The Influence Mechanisms And Simulations Of Precipitation Stable Isotopes Over The Southern Tibetan Plateau

The Tibetan Plateau,as the "Third Pole" of the earth,is drawing worldwide interest for its sensitivity in terms of climate change.It contains a large number of glaciers,from which ice cores can be drilled and information on past climates can be inferred from the archived isotopic composition of the precipitation.The exact processes controlling the variability of precipitation isotopic composition need to be deciphered and understood.This research helps to understand the modern hydrological processes responsible for the isotopic variability over the TP.Moreover,it is necessary to evaluate model biases based on the precipitation isotopic seasonality,which contributes to understanding the influence of model biases on the simulation of the isotopic composition for past climates and provides a valuable reference for interpretation on paleo-climatic record based on iso-GCMs simulations.This study investigates the factors controlling the variability in the isotopic composition of precipitation are investigated at the synoptic,intra-seasonal and seasonal scales,which is achieved using precipitation isotopic composition observed in southeastern Tibetan Plateau and water vapor isotopic measurements derived from satellites data.The ability of the general circulation models to capture the variability and its controlling processes is also evaluated.Moreover,the factors controlling the inter-model spread in precipitation isotopic seasonality are discussed in the southern and northern Tibetan Plateau,which is used to evaluate the model errors.Main results of our study are as follows:1)The ?D and ?18O of precipitation sampled from Lijiang city exhibit the following relationship:?D=7.97?18O+3.66.The ?18O values of precipitation range between-23.60‰ and 2.70‰ with a mean of-14.8‰ and a standard deviation of 5.24‰,while the ?D values of precipitation range between-185.00‰ and 18.20‰ with a mean of-87.1‰ and a standard deviation of 42.25‰ from March,2017 to August,2018.The slope of the local meteoric water line derived from three sampling sites at Mount Meili is around 8,which is near the slope of global meteoric water line.The?18O values of precipitation vary from-28.02‰ to 4.66‰ with a mean value of-12.89‰and a standard deviation of 6.30‰,while the ?D values of precipitation vary from-210.10‰ to 41.54‰ with a mean of-96.94‰ and a standard deviation of 50.29‰.At the seasonal and intra-seasonal scale,the ?D of precipitation at Lijiang show an obvious correlation with local wind direction.The significantly negative relationships between precipitation ?D and relative humidity and temperature are also found at the seasonal scale.The "temperature effect" and "precipitation amount effect" are obvious for the?D and ?18O of precipitation with a low correlation coefficient at the synoptic scale.Local meteorological variables are not sufficient to explain the precipitation ?D variability.These results highlight the importance of upstream deep convection and rain evaporation process.2)Combined with the observed precipitation ?D and water vapor ?D derived by the IASI satellite,the different processes are quantified at seasonal to synoptic scale.At the seasonal scale,the isotope composition in precipitation is controlled by processes along air mass trajectories and by local processes.Firstly,local processes and are main drivers.The third contribution ?·(Rveq-RvLS)contributes to 69%of the precipitation ?D variations.The role of local circulation is further supported by the correlation between the third contribution and local wind direction(r=0.97,p<0.01).The role of rain evaporation is also further supported by the correlation between the third contribution and relative humidity(r=-0.91,p<0.01).The isotope composition in precipitation is also controlled by processes along air mass trajectories affecting the water vapor at the large scale(>200 km),especially upstream deep convection and air mass origin.The second contribution ?·RvLS contributes to 27%of the precipitation?D variations and ?D in precipitation is significantly anti-correlated with the cumulated precipitation amount at 2 days preceding the rainy event(r=-0.89,p<0.01),highlighting the important role of deep convection along air mass trajectories.This could also include an effect of the moisture origin:the second contribution also correlates significantly with local wind direction(r=0.62,p<0.01),which at the seasonal scale reflects both local and large-scale circulation.At the intra-seasonal scale,the third contribution ?·(Rveq-RvLS)contributes to 115%of the precipitation ?D variations.This indicates that processes(rain evaporation and local circulation)are the main drivers.However,precipitation ?D is negatively correlated with the second contribution ?·RvLS,indicating that processes along trajectories blur or dampen the precipitation ?D variability.At the synoptic scale,local processes transforming the large-scale water vapor isotopic composition into the precipitation composition variability dominate,contributing to 68%of the variability in precipitation ?D.Small-scale vertical and horizontal heterogeneities in water vapor ?D could be the main drivers of this contribution.We thus highlight the dependence of the isotopic drivers on the time scale of variability.3)Based on the water vapor ?D derived from the TES and GOSAT satellite,simulations from iso-GCMs and the observations,the spatial scale of precipitation ?D signal from seasonal to synoptic time scale at Lijiang is further explored.The precipitation ?D at Lijiang show a consistent seasonal variations in a wide spatial range,and is strongly correlated with precipitation ?D at the three sites at Mount Meili.Moreover,the water vapor ?D at Lijiang correlates significantly with the water vapor?D everywhere around(r>0.8).At the intra-seasonal and synoptic scale,the correlation becomes insignificant with other sites at Mount Meili,and the spatial scale of precipitation ?D variations have a smaller spatial imprint,and thus the importance of local processes.The shorter the time scale,the smaller the spatial imprint of isotopic variations.4)With the monthly output from the SWING2 models and daily simulations from the LMDZ-iso model of precipitation and water vapor ?D,we have quantified the relative contributions of different processes to the isotopic variability at Lijiang,and to what extent are these different contributions and isotopic variability captured by the models?Results show that isotope-enabled general circulation models can qualitatively capture the seasonal variations in ?D,but the contribution of the second contribution?·RvLS simulated by SWING2 and LMDZ-iso is overestimated,with greater than observation(27%),and has a lower third contribution ?·(Rvep-RvLS)than that of observation(69%).At the intra-seasonal and synoptic scale,LMDZ-iso also overestimate the contribution of processes along trajectories and underestimate the contribution of local processes.IASI errors can not dramatically modify the results and the coarse horizontal resolution may be the main reason.5)We use the isotopic seasonality as a benchmark and explore what is the skills of iso-GCMs to simulate the stable isotope seasonality of the precipitation over the Tibetan Plateau?It could be found that the inter-model spread in precipitation isotopic seasonality is related to the inter-model spread in climatic seasonality.In the southern Tibetan Plateau,the inter-model spread in isotopic seasonality is controlled by the precipitation seasonality,whereas in the northern Tibetan Plateau,it is affected by the zonal wind seasonality.If the monsoon circulation extends further north,then zonal wind seasonality is more negative.At the same time,vapor isotope in June-July-August is more depleted because more vapor comes from the monsoon convection region,leading to weaker ?18O seasonality.For an iso-GCM to simulate isotopes well,we need an iso-GCM that simulates the climate well.