Studies On Extraction Of Impervious Surfaces Using Landsat-5 TM Data

With the development of social economy,the expansion of urban is speeding up.The knowledge of impervious surface can reflect the degree of urbanization directly,and it is also very important to urban planning and resources management.Moreover,the size,distribution and location of impervious surfaces are significant to a series of issues on the global climate change and the inter-action between human beings and environment.To this end,accurately extracting the information of impervious surfaces in urban areas is essential to urban planning,ecological assessment and regional sustainable development.In recent years,due to the rapid development of remote sensing technology,rich and multi-source remote sensed images in urban areas can be captured.Therefore,remote sensed images have been important data resources in the study of extraction of impervious surfaces.Remote sensed imagery can be grouped into three categories based on spatial resolution,including coarse-resolution images,medium-resolution images and high-resolution images.Especially,Landsat series images have been wide-used in the extraction of impervious surfaces in urban areas,due to its advantages of rich spectral information,wide cover range and cost free.Thus,in this paper,Landsat 5 TM data is used for extracting the information of impervious surfaces in urban areas.The main work of this paper is as follows:1)By analyzing the traditional spectral unmixing methods,the features of minimum noise fraction(MNF)feature space based on the initial spectral bands in Landsat 5 satellite is used to select endmember models.Then,the all-limited liner spectral mixture analysis(LSMA)method is used for obtaining the abundance of impervious surfaces.2)Based on the idea of feature extraction after analyzing the typical land covers in study areas,spectral mixture analysis(SMA)is conducted on feature bands instead of original spectral bands.This novel algorithm combines Biophysical Composition Index(BCI),Normalized Difference Water Index(NDWI)and Soil-Adjusted Vegetation Index(SAVI)to construct endmember models.Then,the liner spectral mixture analysis(LSMA)algorithm was carried out to extract the fraction image of impervious surface.3)Analyze the unmixing performance of these two methods.Results showed that the novel algorithm developed in this paper is superior to the traditional unmixing algorithm which based on the original spectral space.This novel algorithm can suppress the spectral-similar objects,such as bared soil,and it can also improve the accuracy of unmixing in the developed area where some building shadows existing.