Towards accurate extreme event likelihoods from diffusion model climate emulators
Abstract
ML climate model emulators are useful for scenario planning and adaptation, allowing for cost-efficient experimentation. Recently, the diffusion model Climate in a Bottle (cBottle) has been proposed for generation of atmospheric states compatible with boundary conditions of solar position and sea surface temperatures. Crucially, cBottle can be guided to generate extreme events such as Tropical Cyclones (TCs) over locations of interest. Diffusion models such as cBottle work by approximating the probability density of the training data. Here, we show use cases of the probability density estimates of atmospheric states obtained from this climate emulator. Most importantly, these estimates allow us to calculate likelihoods of extreme events under guidance. When guiding the model towards states including TCs, comparing the probability density under the guided and unguided model enables us to quantify how much more likely the guidance has made the TC. We show how these odds ratios allow us to importance-sample from the TC distribution, reducing the standard error of the probability estimate compared to simple Monte Carlo sampling. Furthermore, we discuss results and limitations of the application of model probability densities to extreme event attribution-like experiments. We present these early but encouraging results hoping they will spur more research into probabilistic information that can be gained from diffusion models of the atmosphere.
Related
- SamudrACE: Fast and Accurate Coupled Climate Modeling With 3D Ocean and Atmosphere Emulators
- Climate warming amplifies coastal hypoxia response to extreme river discharge
- On the Inherent Ambiguity of Extreme Event Costs Attributed to Climate Change
- Evidence from 2024 Storm Boris indicates anthropogenic climate change amplifies floods more than extreme precipitation
- Passive retrofit strategies for climate-resilient high-rise social housing: evidence from sydney under current and future extreme climate conditions
BibTeX
@misc{manshausen2026towards,
title = {{Towards accurate extreme event likelihoods from diffusion model climate emulators}},
author = {Peter Manshausen and Noah Brenowitz and Julius Berner and Karthik Kashinath and Mike Pritchard},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2605.03802},
eprint = {2605.03802},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.03802},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Towards accurate extreme event likelihoods from diffusion model climate emulators
AU - Peter Manshausen
AU - Noah Brenowitz
AU - Julius Berner
AU - Karthik Kashinath
AU - Mike Pritchard
PY - 2026
DA - 2026/10/12
PB - arXiv
DO - 10.48550/arxiv.2605.03802
UR - https://arxiv.org/abs/2605.03802
AB - ML climate model emulators are useful for scenario planning and adaptation, allowing for cost-efficient experimentation. Recently, the diffusion model Climate in a Bottle (cBottle) has been proposed for generation of atmospheric states compatible with boundary conditions of solar position and sea surface temperatures. Crucially, cBottle can be guided to generate extreme events such as Tropical Cyclones (TCs) over locations of interest. Diffusion models such as cBottle work by approximating the probability density of the training data. Here, we show use cases of the probability density estimates of atmospheric states obtained from this climate emulator. Most importantly, these estimates allow us to calculate likelihoods of extreme events under guidance. When guiding the model towards states including TCs, comparing the probability density under the guided and unguided model enables us to quantify how much more likely the guidance has made the TC. We show how these odds ratios allow us to importance-sample from the TC distribution, reducing the standard error of the probability estimate compared to simple Monte Carlo sampling. Furthermore, we discuss results and limitations of the application of model probability densities to extreme event attribution-like experiments. We present these early but encouraging results hoping they will spur more research into probabilistic information that can be gained from diffusion models of the atmosphere.
ER -