Impervious surface coverage in a region is not only an indicator of the degree of urbanization but also a major indicator of environmental quality.Most of the existing methods of extracting impervious surface based on remote sensing concentrate on an urban scale

but the rapid and accurate methods of extracting impervious surfaces in a basin scale are nearly nonexistent in China and abroad.In this study

we used Landsat images acquired in same season covering the entire Hai Basin as data source

and generated a mask for removing the non-impervious surfaces using a land-use data set of roads

and urban

rural

and industrial land.Then

by selecting bright and dark vegetation endmember

high albedo and low albedo impervious surface endmember

and dry and wet soil endmember

we applied a Multiple Endmember Spectral Mixture Analysis（MESMA） model to extract impervious surfaces in the basin scale.The accuracy assessment results showed high accuracy

in that the mean relative error（MRE） and correlation coefficient（R） of all samples were 12.1% and 0.83

respectively

which indicated that the method of extracting impervious surfaces in a basin scale was feasible.