Cerfacs Enter the world of high performance ...

PhD Defense : Saloua PEATIER :”Exploring the uncertainty of climate feedbacks from a perturbed parameters ensemble of atmospheric simulations”

  Wednesday 30 November 2022 at 15h00

  Phd Thesis       Conference room, CERFACS, Toulouse    

You Tube link : https://youtu.be/jMltW3QmBM4

Abstract :

The uncertainty associated with the climate model calibration is rarely represented in climate feedback parameter estimates. Perturbed parameter ensembles (PPE) are ensembles of simulations for which the physical parameters have been perturbed according to a defined sampling. The main objective of this thesis is to explore the parametric dependence of climate (and mainly cloud) feedbacks of an atmospheric model. The methods developed and the results obtained are based on a PPE of the ARPEGE-Climat 6.3 model, atmospheric component of the CNRM-CM6-1 model. In the PPE simultaneously, 30 parameters of the model have been simultaneously perturbed, coming from the parametrizations of turbulence, convection, clouds microphysic and their radiative properties. The first chapter of the thesis illustrates the spatial trade-offs that can be made by model errors when minimizing an integrated metric, questioning the potential diversity of climate responses within the PPE. The second chapter addresses this very issue by proposing a quasi-automatic optimization method for selecting a sub-set of optimal calibrations with a diversity of feedback parameter values λ. Despite the approximation introduced by the use of statistical emulations, 15 simulations configured with different combinations of the 30 parameters values are identified, presenting a score comparable to that of the CFMIP6 multi-models, while covering an estimated ECS interval of [4.1 – 6.1 K]. These results illustrate the impact of model calibration on the ECS, highlighting the importance of better quantifying calibration uncertainty in climate projections. The thesis also explored the representations of clouds and their feedbacks in two different PPEs (the ARPEGE-Climate PPE and the HadGEM3-GA7 PPE). Several major differences were found between the two models, including the climate response of high clouds, leading to opposite SW and LW cloud feedbacks from one PPE to the other. Moreover, the high cloud feedbacks of average optical thickness (Hm clouds) appeared to be much larger in the CNRM-CM6-1 reference simulation than in the rest of the PPE. Parameters related to ice crystal microphysics greatly influence the variances of high cloud fractions in both PPEs. For the CNRM model, the inverse of the time scale characterizing the solid auto-conversion process efficiency is a parameter dominating both the variances of the Hm cloud climatology, and those of their tropical net feedbacks. The perspectives of this thesis include a PPE of coupled simulations with an oceanic model, which would allow a better quantification of the uncertainty related to the model calibration in climate projections.
Keywords :
climate model – uncertainties – tuning – climate sensitivity – perturbed parameter ensemble 
Jury :
Juliette MIGNOT – LOCEAN – – Referee
Frédéric HOURDIN – LMD  –  Referee
Fleur COUVREUX – CNRM – Examiner
Paul KUSHNER – University of Toronto – Examiner
Céline MARI – LA – Examiner
Laurent TERRAY –  CERFACS –  Director
Benjamin SANDERSON – CICERO  – Co-director