A New Method for Fast Information Extraction of Water Bodies Using Remotely Sensed Data

The construction of digital watershed plays an important role in the modernization process of water resources management and utilization in China.Using remote sensing technology to extract water information has become a significant way of providing data resources for digital watershed construction.In this paper,the unique characteristics of water bodies' simultaneous strong absorption at near and middle infrared bands(corresponding to the Band 4,Band 5 and Band 7 of Landsat TM/ETM+ imagery,respectively) has been revealed by a careful study of spectral features of water bodies with different surrounding land use/cover types,and based on that,a new water index(NWI) for water information extraction has been proposed.The new index distinguishes itself from other conventional water indexes by its usage of Band 7 of Landsat data which has rarely been reported in the literature.Using a ETM+ image acquired on March 4,2001(path 119,row 43) and taking Xiamen Island as study area,the NWI,with atmospheric correction,has been tested in the reservoir,lake,sea and cage culture areas with various backgrounds.The overall accuracy of the water information extraction result reaches 90.4%.Therefore,it can be concluded that the NWI is effective in quickly and accurately extracting water information using remotely sensed data.