Powered by OpenAIRE graph
Found an issue? Give us feedback
addClaim

Deep recurrent learned dynamic downscaling

Authors: Djamen-Kepaou, Jean-Yves;

Deep recurrent learned dynamic downscaling

Abstract

Les modèles climatiques mondiaux reproduisent les principaux composants du système climatique de la planète afin de générer des réalisations à long terme, éparses et précises, d'événements climatiques futurs sur l'ensemble du globe. La réduction d'échelle est la méthode par laquelle ces simulations à basse résolution sont converties en simulations à haute résolution d'événements climatiques qui peuvent ensuite être utilisées par les parties prenantes et les décideurs. Les modèles climatiques régionaux réduisent dynamiquement l'échelle du climat simulé en conditionnant les modèles climatiques mondiaux à des processus physiques spécifiques à un lieu. Bien que ces modèles soient robustes et fiables, ils sont coûteux en termes de calcul par rapport aux approches statistiques permettant de modéliser une relation générale entre le comportement du climat mondial et le comportement du climat local. Par conséquent, il existe un besoin pour des méthodes de réduction d'échelle qui tirent parti de l'efficacité de calcul des modèles statistiques tout en maintenant la performance des modèles climatiques régionaux.Dans cette thèse, nous nous appuyons sur les méthodes d'apprentissage profond proposées précédemment pour la réduction d'échelle dynamique par l'estimation d'un modèle climatique régional. Le modèle que nous proposons est un réseau antagoniste génératif qui exploite les effets des dépendances temporelles dans les événements climatiques spatio-temporels

Global climate models represent major climate system components of the planet in order to generate long term, sparse, accurate realizations of future climatic events across the entire globe. Downscaling is the method by which these low resolution realizations are converted into high resolution simulations of climate events which can then be used by stakeholders and policy makers. Regional climate models dynamically downscale simulated climate by conditioning global climate models on location-specific physical processes. Although these models are robust and reliable, they are computationally expensive when compared to statistical approaches for modeling a general relationship between global climate behaviour and local climate behavior. Therefore, there is need for downscaling methods that leverage the computational efficiency of statistical models while maintaining the performance of regional climate models.In this thesis, we build upon previously proposed deep learning methods for dynamical downscaling through estimation of a regional climate model. Our proposed model is a generative adversarial network that leverages the effects of temporal dependencies within spatio-temporal climate events

Neslehova, Johanna (Supervisor)

Country
Canada
Related Organizations
Keywords

Mathematics and Statistics

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average
Upload OA version
Are you the author of this publication? Upload your Open Access version to Zenodo!
It’s fast and easy, just two clicks!