QUINTET
Quantifying uncertainties, tuning and equilibrating climate models
Project Outline
Take into account the uncertainties of climate models in current climate simulations and future projections.
Project lead:
Julie DESHAYES (CNRS – LOCEAN-IPSL)
Co-leads:
Aurore VOLDOIRE (Météo-France – CNRM)
Romain ROEHRIG (Météo-France – CNRM)
The core project QUINTET addresses the scientific and technical issues related to uncertainties in climate models (parameter calibration, simulation initialisation and uncertainties related to the spatial resolution of the models) and how these are taken into account in current climate simulations and future projections.
Quantifying uncertainties in climate simulations remains a challenge despite the efforts made, in particular in the CMIP6 project. There is an urgent need to adopt uncertainty quantification techniques that are less costly in terms of human and computational resources, while at the same time meeting societal needs for climate services. Parameter calibration is crucial, but has so far been mainly empirical.
A new semi-automatic framework developed by the French climate modelling community integrates uncertainty quantification methods and artificial intelligence to explore different acceptable configurations. However, the application of this framework to different timescales remains a challenge, as does the validation of models over transient periods.
The PC6 QUINTET aims to address these issues, with the overall objective of facilitating the exploration of parametric uncertainty (by generating perturbed physical ensembles and evaluating low-probability, high-risk scenarios) and preparing for higher resolution configurations. Emphasis will be placed on the representation of the current climate, the balancing of long time scales and the consideration of these uncertainties in transient simulations, i.e. past, present and future climate change.
Keywords: uncertainty quantification, parameter calibration, acceleration of spin-up to equilibrium, perturbed physics ensembles (PPE)
Tasks
Our Research
Improve MSEs: calibration and parametric sensitivity to current climate (time scales of 1 to 50 years)
Identify the relevant metrics for validating current climate models.
Develop existing tools to better estimate the parametric sensitivity of models for these metrics.
Assess the representation of variability mechanisms in current climate models.
Balance long time scales (>100 years)
Identify the information to be extracted from climate model trends on long time scales.
Reduce the calculation time (and associated cost) needed to bring models into equilibrium.
Separate the numerical drift of models from the mechanisms of climate variability, for future projections.
Current transient climate (time scales: approximately 100 years of climate change)
Identify constraints derived from observations that can be used to reduce uncertainty on the basis of available climate trends.
Extend the scope of climate projections to new variables and finer spatial scales.
The main objective of PC6 QUINTET is to quantify the uncertainties in climate simulations with the two French Earth System Models (ESM) and to provide observationally constrained projections as baseline references for climate services, by combining the latest observations, AI-based techniques and an in-depth understanding of climate modelling. The overall result will be a better understanding of the limitations of climate predictions, as well as the production of better climate projections to guide adaptation and mitigation.
What makes a good climate model? How can we make progress in quantifying the uncertainties in climate models, while limiting the environmental footprint of climate modelling? Is it possible to combine information drawn from observations of the current climate and numerical simulations of climate models to better guide adaptation and mitigation? The PC6 QUINTET project aims to answer these 3 questions, which are fundamental to the development of climate services, from a societal point of view.
36 participants + 7 planned recruitments.
The consortium
CNRS, Météo-France
- Centre National de Recherches Météorologiques (CNRM)
- Institut Pierre-Simon Laplace (IPSL) including LMD, LOCEAN, LSCE
