The retrieval of snow grain size and subpixel of snow cover using the domestic hyperspectral data

The remoter sensor from TG-1 platform can provide the hyperspectral and high-resolution images. In the present study, the snow cover fraction and snow size were retrieved from domestic hyperspectral short-wave infrared (SWI) data at Heihe upstream on Qilian Mountains to test its application in the civilian area. A new algorithm that combined Vertex Component Analysis (VCA) component automatic extraction techniques with sparse regression pixels unmixed techniques were developed and to produce snow cover fraction map. The results were verified using the higher spatial resolution images that also been achieved by TG-1 platform. Initial analysis indicated the root mean square error(RMSE) and the correlation coefficient of both validate area was 0.24 and 0.27, 0.72 and 0.84 compared with reference images, respectively. In addition, the ART radiative transfer theory referred to snow particle shape have been approved by the optimizing the shaper factor and retrieval band. The improved algorithm then provided the snow grain size map by domestic hyperspectral data. The maps were validated indirectly by hyperion data which covered the similar field area. The results showed that the domestic hyperspectral data were appropriate to produce SFC and snow grain size map and were feasible to hydrological and climate models.