Historical Changes And Future Projections Of Precipitation Phase In The Cold Season Over The Tibetan Plateau

In the background of global warming,the rapidly increasing temperature in the Tibetan Plateau(TP)has caused significant changes in climate and hydrology,which have serious impacts on socio-economic development and human water security in local and downstream areas.The Gravity Recovery and Climate Experiment(GRACE)data shows that Terrestrial Water Storage(TWS)of the TP is continuously decreasing,with a more significant decline rate in the cold season.As the main process responsible for TWS replenishment,the precipitation in the cold season over the TP has been widely studied by domestic and overseas scholars.In addition to changes in precipitation amount,changes in precipitation phase caused by climate warming are also related to TWS changes.Different phases of precipitation(solid snowfall or liquid rainfall)have varying effects on TWS replenishment,among which snowfall being particularly important for maintaining local snow cover,glaciers,and frozen ground.However,the characteristics of precipitation phase changes in the cold season over the TP and their contributions to TWS are still unclear.The simulation ability of CMIP6 also need to be studied.Therefore,we mainly used the dara form ERA5,CMFD and CMIP6 multimodels to investigate the change characteristics of precipitation phase in the cold season over the TP,clarify the physical processes causing precipitation phase changes due to climate warming that affect TWS,evaluate the ability of CMIP6 multimodels to simulate precipitation phase,and project the precipitation phase changes in the future.The main research results obtained in this article are as follows:(1)ERA5 data showed that precipitation phase in the cold season over the TP gradually transitioned from snowfall to rainfall from 1979 to 2019.The rise in atmospheric melting level height(MLH)played a crucial role in this process.MLH increased by 1m in the cold season,resulting in a corresponding increase of 0.034 mm in rainfall and a decrease of 0.061 mm in snowfall.we quantified the effects of slow processes(snowfall and evaporation)and fast processes(rainfall and runoff)on TWS using the land-surface water balance equation.It was found that the weakening of slow processes related to snowfall is responsible for reducing the effective recharge ability in most areas of the TP.The combined effects of changes in atmospheric processes(precipitation phase changes)and surface processes(e.g.,glacier melting and increased runoff)have caused significant loss of TWS over the TP.(2)Comparing CMIP6 simulation results with CMFD data from 1979 to 2017,our study found that the CMIP6 multi-model ensemble(MME)underestimated the snowfall-to-precipitation ratio(SPR).Specifically,the SPR in the southern edge of the TP was underestimated by nearly 30%.We assessed the influence of different factors on SPR and found that the correlation coefficient between SPR bias and temperature bias(-0.701)was significantly greater than that between SPR bias and total precipitation bias(-0.333),indicating that SPR bias was largely caused by temperature bias.Additionally,we evaluated the ability of 23 CMIP6 models to simulate precipitation phase in terms of time and space,and selected the top seven models as the best model group.They exhibited better performance in historical SPR than the MME,so they can be applied for future projections.(3)Two methods were used to project the future changes of precipitation phase in the cold season.The results show that the SPR will continue to decline rapidly.There were significant differences among different emission scenarios,indicating that climate warming will have great impact on SPR in the future.Precipitation phase changes in the cold season will threaten the water resources security of the TP.The results of emergence constraint showed that by the end of this century(2050-2099),the estimated SPR reduction under the SSP245 and SSP585 scenarios was-9.79% and-16.38%,respectively,which was close to the results of best model group(-9.22% for SSP245and-9.22% for SSP585).However,compared to the latter,the uncertainty range of emergence constraint method under the SSP245/SSP585 was reduced by 54.89%/58.26%,and the reliability has significantly improved after emergence constraint.