基于ETM＋制作土地利用覆盖图——以制作北京1：5万土地利用覆盖图为例

简要介绍了基于LANDSAT7 ETM＋影像,采用计算机非监督分类、监督分类与人工解译相结合的方法制作土地利用覆盖图的过程和所采用的关键技术,给出了适用于规模化生产土地利用覆盖数据的工艺流程图。使用该方法制作的十一种分类要素的北京地区1：5万土地利用覆盖图,平均分类精度为84．85％,可以满足一般用户对土地利用覆盖图的要求。

The Process of Generating the Landuse/Landcover Map Based on ETM+

The paper describes the process on using the method combined the unsupervised classification, supervised classification and manual interpretation technology based on the LANDSAT7 ETM+ to make the landuse and landcover map, and also works out the work flow to produce the landuse/landcover data. By this way, the Beijing 1: 50000 landuse and landcover map which has been mapped in the 10classes including: deciduous forest, evergreen forest, shrub, grass land, orchard, irrigated land, dryland, urban/built up, village and water, has been generated. The result of image classification accuracy assessment comparing the "map" generated by this way against the referenCe data from the airphoto(scale l:20,000), shows the classification accuracy of the map can be expected to 84. 85 percent.