多时相MODIS影像水田信息提取研究

Study on Extraction of Paddy Rice Fields from Multitemporal MODIS Images

水稻种植及其分布信息是土地覆被变化、作物估产、甲烷排放、粮食安全和水资源管理分析的重要数据源。基于遥感的水田利用监测中,通常采用时序NDVI植被指数法和影像分类法分别进行AVHRR和TM影像的水田信息获取。针对8天合成MODIS陆地表面反射比数据的特点和水稻生长特征,选取水稻种植前的休耕期、秧苗移植期、秧苗生长期和成熟期等多时相MODIS地表反射率影像数据,通过归一化植被指数、增强植被指数及利用对土壤湿度和植被水分含量较敏感的短波红外波段计算得到的陆表水指数进行水田信息获取。将提取结果与基于ETM+影像的国土资源调查水田数据,通过网格化计算处理并进行对比分析,结果表明,利用MODIS影像的8天合成地表反射率数据,进行区域甚至全国的水田利用监测是可行的。

Information of paddy rice fields and their spatial distribution is the data source for land use and land cover changes, crop yield estimates, methane emission estimates, food security, and water resources management. A number of studies have explored the potential of remote sensing images to identify Paddy rice fields. Those studies that identified paddy rice fields used image classification procedures for TM images and temporal development of the NDVI for AVHRR images. 8-day composite MODIS （ Moderate Resolution Imaging Spectroradiometer） Surface Reflection products （MOD09A1） provide the potential for the improved characterization of vegetation at large spatial scale. According to the different paddy rice growth stages, multitemporal MODIS images were selected by the flooding and transplanting period, growing period, maturation period, and the fallow period after harvest. Three vegetation indices, Normalized Difference Vegetation Index （NDVI）, Enhanced Vegetation Index （EVI）, and Land Surface Water Index （LSWI） using the shortwave infrared band sensitive to soil moisture and leaf moisture content, were calculated to extract information of paddy rice fields. We compared the paddy rice fields dataset to the dataset extracted from ETM ＋ images and produced at 40-km gridded data, the results show that the paddy rice extracting algorithm of time-series MODIS data could be applied to monitor paddy rice fields at large spatial scales.