GCMagicc v1: a fast generative emulator for multivariate climate-impact ensembles
Abstract
Projecting the impacts of climate change requires large ensembles of climate variables that match historical observations, align with the warming ranges assessed by the IPCC, and can efficiently run new future emissions scenarios, including the newest generation of climate model scenarios (CMIP7) and pathways consistent with countries' Paris Agreement pledges. Generating such ensembles at the scale needed for impact studies is normally computationally prohibitive. We close this gap with GCMagicc, a hybrid model that pairs a simple physical climate model with machine learning to generate ensembles of 10 climate variables at the resolution of full-scale Earth system models, without relying on GPU resources or retraining for new scenarios. Trained on 32 CMIP6 Earth system models and observational/reanalysis data, GCMagicc complements rather than replaces Earth system models. We apply it to a range of future pathways: the canonical SSP scenarios of the latest IPCC report (1.2-6.1°C warming, min-max across scenarios of 5-95 percentile ranges), current policies (2.3-4.0°C), national pledges under the Paris Agreement (1.5-3.3°C) and the CMIP7 range from the 'VL' to 'H' scenarios (1.2-4.2°C), releasing a large public dataset. As an illustration, we perform an attribution analysis of the severe 2025 Iranian drought using GCMagicc ensembles, with three CMIP6 large ensembles for comparison, with and without anthropogenic forcings. The results suggest a strong anthropogenic signal: a median probability of drought at least as severe as observed of 29% with anthropogenic forcing, and zero under natural-forcing-only simulations. In the future, drought conditions are projected to materially worsen, amplifying the potential for agricultural and food security impacts and geopolitical conflicts that use water scarcity as a weapon. GCMagicc data is available at https://gcmagicc.org.
Related
- Generative modelling for multivariate spatio-temporal downscaling of monthly climate fields
- EnsAI: an emulator for atmospheric ammonia ensembles
- Kilometer-Scale AI Downscaling of Atlantic Hurricanes with Generative Ensembles
- On combining climate models into weighted ensembles
- EnScale: Temporally‐Consistent Multivariate Generative Downscaling via Proper Scoring Rules
BibTeX
@misc{meinshausen2026gcmagicc,
title = {{GCMagicc v1: a fast generative emulator for multivariate climate-impact ensembles}},
author = {Nicolai Meinshausen and Malte Meinshausen and Jared Lewis and Zebedee Nicholls and Sarah Schöngart and Alister Self and Xinwei Shen and Karla Spiller and Elisabeth Vogel},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.08383},
eprint = {2609.08383},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.08383},
}
RIS
TY - UNPB
TI - GCMagicc v1: a fast generative emulator for multivariate climate-impact ensembles
AU - Nicolai Meinshausen
AU - Malte Meinshausen
AU - Jared Lewis
AU - Zebedee Nicholls
AU - Sarah Schöngart
AU - Alister Self
AU - Xinwei Shen
AU - Karla Spiller
AU - Elisabeth Vogel
PY - 2026
DA - 2026/09/08
PB - arXiv
DO - 10.48550/arxiv.2609.08383
UR - https://arxiv.org/abs/2609.08383
AB - Projecting the impacts of climate change requires large ensembles of climate variables that match historical observations, align with the warming ranges assessed by the IPCC, and can efficiently run new future emissions scenarios, including the newest generation of climate model scenarios (CMIP7) and pathways consistent with countries' Paris Agreement pledges. Generating such ensembles at the scale needed for impact studies is normally computationally prohibitive. We close this gap with GCMagicc, a hybrid model that pairs a simple physical climate model with machine learning to generate ensembles of 10 climate variables at the resolution of full-scale Earth system models, without relying on GPU resources or retraining for new scenarios. Trained on 32 CMIP6 Earth system models and observational/reanalysis data, GCMagicc complements rather than replaces Earth system models. We apply it to a range of future pathways: the canonical SSP scenarios of the latest IPCC report (1.2-6.1°C warming, min-max across scenarios of 5-95 percentile ranges), current policies (2.3-4.0°C), national pledges under the Paris Agreement (1.5-3.3°C) and the CMIP7 range from the 'VL' to 'H' scenarios (1.2-4.2°C), releasing a large public dataset. As an illustration, we perform an attribution analysis of the severe 2025 Iranian drought using GCMagicc ensembles, with three CMIP6 large ensembles for comparison, with and without anthropogenic forcings. The results suggest a strong anthropogenic signal: a median probability of drought at least as severe as observed of 29% with anthropogenic forcing, and zero under natural-forcing-only simulations. In the future, drought conditions are projected to materially worsen, amplifying the potential for agricultural and food security impacts and geopolitical conflicts that use water scarcity as a weapon. GCMagicc data is available at https://gcmagicc.org.
ER -