基于MODIS和ENVISAT数据的湖北省四湖地区土地覆盖分类

Land cover classification of Four Lakes Region in Hubei Province based on MODIS and ENVISAT data

依据ENVISAT ASAR数据中后向散射系数的差异对湖北四湖地区的城镇、水域、植被覆盖区进行划分;根据区域种植制度及物候特点,以4月下旬至5月上旬MODIS-NDVI值区别植被覆盖区中的作物植被与非作物植被;并采用基于高分辨率ETM数据的土地分类结果对基于上述规则分类的结果进行验证.结果表明:借助DEM高程数据,可将研究区非作物植被划分为林地与滩地;利用月份NDVI平均值的差异,可将作物划分为中稻、棉花和晚稻,并获得水田和旱地的区分;采用低空间分辨率的MODIS数据获得的土地覆盖分类结果,与采用高分辨率ETM数据分类的结果具有一定程度的相似性,以ETM数据分类结果为标准的总误差率为13.15%;利用上述判别流程进行大尺度土地覆盖分类与制图可以实现对区域土地覆盖变化的快速跟踪.

Based on the differences of back scattering coefficient in ENVISAT ASAR data, a classification was made on the towns, waters, and vegetation-covered areas in the Four Lakes Region of Hubei Province. According to the local cropping systems and phenological characteristics in the region, and by using the discrepancies of the MODIS-NDVI index from late April to early May, the vegetation-covered areas were classified into croplands and non-croplands. The classification results based on the above-mentioned procedure was verified by the classification results based on the ETM data with high spatial resolution. Based on the DEM data, the non-croplands were categorized into forest land and bottomland; and based on the discrepancies of mean NDVI index per month, the crops were identified as mid rice, late rice, and cotton, and the croplands were identified as paddy field and upland field. The land cover classification based on the MODIS data with low spatial resolution was basically consistent with that based on the ETM data with high spatial resolution, and the total error rate was about 13.15% when the classification results based on ETM data were taken as the standard. The utilization of the above-mentioned procedures for large scale land cover classification and mapping could make the fast tracking of regional land cover classification.