Les 25 meilleures formations RSE en 2025
Evaluating biases in climate impact drivers and extreme indices, at scale
Overview
You will work closely with Alexandre (co-founder & CTO) as well as academic partners from TRACCS to compute indicators from climate simulations and evaluate biases & uncertainty, at scale. The results will be integrated into Azard’s product to help companies assess their climate risks and may result in a publication.
Context
Climate change is increasing the frequency and intensity of natural hazards such as heatwaves, droughts, floods, and extreme precipitation. Climate risk assessments help organizations anticipate these risks and invest in the relevant adaptation measures.
These assessments rely on climate indicators, or climatic impact drivers [1], which are computed from raw climate variables produced by global and regional climate models. However, model outputs can suffer from biases and typically have coarse spatial resolution, limiting their direct use. Bias-adjustment and downscaling methods are commonly used to make them more suitable for local-scale impact studies, but also have limitations, especially for multivariate indicators and extremes.
This makes the evaluation step crucial: comparing climate variables or indicators computed from simulations with those derived from historical reference datasets and observations, to identify biases in existing methods and data [2]. In addition to the scientific questions it raises, this evaluation task is also an engineering challenge. Climate data quickly reaches terabyte scale when working globally and across multiple models and scenarios.
What you will do
The goal of the internship is to build, step by step, an evaluation framework to quantify biases in climate indicators computed from CMIP6 simulations by comparing against reanalysis datasets (e.g. ERA5) and/or in-situ observations. You will:
· Compute climate indicators on a global scale (xarray, dask). Appropriate computing resources will be provided (cloud-based or HPC).
· Identify and analyze biases in climate variables and computed indicators.
· Understand shortcomings of statistical downscaling and bias-correction approaches, explore alternatives.
Bibliography
[1] Intergovernmental Panel On Climate Change (Ipcc), Climate Change 2021 – The Physical Science Basis: Working Group I Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, 1st ed. Cambridge University Press, 2023. doi: 10.1017/9781009157896.
[2] D. Allard, M. Vrac, B. François, and I. García De Cortázar-Atauri, “Assessing multivariate bias corrections of climate simulations on various impact models under climate change,” Aug. 28, 2024, Hydrometeorology/Modelling approaches. doi: 10.5194/hess-2024-102.
