Downscaling with AI reveals the large role of internal variability in fine-scale projections of climate extremes
5 affiliations
- 1ARC Centre of Excellence for Weather of the 21st Century, University of New South Wales, Sydney, Australia
- 2Climate Change Research Centre, University of New South Wales, Sydney, Australia
- 3Earth Sciences New Zealand, Auckland, New Zealand
- 4Earth Sciences New Zealand, Wellington, New Zealand
- 5Te Aka Mātuatua School of Science, Hamilton, New Zealand
Abstract
The computational cost of dynamical downscaling limits ensemble sizes in regional downscaling efforts. We present a generative-AI approach to greatly expand the scope of such downscaling, enabling fine-scale future changes to be characterised including rare extremes that cannot be addressed by traditional approaches. We test this approach for New Zealand, where heterogenous regional climate effects are anticipated. End-of-century projected daily precipitation extremes become progressively more responsive to climate change as rarer events are considered, on average increasing at 5.6%∘ C−1 for annual extremes to 9%∘ C−1 for 100-year extremes. The internal variability of precipitation extremes increases with warming much faster than the mean and nearly twice that of Clausius-Clapeyron scaling, dominating other sources of uncertainty. This precipitation “tail stretching" phenomenon for internal variability is accompanied by similar stretching of the extreme 850 hPa specific humidity distribution. Overall, fine-scale precipitation changes are less predictable than widely assumed and require substantially larger ensembles for reliable assessment than previously recognised.
Related
- A dynamical downscaling approach for the improved representation of sub-annual coastal sea level variability: the importance of shelf and slope ocean dynamics
- Assessing the potential of Individual Foraminifera Analyses ( δ ¹⁸ O) to reconstruct variability, seasonality and extremes in the tropical Indian Ocean
- Disentangling Internal and Forced Climate Variability with Convolutional Neural Networks using Multivariate Fields
- A 256-year daily rainfall record reveals multiscale rainy-season variability and emerging hydroclimatic hazard over northern East Asia
- Decadal variability reveals contrasts in solar-resource stability across China
BibTeX
@article{rampal2026downscaling,
title = {{Downscaling with AI reveals the large role of internal variability in fine-scale projections of climate extremes}},
author = {Neelesh Rampal and Peter B. Gibson and Steven C. Sherwood and Laura E. Queen and Hamish Lewis and Gab Abramowitz},
journal = {Climate Dynamics},
year = {2026},
month = jul,
doi = {10.1007/s00382-026-08269-y},
url = {https://doi.org/10.1007/s00382-026-08269-y},
note = {CC BY},
}
RIS
TY - JOUR
TI - Downscaling with AI reveals the large role of internal variability in fine-scale projections of climate extremes
AU - Neelesh Rampal
AU - Peter B. Gibson
AU - Steven C. Sherwood
AU - Laura E. Queen
AU - Hamish Lewis
AU - Gab Abramowitz
PY - 2026
DA - 2026/07/23
JO - Climate Dynamics
DO - 10.1007/s00382-026-08269-y
UR - https://doi.org/10.1007/s00382-026-08269-y
AB - The computational cost of dynamical downscaling limits ensemble sizes in regional downscaling efforts. We present a generative-AI approach to greatly expand the scope of such downscaling, enabling fine-scale future changes to be characterised including rare extremes that cannot be addressed by traditional approaches. We test this approach for New Zealand, where heterogenous regional climate effects are anticipated. End-of-century projected daily precipitation extremes become progressively more responsive to climate change as rarer events are considered, on average increasing at 5.6%^∘ C^−1 for annual extremes to 9%^∘ C^−1 for 100-year extremes. The internal variability of precipitation extremes increases with warming much faster than the mean and nearly twice that of Clausius-Clapeyron scaling, dominating other sources of uncertainty. This precipitation “tail stretching" phenomenon for internal variability is accompanied by similar stretching of the extreme 850 hPa specific humidity distribution. Overall, fine-scale precipitation changes are less predictable than widely assumed and require substantially larger ensembles for reliable assessment than previously recognised.
SN - 0930-7575
SN - 1432-0894
ER -