Remotely Sensed Estimation Method For Near-surface Air Temperature And Its Lapse Rate Over The Tibetan Plateau

Near-surface air temperature（NSAT）and near-surface air temperature lapse rate（NSATLR）are key criteria for judging atmospheric stability.They are also important parameters in glacier,hydrology,ecology,climate and meteorology in high mountains such as the Tibetan Plateau（TP）.However,the researches of NSAT and NSATLR based on remote sensing data still exist the following problems:the sparse meteorological stations and their uneven distribution over the TP results in an insufficient spatial representation of the station;affected by the clouds and complex terrain of the TP,there are some deficiencies in the spatial dimension in remotely sensed data;the estimated NSATLR lacks validation dataset.To address the above issues,the main research works of this thesis are as follows:（1）The NSATLR_in-situ values of 86 stations over the TP are obtained by linear regression of the in-situ data and DEM.The NSATLR_in-situ values are used to analyze the reliability of the remotely sensed NSATLR.Based on the in-situ data,the linear regression method is used to obtain the NSATLR_in-situ value of each station with strict control of the input data.The results showes that the annual average NSATLR_in-situ in the TP ranged from 0.44 to 9.09°C/km.（2）A novel method for estimating NSAT and NSATLR based on MOD07＿L2product is developed in this thesis.Firstly,the daily MODIS NSAT over the TP is developed entirely based on MOD07＿L2 product.The time range of MODIS NSAT and NSATLR is from 2010 to 2018.The validation results based on the in-situ NSAT show that the coefficient of determination（R²）is 0.92,root mean square error（RMSE）is2.66°C,and mean bias error（MBE）is 1.04°C.Then,based on the MODIS NSAT and DEM data,the 5 km MODIS NSATLR is estimated by combining moving window and specified criteria.Cross-comparison based on NSATLR_in-situ shows that MODIS NSATLR is in well agreement with NSATLR_in-situ in the eastern TP,with RMSE of 1.51°C/km and MBE of-0.78°C/km.Larger deviation exists in the southern TP.（3）The all-weather NSAT estimation method is constructed by spatial interpolation and random forest methods.The input parameters of this method include in-situ data,all-weather land surface temperature product（TRIMS LST）,and several auxiliary variables,and the daily 5 km TRIMS NSAT dataset are generated for the TP from 2010-2018.The validation results based on the independent in-situ data shows that the R² is 0.91,RMSE is 2.55°C,and MBE is-1.08°C.Then,the TRIMS NSATLR dataset are produced by TRIMS NSAT and DEM data.Cross-comparison based on NSATLR_in-situ shows that the RMSE value in the western TP is small,with a multi-year average of 1.32°C/km,and the systematic deviation is small.The NSAT and NSATLR estimation methods proposed in this thesis provide the possibility of generating spatially continuous NSAT and NSATLR across the entire TP and even larger areas,and providing data support for related studies such as ecology,hydrology and climatology.