Selecting and weighting dynamical models using data-driven approaches

Autores
Le Bras, Pierre; Sévellec, Florian; Tandeo, Pierre; Ruiz, Juan Jose; Ailliot, Pierre
Año de publicación
2024
Idioma
inglés
Tipo de recurso
artículo
Estado
versión publicada
Descripción
In geosciences, multi-model ensembles are helpful to explore the robustness of a range of results. Toobtain a synthetic and improved representation of the studied dynamic system, the models are usually weighted.The simplest method, namely the model democracy, gives equal weights to all models, while more advancedapproaches base weights on agreement with available observations. Here, we focus on determining weights forvarious versions of an idealized model of the Atlantic Meridional Overturning Circulation. This is done by assessing their performance against synthetic observations (generated from one of the model versions) within a dataassimilation framework using the ensemble Kalman filter (EnKF). In contrast to traditional data assimilation, weimplement data-driven forecasts using the analog method based on catalogs of short-term trajectories. This approach allows us to efficiently emulate the model’s dynamics while keeping computational costs low. For eachmodel version, we compute a local performance metric, known as the contextual model evidence, to compareobservations and model forecasts. This metric, based on the innovation likelihood, is sensitive to differences inmodel dynamics and considers forecast and observation uncertainties. Finally, the weights are calculated usingboth model performance and model co-dependency and then evaluated on averages of long-term simulations.Results show good performance in identifying numerical simulations that best replicate observed short-termvariations. Additionally, it outperforms benchmark approaches such as strategies based on model democracy orclimatology when reconstructing missing distributions. These findings encourage the application of the proposedmethodology to more complex datasets in the future, like climate simulations.
Fil: Le Bras, Pierre. Imt Atlantique Bretagne Pays de la Loire.; Francia
Fil: Sévellec, Florian. Imt Atlantique Bretagne Pays de la Loire.; Francia
Fil: Tandeo, Pierre. Imt Atlantique Bretagne Pays de la Loire.; Francia
Fil: Ruiz, Juan Jose. Consejo Nacional de Investigaciones Científicas y Técnicas. Oficina de Coordinación Administrativa Ciudad Universitaria. Centro de Investigaciones del Mar y la Atmósfera. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Centro de Investigaciones del Mar y la Atmósfera; Argentina. Instituto Franco-Argentino sobre Estudios del Clima y sus Impactos; Argentina. Universite de Bretagne Occidentale; Francia
Fil: Ailliot, Pierre. Imt Atlantique Bretagne Pays de la Loire.; Francia
Materia
Model selection
Data assimilation
Machine learning
Nivel de accesibilidad
acceso abierto
Condiciones de uso
https://creativecommons.org/licenses/by/2.5/ar/
Repositorio
CONICET Digital (CONICET)
Institución
Consejo Nacional de Investigaciones Científicas y Técnicas
OAI Identificador
oai:ri.conicet.gov.ar:11336/261257

id CONICETDig_49900b0a294593de1d3c22269dddc9a3
oai_identifier_str oai:ri.conicet.gov.ar:11336/261257
network_acronym_str CONICETDig
repository_id_str 3498
network_name_str CONICET Digital (CONICET)
spelling Selecting and weighting dynamical models using data-driven approachesLe Bras, PierreSévellec, FlorianTandeo, PierreRuiz, Juan JoseAilliot, PierreModel selectionData assimilationMachine learninghttps://purl.org/becyt/ford/1.5https://purl.org/becyt/ford/1In geosciences, multi-model ensembles are helpful to explore the robustness of a range of results. Toobtain a synthetic and improved representation of the studied dynamic system, the models are usually weighted.The simplest method, namely the model democracy, gives equal weights to all models, while more advancedapproaches base weights on agreement with available observations. Here, we focus on determining weights forvarious versions of an idealized model of the Atlantic Meridional Overturning Circulation. This is done by assessing their performance against synthetic observations (generated from one of the model versions) within a dataassimilation framework using the ensemble Kalman filter (EnKF). In contrast to traditional data assimilation, weimplement data-driven forecasts using the analog method based on catalogs of short-term trajectories. This approach allows us to efficiently emulate the model’s dynamics while keeping computational costs low. For eachmodel version, we compute a local performance metric, known as the contextual model evidence, to compareobservations and model forecasts. This metric, based on the innovation likelihood, is sensitive to differences inmodel dynamics and considers forecast and observation uncertainties. Finally, the weights are calculated usingboth model performance and model co-dependency and then evaluated on averages of long-term simulations.Results show good performance in identifying numerical simulations that best replicate observed short-termvariations. Additionally, it outperforms benchmark approaches such as strategies based on model democracy orclimatology when reconstructing missing distributions. These findings encourage the application of the proposedmethodology to more complex datasets in the future, like climate simulations.Fil: Le Bras, Pierre. Imt Atlantique Bretagne Pays de la Loire.; FranciaFil: Sévellec, Florian. Imt Atlantique Bretagne Pays de la Loire.; FranciaFil: Tandeo, Pierre. Imt Atlantique Bretagne Pays de la Loire.; FranciaFil: Ruiz, Juan Jose. Consejo Nacional de Investigaciones Científicas y Técnicas. Oficina de Coordinación Administrativa Ciudad Universitaria. Centro de Investigaciones del Mar y la Atmósfera. