The increase of various impervious land surfaces constitutes one of the main features in urban development, which

results in serious adverse impacts on regional environment. In this study, a new methodology was developed to model urban land

imperviousness based on multi-spectral features by using SPOT image. The results show that the combination of multi-spectral

features can efficiently predict land imperviousness. The significant relations between land imperviousness and SPOT based

spectral features can reach 0.818 (p<0.001). The distribution pattern of urban imperviousness was extracted based on the

developed impervious index and object-oriented classification. The results show that more than 70% lands of the city center are

estimated as being of high or middle imperviousness. The average size of these impervious patches is large with a heterogeneous

and fragmented distribution pattern. The tests on scale impacts show that the accuracy of surface imperviousness derived from the

lower spatial resolution is higher than that from the high spatial resolution image. Accordingly, the impervious surface patterns

are obviously different.