Retrieve soil moisture from mixed-pixels based on scale transformation using hyperspectral data

Soil moisture is a key parameter for drought monitoring. Crops distribute so fragmentally in China that mixed pixels account for a large proportion in moderate and coarse resolution remote sensing images. The soil moisture retrieved from vegetation-soil mixed pixels is a very important problem for drought monitoring and ecological study. Focusing on vegetation-soil mixed pixels, a new method for retrieving soil moisture from hyperspectral data is provided based on scale transformation method. Yingke Oasis, Zhangye, Gansu province was selected as validation area. A Hyperion/EO-1 data acquired on Jul.15, 2008 was pre-processed and linearly interpolated to 180m and 1080m resolution images. Then a multi-scale image series was obtained. Using the above method, the soil moisture of pixels whose space resolution is 1080m were calculated. The retrieved results were verified by synchronized ground observation data. The results show that the proposed method is reliable.