【摘要】 该文以包含平坝地及丘陵区的德阳地区为研究区,利用该区水稻拔节(7月10日)和乳熟(8月25日)两景双极化(HH/HV)ALOS_PALSAR雷达数据,参照2009年6月3日TM(129/038,130/038)光学影像,分析主要地物影像特征,对比分析水稻在不同时相、不同极化方式后向反射系数,结合GIS和GPS技术获取的地面调查资料,对水稻种植区域进行识别和面积监测,最后利用水稻本底调查资料进行对比,探讨利用ALOS_PALSAR数据进行水稻遥感监测的可行性和准确性.该文采用了监督分类和决策树分类两种方法,其中决策树分类以后向反射系数HH/HV比值和两时相HV-HV差值为基础,以HH-HV≥1.5和0.2≤HH/HV≤0.6(7月10日)和0.15≤HH/HV≤0.6(8月25日)为水稻阈值区域.结果表明:与2009年水稻本底解译随机选取的100个水稻图斑相比较,监督分类准确率为91%,区域水稻种植总解译面积比本底增加了8.2%;决策树分类准确率为93%,区域水稻种植总解译面积比本底增加了3.48%.结合地面样方调查资料,利用ALOS_PALSAR多时相影像进行水稻识别和面积测量可以达到较高精度.更多还原

【Abstract】 Rice is an important crop in China.Timely acquisition of the data of rice planting area changes is an urgent need for food policy and food security of China,but monitoring of rice crops using optical remote sensing is particularly difficult due to the cloud cover since the rice growing seasons often coincide with the rainy seasons.In this paper,with Deyang county in Sichuan province in southwest of China as the study area,based on the multi-temporal and dual polarization ALOS-PALSAR data(July 10th and August 25th) in 2010,L band and HH/HV channel,after examination the dependences of the multi temporal backscatter coefficient on land cover types,the authors applied supervised and decision tree classification methods for rice planting recognition in plain and hilly area.The rice map was validated by rice background inventory data and ETM image.The results showed that multi-temporal PALSAR dual polarization data could be used to monitor rice planting maps for plain and hilly land using the proposed method.The user accuracy was higher than 90%.In conclusion,it is possible to discriminate rice based on multi-temporal dual polarization ALOS PALSAR image feature in this test site.更多还原