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Centro de Investigaciones del Mar y la Atmósfera; Argentina. Instituto Franco-Argentino sobre Estudios del Clima y sus Impactos; Argentina. Universite de Bretagne Occidentale; FranciaFil: Ailliot, Pierre. Imt Atlantique Bretagne Pays de la Loire.; FranciaEuropean Geosciences Union2024-07info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionhttp://purl.org/coar/resource_type/c_6501info:ar-repo/semantics/articuloapplication/pdfapplication/pdfhttp://hdl.handle.net/11336/261257Le Bras, Pierre; Sévellec, Florian; Tandeo, Pierre; Ruiz, Juan Jose; Ailliot, Pierre; Selecting and weighting dynamical models using data-driven approaches; European Geosciences Union; Nonlinear Processes in Geophysics; 31; 3; 7-2024; 303-3171607-7946CONICET DigitalCONICETenginfo:eu-repo/semantics/altIdentifier/url/https://npg.copernicus.org/articles/31/303/2024/info:eu-repo/semantics/altIdentifier/doi/10.5194/npg-31-303-2024info:eu-repo/semantics/openAccesshttps://creativecommons.org/licenses/by/2.5/ar/reponame:CONICET Digital (CONICET)instname:Consejo Nacional de Investigaciones Científicas y Técnicas2025-09-29T09:44:28Zoai:ri.conicet.gov.ar:11336/261257instacron:CONICETInstitucionalhttp://ri.conicet.gov.ar/Organismo científico-tecnológicoNo correspondehttp://ri.conicet.gov.ar/oai/requestdasensio@conicet.gov.ar; lcarlino@conicet.gov.arArgentinaNo correspondeNo correspondeNo correspondeopendoar:34982025-09-29 09:44:28.845CONICET Digital (CONICET) - Consejo Nacional de Investigaciones Científicas y Técnicasfalse
dc.title.none.fl_str_mv Selecting and weighting dynamical models using data-driven approaches
title Selecting and weighting dynamical models using data-driven approaches
spellingShingle Selecting and weighting dynamical models using data-driven approaches
Le Bras, Pierre
Model selection
Data assimilation
Machine learning
title_short Selecting and weighting dynamical models using data-driven approaches
title_full Selecting and weighting dynamical models using data-driven approaches
title_fullStr Selecting and weighting dynamical models using data-driven approaches
title_full_unstemmed Selecting and weighting dynamical models using data-driven approaches
title_sort Selecting and weighting dynamical models using data-driven approaches
dc.creator.none.fl_str_mv Le Bras, Pierre
Sévellec, Florian
Tandeo, Pierre
Ruiz, Juan Jose
Ailliot, Pierre
author Le Bras, Pierre
author_facet Le Bras, Pierre
Sévellec, Florian
Tandeo, Pierre
Ruiz, Juan Jose
Ailliot, Pierre
author_role author
author2 Sévellec, Florian
Tandeo, Pierre
Ruiz, Juan Jose
Ailliot, Pierre
author2_role author
author
author
author
dc.subject.none.fl_str_mv Model selection
Data assimilation
Machine learning
topic Model selection
Data assimilation
Machine learning
purl_subject.fl_str_mv https://purl.org/becyt/ford/1.5
https://purl.org/becyt/ford/1
dc.description.none.fl_txt_mv In geosciences, multi-model ensembles are helpful to explore the robustness of a range of results. Toobtain a synthetic and improved representation of the studied dynamic system, the models are usually weighted.The simplest method, namely the model democracy, gives equal weights to all models, while more advancedapproaches base weights on agreement with available observations. Here, we focus on determining weights forvarious versions of an idealized model of the Atlantic Meridional Overturning Circulation. This is done by assessing their performance against synthetic observations (generated from one of the model versions) within a dataassimilation framework using the ensemble Kalman filter (EnKF). In contrast to traditional data assimilation, weimplement data-driven forecasts using the analog method based on catalogs of short-term trajectories. This approach allows us to efficiently emulate the model’s dynamics while keeping computational costs low. For eachmodel version, we compute a local performance metric, known as the contextual model evidence, to compareobservations and model forecasts. This metric, based on the innovation likelihood, is sensitive to differences inmodel dynamics and considers forecast and observation uncertainties. Finally, the weights are calculated usingboth model performance and model co-dependency and then evaluated on averages of long-term simulations.Results show good performance in identifying numerical simulations that best replicate observed short-termvariations. Additionally, it outperforms benchmark approaches such as strategies based on model democracy orclimatology when reconstructing missing distributions. These findings encourage the application of the proposedmethodology to more complex datasets in the future, like climate simulations.
