Numerical modelling of “deeptime” (geological timescale) paleoclimates through the use of general circulation models has been widely developed since the 1980s. This approach is particularly useful to understand the interactions within the climate system either in states very different from present-day (e.g. last glacial maximum) or similar to what is expected with the ongoing, human-induced climate change (e.g. the Eocene thermal maximum, 55 million-years ago, characterized by very high atmospheric CO2 concentration). Still paleoclimate modelling is hampered by the computing cost of long integration with GCMs and more lately ESMs, that prevents transient climate simulations of several millions of years that would be useful to understand climate change mechanisms. The widespread strategy is thus to simulate steady-state, equilibrium climate states in response to pre-defined boundary conditions like paleogeography, greenhouse gas concentrations and orbital parameters.

In recent years, statistical approaches involving gaussian process emulators have been designed and applied for specific time periods (Van Breedam et al., 2021; Sablon et al., 2025). Others have used a combination of climate models simulations at equilibrium with  kriging on isotope-based temperature proxies to interpolate climate in time.  The aim of this internship is to design a new mathematical strategy to interpolate in time steady-state climate simulations from different geological periods, based on the knowledge of finer timescale variations of temperature, insolation and greenhouse gas coming from proxy data and reconstructions.

Requirements:

• Proficiency in Python, including experience with relevant libraries such as NumPy, Pandas,  Matplotlib. and preferably with xarray and dask

• Strong background in statistics/machine learning and data analysis, particularly applied to climate or environmental data.

• (Optional) Familiarity with climate or weather data (e.g., ERA5, CMIP6).

• Strong problem-solving skills and the ability to work both independently and as part of a  research team.

Références:

* Lguensat et al. 2017 “ The Analog data Assimilation” https://journals.ametsoc.org/view/journals/mwre/145/10/mwr-d-16-0441.1.xml

*Tandeo et al.. 2023. « Data-Driven Reconstruction of Partially Observed Dynamical Systems ». Nonlinear Processes in Geophysics 30 (2): 129‑37. https://doi.org/10.5194/npg-30-129-2023.

* Sablon et al. 2025. « An Emulator-Based Modelling Framework for Studying Astronomical Controls on Ocean Anoxia with an Application on the Devonian ». Prépublication, Climate and Earth system modeling, juin 2. https://doi.org/10.5194/egusphere-2025-1696.

* Van Breedam, et al.  2021. « A Gaussian Process Emulator for Simulating Ice Sheet–Climate Interactions on a Multi-Million-Year Timescale: CLISEMv1.0 ». Geoscientific Model Development 14 (10): 6373‑401. https://doi.org/10.5194/gmd-14-6373-2021.

* Tardif, et al.. 2025. « Generating Spatialised and Seasonal Deep-Time Palaeoclimatic Information: Integration Into an Environmental-Dependent Diversification Model ». Global Ecology and Biogeography 34 (4): e70024. https://doi.org/10.1111/geb.70024.