Using Water Stable Isotopes to Assess past, present and future tropical precipitation change – ISOTROPIC

Projections of tropical precipitation strongly differ from one climate model to another, both in sign and in amplitude. How to assess which of the model projections are likely to be the most reliable? By identifying the models that are most reliable both in simulating the processes that control present-day precipitation and in reproducing precipitation changes associated with past climate changes, one may reduce the uncertainty. To attain this objective, a four-step strategy will be followed:
(1) To evaluate paleoclimate simulations realism, we need to reconstruct past precipitation changes in the tropics. For this purpose, we will use the isotopic composition (deuterium and oxygen 18) of past precipitation stored in some well-known climate archives. This requires refining the relationship between the isotopic composition of precipitation and the precipitation amount (i.e. the isotopic calibration). Based on new isotopic observations in water vapor and in precipitation (deuterium, oxygen 18 and 17, and their combinations), we will improve our knowledge of isotopic fractionation between water vapor and precipitation phases to provide a robust calibration. We will focus on two contrasted regions of the tropics where climate archives were studied and where precipitation projections are of high societal significance but also particularly uncertain: the Sahel in Western Africa and the Andes in Western South America.
(2) In turn, we will explore how those new present-day isotopic observations may help us to decipher and to quantify the influence of some physical processes playing a key role in convective atmospheres (rain reevaporation, subsidence) and to evaluate moist biases in climate models fitted with water stable isotopes (LMDZ-iso, ECHAM-iso and models distributed from the SWING 2 data base) which allow a direct comparison between observed and simulated isotopic data, independently of the isotopic calibration. The best models arising from this comparison will be used to go one step further in the isotopic calibration (temporal stability? Spatial homogeneity?).
(3) Thanks to the first step, we will provide robust reconstructions of past precipitation changes in Western Africa and Western South America from archives as ices cores, speleothems, ostracods, etc. covering the Last Glacial Maximum and/or the Holocene, periods simulated by climate models. Therefore, we will have the opportunity to explore past climate variability from interannual to glacial-interglacial timescales as well as tropical climate teleconnections and sensitivity.
(4) Then, to attain our initial objective, we will analyze multi-model simulations from the international CMIP5 project (a coordinated set of present-day, past and future climate change simulations performed by the same ensemble of models and thus having the same physics from one period to another for a given model) to evaluate the ability of climate models to reproduce both present-day and past precipitation according to our reconstructions.
Finally, we should be able to select the climate models that present the highest skills in precipitation changes evaluation.
This project thus proposes an interactive and coupled model-data approach to evaluate climate models and to assess which of the tropical precipitation projections seem the most credible. In turn, by answering this burning question, we will progress on our knowledge in three major fields in climatology: the physic of water stable isotopes and its use in climatology, the past precipitation variability in the tropics and the present-day key processes controlling convective atmospheres.