Fil: Le Bras, Pierre. Imt Atlantique Bretagne Pays de la Loire.; Francia
Fil: Sévellec, Florian. Imt Atlantique Bretagne Pays de la Loire.; Francia
Fil: Tandeo, Pierre. Imt Atlantique Bretagne Pays de la Loire.; Francia
Fil: Ruiz, Juan Jose. Consejo Nacional de Investigaciones Científicas y Técnicas. Oficina de Coordinación Administrativa Ciudad Universitaria. Centro de Investigaciones del Mar y la Atmósfera. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Centro de Investigaciones del Mar y la Atmósfera; Argentina. Instituto Franco-Argentino sobre Estudios del Clima y sus Impactos; Argentina. Universite de Bretagne Occidentale; Francia
Fil: Ailliot, Pierre. Imt Atlantique Bretagne Pays de la Loire.; Francia
description In geosciences, multi-model ensembles are helpful to explore the robustness of a range of results. Toobtain a synthetic and improved representation of the studied dynamic system, the models are usually weighted.The simplest method, namely the model democracy, gives equal weights to all models, while more advancedapproaches base weights on agreement with available observations. Here, we focus on determining weights forvarious versions of an idealized model of the Atlantic Meridional Overturning Circulation. This is done by assessing their performance against synthetic observations (generated from one of the model versions) within a dataassimilation framework using the ensemble Kalman filter (EnKF). In contrast to traditional data assimilation, weimplement data-driven forecasts using the analog method based on catalogs of short-term trajectories. This approach allows us to efficiently emulate the model’s dynamics while keeping computational costs low. For eachmodel version, we compute a local performance metric, known as the contextual model evidence, to compareobservations and model forecasts. This metric, based on the innovation likelihood, is sensitive to differences inmodel dynamics and considers forecast and observation uncertainties. Finally, the weights are calculated usingboth model performance and model co-dependency and then evaluated on averages of long-term simulations.Results show good performance in identifying numerical simulations that best replicate observed short-termvariations. Additionally, it outperforms benchmark approaches such as strategies based on model democracy orclimatology when reconstructing missing distributions. These findings encourage the application of the proposedmethodology to more complex datasets in the future, like climate simulations.
publishDate 2024
dc.date.none.fl_str_mv 2024-07
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
http://purl.org/coar/resource_type/c_6501
info:ar-repo/semantics/articulo
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/11336/261257
Le Bras, Pierre; Sévellec, Florian; Tandeo, Pierre; Ruiz, Juan Jose; Ailliot, Pierre; Selecting and weighting dynamical models using data-driven approaches; European Geosciences Union; Nonlinear Processes in Geophysics; 31; 3; 7-2024; 303-317
1607-7946
CONICET Digital
CONICET
url http://hdl.handle.net/11336/261257
identifier_str_mv Le Bras, Pierre; Sévellec, Florian; Tandeo, Pierre; Ruiz, Juan Jose; Ailliot, Pierre; Selecting and weighting dynamical models using data-driven approaches; European Geosciences Union; Nonlinear Processes in Geophysics; 31; 3; 7-2024; 303-317
1607-7946
CONICET Digital
CONICET
dc.language.none.fl_str_mv eng
language eng
dc.relation.none.fl_str_mv info:eu-repo/semantics/altIdentifier/url/https://npg.copernicus.org/articles/31/303/2024/
info:eu-repo/semantics/altIdentifier/doi/10.5194/npg-31-303-2024
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
https://creativecommons.org/licenses/by/2.5/ar/
eu_rights_str_mv openAccess
rights_invalid_str_mv https://creativecommons.org/licenses/by/2.5/ar/
dc.format.none.fl_str_mv application/pdf
application/pdf
dc.publisher.none.fl_str_mv European Geosciences Union
publisher.none.fl_str_mv European Geosciences Union
dc.source.none.fl_str_mv reponame:CONICET Digital (CONICET)
instname:Consejo Nacional de Investigaciones Científicas y Técnicas
reponame_str CONICET Digital (CONICET)
collection CONICET Digital (CONICET)
instname_str Consejo Nacional de Investigaciones Científicas y Técnicas
repository.name.fl_str_mv CONICET Digital (CONICET) - Consejo Nacional de Investigaciones Científicas y Técnicas
repository.mail.fl_str_mv dasensio@conicet.gov.ar; lcarlino@conicet.gov.ar
_version_ 1844613399759028224
score 13.070432