Extraction of impervious surface based on multi-source satellite data of Qinhuai River basin from 1979–2009

Impervious surface (here after IMP) is a typical characteristic of urban area and is one of the most important environmental indicators. A 30 year time series (1979-2009) of Landsat imagery and CBERS imagery for Qinhuai River basin was analyzed to estimate the IMP. A new approach was proposed to quantify impervious surface as a continuous variable by using multi-temporal and multi-source datasets. The principal component analysis (PCA) approach was performed in CBERS imagery to increase information of image. Linear Spectral Mixture Analysis (LSMA) was used to determine the fractional composition of vegetation, high- and low-albedo and soil for each pixel of the normalized data. Supervised classification technique and MNDWI (Modified Normalized Difference Water Index) method were used to extract water. IMP was then estimated by adding all of high-albedo and part of low-albedo fraction images. Temporal rule, that minimized classification error, was developed based on each pixel´s classified trajectory over the time series of imagery. Overall cross-date classification accuracies for impervious vs. non-impervious surface were greater than 85%. The results indicated that the area of impervious surface in the Qinhuai River basin increased by 963% over 30 years, and impervious surface rate was from 1.70% in 1979 to 18.02% in 2009. The increase rate of IMP was 7.1% before 2003 and 12.9% after 2003. This approach demonstrated that impervious surface distribution could be derived from multi-temporal and multi-source satellite datasets with promising accuracy.