三种土地覆盖遥感数据在中国区域的精度分析

土地覆盖遥感数据的精度分析是数据应用的前提和基础。该文以中国科学院资源环境科学数据中心提供的2000年中国土地覆盖数据作为参考,基于对土地覆盖类型的归并,在各种土地覆盖类型的分布面积、空间位置等方面分析了GLC2000、MODIS V004和MODIS V005三种数据描述中国土地覆盖状况的能力,并引入差异性指数,深入分析影响土地覆盖数据精度的原因,得到以下主要结论:总体上各数据均能反映中国主要土地覆盖类型的分布情况,GLC2000、MODIS V004和MODIS V005数据的总体精度分别为62.8%、61.7%和61.9%;3种数据的乔木林地面积较参考数据均偏大,而建筑用地、湿地和水体的面积均偏小;乔木林地、灌木草地、耕地和裸地冰川的精度较高,建筑用地和湿地的精度较低;GLC2000数据在省级水平上对乔木林地和耕地的描述最准确,MODIS V005数据的平均用户精度和生产者精度最高;土地覆盖数据的精度随着数据差异性指数的增加而降低,随着差异性指数为1的区域面积比例的增加而升高;MODISV005数据的精度受土地覆盖差异性的影响相对较小。

Precision analysis of three land-cover types in China region

Accuracy evaluation is the precondition and foundation for the application of land cover data. In order to study the ability of describing the land cover status in China, GLC2000, MODIS V004 and MODIS V005 land cover data on distribution area and spatial distribution were compared with the reference data in 2000 from RESDC (Data Center for Resources and Environmental Sciences, Chinese Academy of Science) based on land cover types aggregating. Land cover diversity index was introduced in this study to analyze the impact of land cover data accuracy. The results showed that: all the three data can describe the spatial distribution of land cover well in China; the precision of GL2000, MODIS V004 and MODIS V005 were 62.8%, 61.7% and 61.9% respectively; the proportion of forest/woods for the three land cover data were larger than that of reference data, the areas of building, wetlands and water bodies were smaller; the accuracy of forest/woods, grass/shrub lands, croplands and barren/ice were higher than that of urban and wetlands; the precise of forest/woods and croplands of GLC2000 were the most in provincial level, MODIS V005 has the highest user and producer accuracy on average; the precision of the three data were decreased with the increase of land cover diversity index, and it was increased with the increase of the proportion of land cover diversity index that equals to 1, land cover heterogeneity has the least impact on MODIS V005 data.