A machine-learning analysis of lagrangian low-cloud evolution in climate models
Une analyse par apprentissage automatique de l'évolution lagrangienne des nuages ??bas dans les modèles climatiques
Lewis, Hamish ; Bellon, Gilles
Année de publication
2025
It is well documented that upstream meteorological conditions are highly significant in determining the cloud state within subtropical marine boundary layers. These upstream conditions are both significant and systematic enough to be the dominant factors in the observed low-cloud coverage (LCC) climatology. At the time of writing there exists only one study which aims to understand the role of these upstream conditions in climate models. Here we assess the simulation of the impacts of the upstream conditions on the LCC climatology within 9 general circulation models (GCMs). We use machine-learning statistical models (random forests) applied to monthly fields to determine the influence of local and upstream large-scale conditions on GCM LCC in three regimes determined by the tercile of mean LCC: trade-cumulus, stratocumulus transition, and stratocumulus regimes. Only the effects of upstream thermodynamic stratification for stratocumulus regime are simulated by the GCMs, otherwise there is either a misrepresentation, or no relationship between GCM LCC and both local and upstream large-scale conditions. In particular, the GCMs do not reproduce the observed upstream influence of SST. There are significant differences between models, but generally we find that GCM LCC variability is too strongly and too exclusively dependent on the upstream stratification compared to observations.</div>
Texte intégral
Accès à la notice sur le site du portail documentaire de Météo-France