PhD CIFRE with AXA and INRIA - AI Weather Models for Extreme Events Risk Estimation

:

MSc. in computer science AI data science applied mathematics statistics or equivalent.

Good experience in programming in python and deep learning libraries (e.g. pytorch...

Very good knowledge in deep learning machine learning and statistical modelling.

Advanced level in English (technical discussions presentations and paper writing are expected)

Preferred:

Education research experience projects in climate sciences extreme events risk estimation.

Previous research experience: research projects internships publications...

Recent advancements in AI weather models have shown promising capabilities in weather prediction from short to mediumterm 0 to 10 days lead time) comparable to physicbased stateoftheart models. Notable examples include FourCastNet SFNO GraphCast GenCast PanguWeather which have adopted a trend observed in other fields (e.g. language models vision models multimodal models): they leverage large architectures with an extensive pretraining on very large datasets. This type of large models (also called foundation models) facilitates the development of downstream applications.

In the context of weather forecasting these AI models are typically trained on the gold standard global reanalysis dataset ERA5 developed by the European Centre for MediumRange Weather Forecasts. ERA5 is widely used for weather and climate studies due to its high resolution comprehensive coverage and accurate representation of atmospheric conditions from 1979 to the present day. AI models such as FourCastNet or GraphCast are trained on ERA5 data: they take a snapshot of the atmosphere state at time (t) and they are trained to predict the atmospheric conditions at time (t1 (typically 6 hours later).

This PhD aims to assess and develop the capabilities of AI weather models in simulating extreme events a promising avenue for risk estimation and risk prevention. However potential issues may prevent accurate simulations: for instance model biases towards average weather conditions (e.g. reliance on RMSE metric) and the underrepresentation of extreme events in the ERA5 dataset.

The PhD (under a CIFRE scheme) will be hosted by the AI research team at AXA Group Operations in collaboration with the Natural Catastrophe R&D modelling team at AXA and the ARCHES team at INRIA. The primary objective is to develop knowledge and methodologies to better understand climate risks using AI weather models. The PhD student will evaluate and enhance the capabilities of AI weather models to simulate plausible and unobserved extreme events focusing on identifying or developing datasets and metrics to assess model performances. The PhD student will explore solutions to improve existing models or create new approaches to address the limitations in extreme event simulation. The downscaling of the simulations to a higher resolution than the native ERA5 resolution will also be addressed during the PhD. A key component of this work will involve estimating climate risks particularly to determine the optimal set of initial conditions to input AI weather models to get in fine an accurate estimation of the risk of extreme weather event across various geographical locations. While this research is open to any major climate risks (tropical cyclones extratropical cyclones windstorms hail floods heatwaves etc. the student may focus on a subset of perils.

Expected Contributions

During the thesis the PhD student is expected to produce research articles to be submitted to highquality peerreviewed ML workshops conferences and journals (e.g. ICML IJCAI NeurIPS JMLR.... Algorithmic implementations of the conceived methodology will be made available through libraries. The algorithms and methodologies developed during the PhD will be applied to realworld usecases.

Working Environment

The PhD (under a CIFRE scheme) will be jointly hosted by the AI research team at AXA Group Operations in collaboration with the Natural Catastrophe R&D modelling team at AXA and the ARCHES team at INRIA.

The PhD student will also benefit from interactions with other researchers from AXA AI Research ecosystem in particular the AXASorbonne University Joint Research Lab (TRAIL) EPFL campus in Lausanne (Switzerland) and the Stanford University campus in Palo Alto (US).

As a worldleading insurance company we act for human progress by protecting what matters. With 153000 employees in 54 countries working with 105 million customers weve created a truly dynamic and vibrant community. Inclusion and diversity link closely with our values and together were nurturing a culture of
respect for each other for our customers and the communities around us. Join AXA and youll feel like you belong are included and can thrive. Youll be able to shape the way you work and truly grow your potential as you seek out new opportunities push boundaries and benefit people in critical moments of their lives. This is your chance to build the tomorrow you want. Know you can.