Private

Meryl WIMMER

Post-Doctoral Researcher on Atmospheric Modelling

PhD

Model Error representation in the convective-scale Ensemble Prediction System AROME-EPS

Abstract

Despite a continous improvement of numerical weather prediction models, some forecast busts still occur due to a presence of error in models. The representation of the different origins of model uncertainty is an important aspect, in particular in Ensemble Prediction Systems (EPS). The regional Ensemble Prediction System used at Météo-France, AROME-EPS, currently represents model uncertainties through the perturbation of global output tendencies of physical parameterization. However, this method presents some disadvantages such as a difficult physical interpretation of results. Thus, this PhD-thesis aims to study more physical model error representation methods, based on the perturbation of input parameters of the physical parameterization schemes.

Following advices of parameterization experts, 21 parameters to perturb, whose values are uncertain, have been selected. Sensitivity analyses using the Morris screening and Sobol' sensitivity indices, have led to reduce this list to eight parameters with a high impact on AROME forecasts. With the Morris method, the sensitivity of the AROME model to the parameters have been fully detailed according the seasons, days, hours and grid points. In parallel, the Sobol’ method mostly confirms Morris’ results and gives an insight of parameters interactions.

Several perturbed parameters techniques have then been set up and evaluated over long periods. They largely improve AROME-EPS performances for most near-surface variables including wind speed and accumulated precipitation. Different optimizations improving the statistical CRPS score have also been tested. Thus, a set of parameters have been identified for each AROME-EPS member. However, they induce a systematic bias of AROME-EPS members. Reducing the perturbation to the eight most influential parameters has shown similar results as the version perturbing the full set of parameters, suggesting a possible cheaper setting of weather prediction models. Finally, in a more pratical way, some study cases have shown the possibility to use these perturbed parameters methods in order to represent model uncertainty.

Keywords : Predictability ; Ensemble Prediction System ; Parameterization ; Mesoscale processes ; Sensitivity Analysis ; Model error ; Perturbed Parameters ;

Examination Committee

Dr. Jean-Pierre CHABOUREAU – LAERO (President)
Dr. Eric BLAYO – UGA (Rapporteur)
Dr. Clémentine PRIEUR – UGA (Rapporteur)
Dr. Stéphane VANNITSEM – IRM (Examinator)
Dr. Loik BERRE – CNRM/GMAP (Director)
Dr. Laurent DESCAMPS – CNRM/GMAP (Co-director)
Dr. Agnès LAGNOUX – UT2J (Invited)

Documents

Dissertation:
Defense announcement:
Defense's presentation:
Other documents

Research Articles

Wimmer, M. , Raynaud, L., Descamps, L., Berre, L. and Seity, Y.(2022) Sensitivity analysis of the convective-scale AROME model to physical and dynamical parameters. Quarterly Journal of the Royal Meteorological Society , 148, 743, 920– 942. Available from: doi : 10.1002/qj.4239

Supplementary:

Presentations (English)

EMS 2023

Wimmer, M. et al. : « Development and evaluation of an "optimal" perturbed parameter approach in the convective-scale AROME-EPS », EMS Annual Meeting 2023 , 04-08th September 2023, online and at Bratislava, Slovakia, doi: 10.5194/ems2023-310

Abstract:

Model Uncertainty Workshop

Wimmer, M. et al. : « Development and evaluation of an "optimal" perturbed parameter approach in the convective-scale AROME-EPS », Workshop on Model Uncertainty , ECMWF, 9-12th May 2022, online and at Reading, United-Kingdom

Abstract:

Accord All Staff Workshop

Raynaud, L. , Wimmer, M. et al. : « AROME-France EPS: current status and future directions », 2nd ACCORD All Staff Workshop , 4-8th April 2022, online and at Ljubljana, Slovenia
Raynaud, L. , Wimmer, M. et al. : « AROME-France EPS: current status and future directions », 1st ACCORD All Staff Workshop , 12-16th April 2021, online

HarmonEPS Working Week

Wimmer, M. et al. : « Model error representation in AROME-EPS », HarmonEPS Virtual Presentation Day , 23th November 2020, online
Wimmer, M. et al. : « Representation of model errors in AROME-EPS », invited to participate at the HarmonEPS Working Week , 11-15th November 2019, Barcelona, Spain

Presentations (French)

Wimmer, M. : « Représentation de l’erreur de modélisation dans le système de prévision d’ensemble régional PEARO », Soutenance de thèse de doctorat , Université Paul Sabatier Toulouse III, 7th December 2021, Toulouse, France, doi : 10.13140/RG.2.2.26926.23367

Abstract:
Wimmer, M. et al. : « Représentation des erreurs de modélisation dans le système de prévision d’ensemble régional PEARO », Session poster lors de la Journée des Doctorants du Centre National de Recherches Météorologiques , 3th December 2020, online
Poster:
Slides:
Wimmer, M. et al. : « Représentation des erreurs de modélisation dans le système de prévision d’ensemble régional PEARO », Session poster lors de la Journée des Doctorants du Centre National de Recherches Météorologiques , CNRM, 16th April 2019, Toulouse, France
Poster:
Slides:
Wimmer, M. et al. : « Représentation des erreurs de modélisation dans le système de prévision d’ensemble régional PEARO », Journée des Doctorants du Centre National de Recherches Météorologiques , 16th April 2019, Toulouse, France
Wimmer, M. et al. : « Représentation des erreurs de modélisation dans le système de prévision d’ensemble régional PEARO », Réunion Groupe de Travail Thématique Transverse (G3T) du Centre National de Recherches Météorologiques sur les méthodes ensemblistes en assimilation de données et en prévision , 5th April 2019, Toulouse, France
Raynaud, L. , Wimmer, M. et al. : « Représentation des incertitudes dans la prévision d'ensemble AROME », Ateliers de Modélisation de l’Atmosphère - Calibration de modèles et quantification des incertitudes , 11-13th March 2019, Toulouse, France