[Validation of vegetation coverage and NDVI based on UAV remote sensing method and its response to hydrothermal...

To verify the accuracy of MODIS-NDVI data products in deserts and provide guidance for scientific management of desert grasslands in the context of climate change, we examined the responses of fractional vegetation cover (FVC) and normalized difference vegetation index (NDVI) to hydrothermal gradient in arid desert areas using unmanned aerial vehicle (UAV) remote sensing. In Alxa desert region of Inner Mongolia, GreenSeeker handheld spectrometer was used to obtain NDVI (NDVI_R) of 100 sampling points. NDVI was extracted by MODIS-NDVI data products (NDVI_M), and the accuracy of NDVI_M was verified by NDVI_R. FVC of each sampling point was obtained through unmanned aerial vehicle remote sensing (FVC_U), which was used to examine the FVC that was retrieved by the pixel binary model (FVC_M). In addition, combining meteorological data, we examined the responses of FVC and NDVI to hydrothermal gradient based on UAV remote sensing method. The results showed that MODIS-NDVI data products reflected the real NDVI in Alxa area with an accuracy of 84.2%, but NDVI_M were 15.7% higher than the actual values. FVC_M reflected the vegetation coverage of Alxa region with an accuracy of 83.1%, which were 14.8% lower than the real value. Effects of meteorological factors on NDVI was different, depending on the different acquisition methods. NDVI was affected not only by temperature and precipitation, but also by ground temperature, evaporation and the interaction between evaporation and ground temperature. Because of the different degree of atmospheric influence, NDVI_M was more affected by ground temperature, evaporation and precipitation than NDVI_R, while NDVI_R was more affected by temperature than NDVI_M. To examine the changes of vegetation coverage across hydrothermal gradient in desert area, we should consider not only precipitation and temperature, but also the interaction between evaporation, ground temperature and meteorological factors. The interaction between temperature and rainfall, evaporation and ground temperature, and between temperature and evaporation had greater impacts on FVC_U.