The amount of various impervious land surfaces increases in the process of urban development.Accompanying with the fast urbanization

it has been well known that the drastically increasing impervious land surface has serious impacts not only on urban environment but also on regional and global environment

such as changing rainfall runoff process

causing urban heat islands

changing local microclimate and so on.However

due to the complex components of impervious surface

it is difficult to derive the accurate estimates of impervious cover.Thus

the objective of this study was to directly estimate impervious cover based on multi-spectral features from remote sensing image in city center of Beijing.According to the spectral response of different land cover

a new methodology was explored to directly estimate urban land imperviousness.The object oriented method was applied to classify land cover/use into basic land units within similar spectral features and texture.Then

the multiple principal regression model was explored to estimate the relation of surface imperviousness and TM image based spectral response.The results showed that the combination of multi-spectral features could efficiently predict land imperviousness.Totally

twenty-two spectral indicators were identified to indicate the characteristics of surface imperviousness.Among the spectral indicators

it showed that the four indicators among others

Band 1

Band 5

Band 6 and the Standard Deviation of Band 6

have the closest relation with surface imperviousness.The significant relations of land imperviousness and TM based spectral features could reach 0.851（P

<

0.001）.The model validation showed that the estimated imperviousness based on TM image was accurate（R=0.91）.It proved that the developed method could efficiently estimate land surface imperviousness.In addition

based on the developed impervious model

the distributed pattern of surface imperviousness within Beijing center was extracted.The results showed that the urbanization degree is very high.More than 70% lands of the city center were estimated as high or middle imperviousness

the index of which was between 50%

7

0% or larger than 70%.The average size of these impervious patches was large and the distribution pattern was heterogeneous and fragmented.Moreover

from the core center（within the 2nd ring road） to the urban-rural edge（the 5th ring road） the surface imperviousness patterns were quite different.For example

the 3rd and 4th rings were fast developed in recent decades

containing diverse land use/cover types such as large commercial center

shopping center and residential district.In contrast

more high impervious patches

mainly old buildings

such as old flat residential built-ups and historic sites

filled up the 2nd ring where the development history is thousands of years and new development was strictly limited.The 5th ring was the urban-rural transitional zone

which is the new developmentregion for the city sprawl in recent years.Large industry district

technology district and residential district with high or middle impervious patches occupied around 68.8%.