Author(s): Sooyoung Kim; Jun-Haeng Heo; Joon Heo; Sunghoon Kim
Linked Author(s):
Keywords: Impervious surfaces estimation; Landsat-7 ETM+; High resolution satellite image; Regression tree algorithm; Tasselled cap transformation; NDVI
Abstract: Impervious surface has an influence on urban flood modelling during the rainy season. The increase of impervious surfaces causes meteorological and hydrological changes like urban climate change, urban flood discharge increasing, urban flood frequency increasing, peak discharge increasing and faster concentration time in urban area. Accordingly, the estimation of impervious surfaces is an important factor of urban flood analysis. In this study, the estimation of impervious surfaces is performed by using regression tree algorithms and remote sensing images such as landsat-7 ETM+ and high resolution satellite image for case study area – Jungnang-cheon basin in South Korea. Moreover, a tasselled cap transformation and NDVI transformation apply to landsat-7 ETM+ for various predictive variables. In addition, regression tree algorithm is applied to the training data sets which are collected by overlay between landsat-7 ETM+ and high resolution satellite image for the estimation of impervious surfaces. As a result, the combination of band 3, 4, 5, and 7 of landsat-7 ETM+, the greenness of a tasselled cap transformed image, and NDVI shows the highest correlation between the real values and predicted values of impervious surfaces. Finally, the impervious surfaces map is accomplished by the prediction model.
Year: 2007