Role of the Atmospheric Circulation in the Observed Warming Over Europe Using a Neural Network
Abstract
Quantifying the role of the atmospheric circulation in the temperature increase in Europe over the recent decades remains an open question. Here, we present an innovative dynamical adjustment framework based on a convolutional neural network (UNET), trained on CMIP6 simulations and fine‐tuned on reanalysis, to estimate the circulation‐induced temperature at the daily timescale and the subsequent trends over 1979–2024. Following this methodology, which provides high reconstruction scores at the daily timescale, we find temperature trends induced by the combination of horizontal winds at 850 hPa of 0.05 (−0.03, 0.14)°C/decade annually, 0.08 (−0.00, 0.17)°C/decade in summer and 0.09 (−0.11, 0.29)°C/decade in winter, accounting for between 10% and 20% of the total trends. In addition, we show that using sea‐level pressure rather than upper‐level winds would result in slightly higher trends, especially in summer.
Related
- Droughts are dominated by anomalous circulation not anomalous atmospheric moisture content
- Persistent influx of atmospheric microplastics and nanoplastics to Mount Everest driven by large scale circulation
- Ocean circulation changes rather than global warming drove temperature trends across the Miocene Climatic Optimum
- The representativeness of observed Δt in calculating formation and growth rates of newly formed atmospheric particles – hypothesis and examination
- Role of atmospheric circulation and anthropogenic forcing in the extreme 2025 European summer heatwave
BibTeX
@article{cariou2026role,
title = {{Role of the Atmospheric Circulation in the Observed Warming Over Europe Using a Neural Network}},
author = {Enora Cariou and Julien Cattiaux and Saïd Qasmi and Aurélien Ribes},
journal = {Geophysical Research Letters},
year = {2026},
month = oct,
doi = {10.1029/2026gl121861},
url = {https://hal.science/hal-05714155v1},
note = {CC BY-NC 4.0},
}
RIS
TY - JOUR
TI - Role of the Atmospheric Circulation in the Observed Warming Over Europe Using a Neural Network
AU - Enora Cariou
AU - Julien Cattiaux
AU - Saïd Qasmi
AU - Aurélien Ribes
PY - 2026
DA - 2026/10/16
JO - Geophysical Research Letters
DO - 10.1029/2026gl121861
UR - https://hal.science/hal-05714155v1
AB - Quantifying the role of the atmospheric circulation in the temperature increase in Europe over the recent decades remains an open question. Here, we present an innovative dynamical adjustment framework based on a convolutional neural network (UNET), trained on CMIP6 simulations and fine‐tuned on reanalysis, to estimate the circulation‐induced temperature at the daily timescale and the subsequent trends over 1979–2024. Following this methodology, which provides high reconstruction scores at the daily timescale, we find temperature trends induced by the combination of horizontal winds at 850 hPa of 0.05 (−0.03, 0.14)°C/decade annually, 0.08 (−0.00, 0.17)°C/decade in summer and 0.09 (−0.11, 0.29)°C/decade in winter, accounting for between 10% and 20% of the total trends. In addition, we show that using sea‐level pressure rather than upper‐level winds would result in slightly higher trends, especially in summer.
SN - 0094-8276
SN - 1944-8007
ER -