| The Tibetan Plateau(TP),known as Earth's“Third Pole”,its warming rate has been significantly higher than the global average under the background of global warming in the past few decades.Meanwhile,the TP affects profoundly the atmospheric circulation in the upstream and downstream regions,even the world due to the mechanical and thermal forcing of the unique TP-topography.Currently,the TP observational data are still insufficient,the land surface energy and water budget of the TP,as well as the interactions between the TP and the atmospheres,and surrounding areas need to be investigated in depth.Therefore,the better understanding of the land surface processes and more realistic descriptions of the surface energy and water budget have become a key part of the TP land-atmosphere interactions.In view of the current problem in the field on land surface simulations in the TP,this paper selects the important variable,land surface temperature(LST),which characterizes the land surface energy budget and the energy exchange between land surface and the atmosphere,as the pointcut.We adopted the CLM(Community Land Model)as the main tools,combining with in situ observations and satellite remote sensing data.First,this study developed an improved vegetation emissivity scheme,analyzed the influences of vegetation emissivity on the land surface energy budget and the snow processes of the ground,and further discussed the applicability of this newly developed scheme in the Northern Hemisphere(NH).Secondly,we conducted a long-term and high spatial resolution offline CLM simulation over the TP,evaluated the diurnal LST simulations,and improved the accuracy of the simulated LST by optimizing the physical parameterization schemes.At last,we utilized four atmospheric forcing datasets to analyze the sensitivities of simulated LST to uncertainties in forcing variables and quantified the actual influence s of these uncertainties.The main conclusions are drawn:(1)Developed an improved vegetation emissivity scheme for the CLM version 4.5(CLM4.5)to more accurately simulate the vegetation emissivity.The original scheme of vegetation emissivity can capture the trend of the vegetation emissivity increases with the increase of vegetation leaf and stem area indices;however,this scheme produced an unreasonably low vegetation emissivity(0.70–0.80)when compared with MODIS(Moderate-resolution Imaging Spectroradiometer)emissivity(?0.98),especially during winter and spring.Thus,we developed a new vegetat ion emissivity scheme based on the maximum vegetation emissivity,vegetation leaf and stem area indices,and different plant function types,which can simulate vegetation emissivity more realistically(?0.95)than the original scheme.(2)The newly vegetation emissivity scheme can more accurately simulate the ground snow cover fraction(SCF)in the NH during winter and spring.The new scheme generates larger ground surface net longwave radiation than the original scheme,which implies that more downward longwave radiation came from the vegetation due to the higher emissivity values in the new scheme.O n the other hand,the higher emissivity values from the new emissivity scheme result in a lower vegetation temperature,evaporation from the wetted vegetation is weakened with the new scheme;in the meantime,the colder vegetation attracts more dew/frost to the vegetation surface.Less evaporation and/or more dew/frost on the vegetation surface with the new scheme produce more canopy water.When canopy water exceeds the maximum amount of water the canopy can hold,the water drips off the canopy to the ground surface.Thus,we see a larger snowfall amount on the ground surface during the cold season with the new emissivity scheme.These energy changes and snowfall on the ground surface with new scheme mostly reduce the simulated SCF errors when compared with MODIS SCF data over mid-high latitudes.In general,about 200 and 350 thousand km~2areas for winter and spring experience a decrease in simulated SCF RMSEs(Root-Mean-Square-Errors)with the new scheme,respectively.(3)Evaluated and improved the ability of CLM version 5.0(CLM5.0)in simulating daytime and nighttime LST over the TP,revealed the factors that limit the accuracy of LST simulation from the perspective of physical parameterization schemes.CLM5.0 was forced by CMFD(China Meteorological Forcing Dataset),and its simulations can well capture the spatial distributions of daytime and nighttime LST over the TP(Pattern Correlation Coefficient,PCC>0.75);however,the simulated LST has significant biases that have strong spatial variability when compared with MODIS LST products.The cold biases of modeled daytime LST over bare ground were improved by adjusting the scheme of ground sensible heat roughness length.We optimized the soil thermal conductivity parameterization by introducing the effects of soil organic matter and soil gravel through volume fractions,decreased energy transfer from the surface(deep soil)to deep soil(surface)during daytime(nighttime).The modification of soil evaporation resistance scheme,which is more suitable for the sandier soil of the TP,increased the soil evaporation and then reproduces more accurate evapotranspiration and surface soil moisture.Above three revisions of physical parameterization schemes improve the CLM5.0 simulated LST diurnal variations over the TP.Additionally,the deficiencies of simulated SCF and the atmospheric forcing data were discussed to investigate the possible reasons associated with the LST biases.(4)Evaluated the ability of different atmospheric forcing datasets to simulate the LST over the TP,explored the sensitivities of LST simulation to uncertainties in every forcing variable,and quantified the actual influence of these uncertainties.Four atmospheric forcing datasets,CMFD,CRU-NCEP(C limatic Research Unit-National Centers for Environmental Prediction),GSWP(Global Soil Wetness Project),and WFDEI(Water and Global Change Forcing Data/ERA-Interim),were used to drive the CLM5.0.All simulations can generally capture the spatial distribution of LST over the TP(PCC>0.50),and the results from CMFD agreed best with the characteristics of LST diurnal cycle,seasonal,and annual average over the TP.Our sensitivity analysis showed that seasonal LST is highly sensitive to the uncertainty in air temperature,whereas has lowest sensitivity to the precipitation uncertainty.However,actual influence of uncertainties in forcing variable s to LST simulation uncertainty suggested that longwave radiation,precipitation,air temperature,and solar radiation uncertainties among datasets account for about 39%,19%,18%,and 16%of the total LST variability caused by all the forcing uncertainties for annual average LST over the TP.In addition,precipitation uncertainties account for about 34%and 38%of the LST simulation uncertainties for western TP during winter and sprin,respectively.The strong uncertainties of precipitation data significantly affect the LST simulation of the western TP by affecting the uncertainties of SCF and then absorbed solar radiation during the cold season.This study focused on the evaluation and improvements of land surface processes simulation,and further explored the key factors affecting the LST simulation over the TP.We developed an improved vegetation emissivity scheme to more accurately simulate the effects of vegetation emissivity on land surface energy budget and snow processes in the N H over winter and spring.We also evaluated the modeled LST diurnal cycle and improved the simulation accuracy of the LST diurnal cycle by optimizing the physical parameterization schemes over the whole TP.Meanwhile,we quantified the sensitivities of LST simulation to uncertainties in atmospheric forcing datasets and further estimated the actual influence of these uncertainties.O ur study gives a strong insight into land surface energy and water budget conditions the data-sparse regions in the TP,provide an effective tool to generate better understandings of the TP land surface processes,and further make a foundation for exploring more realistic land-atmosphere interactions on the TP. |
PDF Full Text Request