Abstract:

Quantitative research of the urban rainfall-runoff process is of great importance for exploiting, utilizing, planning, and managing the urban rainwater resources. In this study, an improved linear spectral mixture analysis (LSMA) method is developed by integrating the normalized difference built-up index (NDBI) and normalized difference vegetation index (NDVI). The improved LSMA method needs two steps: (1) the representative endmembers are first selected by combining a high-resolution image from Google Earth; (2) the results of LSMA are post-processed with NDBI and NDVI. During the post-process, if the NDBI value of a pixel is greater than -0.15, that pixel in the low-albedo fraction image is classified into low-albedo impervious surface; otherwise, pixels in the low-albedo fraction image are kept and classified as low-albedo pervious surface. In the low-albedo pervious surface fraction image, pixels with NDVI value less than 0.2 are classified as low-albedo soil fraction, other pixels are classified as low-albedo vegetation fraction. Finally, the impervious surface fraction is equal to the summation of high-albedo and low-albedo impervious surface fractions. The vegetation fraction can be estimated by the addition of original vegetation and low-albedo vegetation fractions. The soil fraction can be calculated by the summation of original soil and low-albedo soil fractions. Combining Landsat 8 OLI image on 18 th October, 2015 in the main urban area of Guangzhou, this improved LSMA method is then implemented to extract the fraction maps of impervious surface, vegetation, and soil. The results show that the improved LSMA with higher accurate results outperforms the conventional LSMA. This method can reduce errors in the low-albedo image. On the whole, the improved LSMA can significantly reduce the bias and root-mean-square error (RMSE). A soil conservation service curve number (SCS-CN) method is applied for simulating and analyzing the surface runoff under different return periods of precipitation. The key parameter CN in the SCS-CN is estimated with the fractions of impervious surface, vegetation, and soil and their corresponding initial CN values. The results show that high-quality impervious surface, vegetation, and soil fractions may be used to calculate the real CN. Higher CNs are located in these regions with higher impervious surface fraction except water. High CN indicates a low infiltration rate. The cumulative infiltration increases with the decreasing of CN. CN has a significant impact on the surface runoff simulation. A distinct spatial difference can be found in the surface runoff simulation. In general, if the precipitation intensity and impervious surface fraction are high, and the surface runoff is deep, the region will be easy to form waterlogging.

Key words: Normalized Difference Vegetation Index, Normalized Difference Built-up Index, Linear spectral mixture analysis, SCS-CN, runoff