Evaluation And Projection Of Terrestrial Evapotranspiration And Its Components Over China Based On The CMIP6 Models

Terrestrial evapotranspiration(ET)is a key climate variable that links water,carbon and energy exchange between the land surface and the atmosphere.The ET mainly includes vegetation transpiration(T),soil evaporation(Es)and canopy interception evaporation(Ec),which represents the migration process of water in soil-vegetation-atmosphere system.In recent decades,China has been one of the sensitive areas to global climate change.Therefore,it is important to accurately simulate and reasonably predict land ET and its components in China.At present,the latest climate models participated in the sixth phase of the Coupled Model Inter-comparison Project(CMIP6)are important tools for understanding the climate system and predicting the climate change.Thus,a comprehensive evaluation of the simulation performance of CMIP6 models in terrestrial ET and its components over China is helpful to clarify the development and shortcomings of the models and provide a basis for model improvement.In addition,reasonable projections of ET and its components over China based on CMIP6 models can provide reference for water resources management and climate regulation.However,there is still a lack of systematic research in the evaluation and projection of ET and its components in China based on CMIP6 models.Therefore,this study conducted the following researches:Firstly,based on the site observations and four remote sensing ET products,this study systematically evaluated the performance of 49 CMIP6 models on ET estimations in China from 1982 to 2014.Secondly,the uncertainties in ET components from CMIP6 models and its causes were investigated.Finally,through the model filtering and voting,this research evaluated the future(2100-2025)changes of ET and its components under four Shared Socioeconomic Pathway(SSP126,SSP245,SSP370 and SSP585 scenarios).The main conclusions are as follows:(1)At the site scale,the R~2between the 49 CMIP6 models ensemble mean and the observed ET was high(0.55).However,there was an overestimation in ET from CMIP6 models among more than two-thirds of the sites,with an average overestimation of 21%.At the regional scale,the spatial patterns of multi-year(1982-2014)mean ET obtained from CMIP6 models showed the great consistence with that from the remote sensing ET products.However,compared with the results from remote sensing ET products(415-440 mm/year),47 CMIP6 models overvalued the annual mean ET,with the values ranging from 339 to 737 mm/year across them.Moreover,the ensemble mean overestimated ET in more than 90%of the China’s land area,and it exhibited the higher reliability in southeast China than that in northwest China.In addition,through the overall evaluation of the annual ET,spatial distribution and trend,it was found that the performance of FIO-ESM-2-0 and GFDL-CM4 models were the best among the all CMIP6 models,and the performance of the CMIP6 models ensemble means was also better than that of most individual models.(2)The spatial distribution of ET components obtained from the CMIP6ensemble mean was reasonable,and highly correlated with the spatial patterns of land cover.However,there were significant discrepancies in ET partitioning among the models:the transpiration ratio(T/ET),soil evaporation ratio(Es/ET)and canopy interception ratio(Ec/ET)in China fluctuated from 18-66%,25-68%and 6-25%,respectively.These uncertainties in ET partitioning among the models were significantly correlated with the bias of simulated LAI.The CMIP6 models overall underestimated T/ET but overestimated LAI,especially in forest ecosystems.This underestimation of the relationship between T/ET and LAI is mainly due to the fact that the models severely underestimate the sensitivity of terrestrial T/ET to LAI((?)(T/ET)/(?)LAI).The(?)(T/ET)/(?)LAI from CMIP6 models ranged from 0.01-0.17 per m~2/m~2,while the result calculated by the satellite observations was 0.18(?)0.04 per m~2/m~2,indicating that the role of vegetation in the land-atmosphere interactions has not been represented well in these models.In addition,it was further found that the bias in(?)(T/ET)/(?)LAI is mainly due to the bias in the magnitude of LAI,which leads to the bias in greening rate.Therefore,the simulation performance of CMIP6 models for LAI should be improved in time.(3)In the future period(2025-2100),ET and its components will increase under all SSP scenarios,and the upward trend will increase as the SSP level increases.The growth of ET mainly occurs in the southeast China with high LAI region,and the area of rapid rise will expand to the northwest China with the increase of scenario level.Compared to historical periods,Ec/ET and T/ET will increase and decrease,respectively,at the end of the 21st century under all scenarios;Under the SSP585scenario,T/ET and Es/ET in the eastern China with high LAI region will decrease and increase significantly,respectively.