season).Landsat TM data (path-row:128-49)acquired on 7April 2001 was used to identify the agricultural area.To detect the planted fields,RADARSAT SCN(ScanSAR Nareow)data were employed.The parameters of SHR are as follows:C-band,HH(horizontal transmit and horizontal receice)polarization,beam mode;W2,s5,s6,orbit:35606 descending,scene center:15°54′N and 102°41′E.Four images were acquired on 0205-2002,9-06-2002,01-08-2002 and 31-08-2002.Arable land area was labeled using unsupervised classification of the TM data.According to statistic resuly about time series of backscatter coefficients of rice,four type of models were built to estimate the monthly change in plante area.Since rice-planting is not carried out simultanrsously and the planted area change monthly,some assumptions were necessary for estimating the planted area.Assuming that the intensity of planting exhibits a normal distribution,there are five peaks due to the monthly planting.The monthly rice-planted area was wstimated based on suoercised classification using the defined model during the rice-planting season.The overall classification accuracy was 91%,and the rice information including arable land is 90%.