Research On Downscaling Of Temperature And Precipitation Factors In Qinba Mountainous Area

Temperature and precipitation are the two meteorological elements most closely related to humans.High-resolution temperature and precipitation data not only provide people with convenience in travel and production,but also important in regional climate change and natural ecological environment research.Qinba Mountain Area is located in western China,with complex topography,high average altitude and few meteorological stations.Therefore,there is a great deviation in the meteorological research of this region only depending on the meteorological station data.With the development of satellite remote sensing technology,we have been able to obtain meteorological data with higher spatial and temporal resolution.However,for some specific studies,the existing data still can not meet the requirements.This paper takes Qinba Mountains as an example,Produce high spatial resolution temperature and precipitation data by downscaling,regression fitting and interpolation,in order to provide reference for climate change and ecological environment in the region.In this paper,use MODIS NDVI,DEM,slope as the impact factor,constructed geographically weighted regression(GWR)model for the Qinba Mountains.Using GWR model,downscaled the surface temperature and precipitation satellite data from 2001 to 2017,and estimate temperature by surface temperature downscaling results,obtained the 250 m spatial resolution temperature and precipitation data in Qinba Mountain Area.Finally,analyze the reliability of the GWR data.The conclusions are as follows:(1)in recent 17 years,NDVI increased steadily in Qinba Mountains,the NDVI value and the NDVI growth rate increase first and then decrease with altitude,maximum value at 1500 m altitude.Vegetation coverage has the greatest response to surface temperature during the same period,while the response to precipitation has hysteresis,and the lag time is one month.(2)GWR downscaling method can improve the spatial resolution of MODIS surface temperature and TRMM precipitation data.The downscaling results are consistent with the original data in spatial and temporal distribution,and can highlight more data details.High-resolution temperature data derived from downscaled surface temperature is also spatially and temporally consistent with station data.(3)By comparing GWR data with raw satellite data,it found that the precision of surface temperature scaling results will decrease in summer,root mean square error increased by 0.2?;The precision of the GWR precipitation data is consistent with the original data,the root-mean-square error is basically unchanged but decreases in winter;The precision of high-resolution air temperature data obtained from the scaling down results of surface temperature is higher,highly correlated with ground observation data.(4)Overall,the data produced by the two methods of GWR and Anusplin are not much different.But,at high altitudes,GWR data is superior to Anusplin data in terms of correlation,accuracy,and applicability.