Applicability evaluation and correction of CLDAS surface temperature products in permafrost region of Qinghai-Tibet Plateau
Land surface temperature(LST)is a crucial parameter to characterize the surface thermal state and conduct research on the surface hydrothermal and ecological processes.The available LST data in the permafrost region of the Qinghai-Tibet Plateau mainly include the observed data of land surface and shallow ground temperature,remote sensing derive data,model simulation data and reanalysis data.CLDAS dataset has a good performance in most regions of China.However,due to the lack of measured data in permafrost regions of the Qinghai-Tibet Plateau and insufficient consideration of the underlying surface of permafrost,the applicability of CLDAS in the Qinghai-Tibet Plateau needs to be further evaluated.Based on the measured LST data of seven stations in the permafrost region,the CLDAS LST data from 2008 to 2018 were evaluated in different freeze-thaw periods and underlying surface types.The results showed significant errors between CLDAS and the measured values(bias=2.09℃,MAE=3.64℃,RMSE=4.67℃,R2=0.83),mainly performed that CLDAS overestimated the measured LST.The applicability of CLDAS LST is good in the thawing period,but poor in the freeze-thaw alternating period(MAE=3.78℃)and freezing period.And it's better in the alpine desert and alpine desert steppe than that in alpine meadow.Therefore,a multiple stepwise regression correction model was established based on the influences of NDVI,NDSI,snow depth,elevation,slope,aspect and soil texture factors on LST.The correction model took the differences of underlying surface conditions into account and improved the simulation accuracy of CLDAS LST.The results show that the correction models constructed by freezing period,thawing period and alternating period separately performed better than a single model.The accuracy of corrected CLDAS LST by three-period models was significantly improved(bias=-0.11℃,MAE=2.42℃,RMSE=3.23℃,R2=0.89).
Land surface temperatureCLDASQinghai-Tibetan PlateauApplicability evaluationMultiple regression correction
国家科技期刊平台
NSTL
万方数据
