Characterization of lake morphology in China using remote sensing and GIS

Abstract

Based on 2010 Landsat TM / ETM+satellite remote sensing imagery data and referring to Google Earth imageries and other historical documents,the authors used object-oriented classification method to extract inland water bodies across China. The authors set imagery segmentation scale at 30 m,and extracted water bodies with ratio index,Normalized Difference Water Index( NDWI) and spectral relation. There are three types of water,i. e.,rivers,lakes,and reservoirs. Finally,the spatial distribution of lakes in China was obtained. Using Arc GIS software package,lake area,shoreline length,shape index( SI),related circumscribing circle( RCC),fractal dimension( FD) and other landscape indices,the authors conducted calculation,and divided China into five lake districts according to previous researches. The differences of lake morphology were comparatively analyzed in the five lake districts. The results show that the object-oriented classification method has better classification accuracy. The overall accuracy of the three types of water is 93%,and the classification accuracy of the lake is over 90%. Statistically,the number of lakes with area greater than 1. 0 km2 is 2 477,totaling 77 934. 72 km2. The Qinghai-Tibet Plateau Lake District( QTPLD) has the largest lake area,accounting for 54. 34% of the total lake area in China. The averages of SI and FD are the minimum,thus the lake morphology is simple in the QTPLD.However,the rainfall is abundant and the water system is well developed in Eastern Plain Lake District( EPLD).Its averages of SI and FD are the maximum among all the sub-lake regions. The lake shoreline is more zigzag,and hence the lake morphology is more complex in the EPLD. In addition,SI and FD have significant correlation( R2=0. 95). The higher the FD of the lake,the greater the observed SI.