A hybrid machine learning-analog framework for downscaling unprecedented climate extremes
Organizations: Institute for Atmospheric and Climate Science, ETH Zurich · Seminar for Statistics, ETH Zurich · Swiss Data Science Center, EPFL and ETH Zurich
Abstract
Extreme weather events are critical for climate impact assessments. While physical models can simulate unprecedented extremes, for example, through ensemble boosting, this approach is computationally demanding, often restricting output to coarse spatial resolutions. Probabilistic generative downscaling offers a solution by enabling efficient sampling from the entire climate distribution, including its extreme tails. Here, we first demonstrate that machine learning generated regional climate model heatwaves over Paris are physically realistic. We then leverage an emulated machine learning dataset to perform analog-based downscaling of coarse global climate model ensemble boosting events. Our hybrid framework effectively bridges high-resolution regional detail with rare-event sampling, providing an efficient tool for extreme event downscaling and capturing multi-model uncertainty.
Figures & tables
| RCM / GCM | CNRM-CM5 | MIROC5 | MPI-ESM1-2-LR |
|---|---|---|---|
| ALADIN63 | model 1 | model 2 | |
| CCLM4-8-17 | model 3 | model 4 | model 5 |
| REMO2015 | model 6 | ||
| RegCM4-6 | model 7 | model 8 |
Appendix figures & tables11 assets
Supplementary material from the paper’s appendix.
Appendix
| CNRM-CM5 | MPI-ESM1-2-LR | MIROC5 | |
|---|---|---|---|
| RCM vs GCM | 0.8497 | 0.7604 | 0.8119 |
| Enscale Members vs GCM (Avg) | 0.7370 | 0.6096 | 0.7075 |
| Enscale Ensemble Mean vs GCM | 0.9339 | 0.8187 | 0.8902 |
| Precipitation Amount | Probability of Precip. | Solar Radiation | ||||
| Model | Slope 1 1 footnotemark: 1 | -value 2 2 footnotemark: 2 | Slope 3 3 footnotemark: 3 | -value 2 2 footnotemark: 2 | Slope 4 4 footnotemark: 4 | -value 2 2 footnotemark: 2 |
| ML Emulator | - | - | - | |||
| RCM | 0.00169 | 0.01798 | 0.61598 | |||
| GCM | 0.00167 | 0.19794 | 0.90798 | |||
| model 1 | model 2 | model 4 | model 6 | model 7 | model 8 | |
| Threshold Bin ( ) | ||||||
| [4.6, 4.8) | 3 | 3 | 3 | 16 | 8 | 10 |
| [4.8, 5.0) | 0 | 1 | 0 | 7 | 6 | 6 |
| [5.0, 5.2) | 2 | 2 | 0 | 1 | 4 | 2 |
| [5.2, 5.4) | 0 | 0 | 0 | 5 | 4 | 1 |
| [5.4, 5.6) | 1 | 0 | 0 | 2 | 1 | 1 |
| Decay Slope ( ) | -value | |
|---|---|---|
| EnScale | ||
| RCM | 0.601 | |
| GCM | 0.898 |
Climate Downscaling
Climate downscaling translates coarse climate simulations and observations into finer regional detail for local impact and risk assessment.
See all 32 papers