Séminaire
In order to address the problem of attribution of climate extreme events, Trenberth et al (2015) suggested to regard the extreme circulation regime as being largely unaffected by climate change. This heuristic approach is called conditional event attribution.
To implement this approach, one needs to be able to simulate the event in question with simulations whose circulation is constrained in some manner, and then assess the impact of known anthropogenic changes in the climate system's thermodynamic state. Here, we propose to do so using data assimilation. We test our implementation with a realistic intermediate complexity atmospheric model(ICTP AGCM, nicknamed SPEEDY), based on a spectral dynamical core.
A synthetic dataset of observations of two extreme precipitation events over North America is extracted from a long SPEEDY simulation under present climatic conditions (i.e. factual conditions). Then, observations of the circulation regime during these events are assimilated twice in the SPEEDY model: first in the factual configuration
of the model and second under its counterfactual, pre-industrial configuration.
We show that conditional attribution can be performed using data assimilation and that our implementation helps identify the specific physical features of the event and its causal signature.