Ensemble Methods for Neural Network‐Based Weather Forecasts
Problem
Deterministic neural network weather forecasts provide no measure of uncertainty, and generating ensembles with a good spread-error relationship is non-trivial.
Why it matters
Uncertainty estimates are essential for making reliable decisions from weather forecasts.
What the AI did
Tested four methods (random initial perturbations, network retraining, random dropout, singular-vector-decomposition perturbations) to convert a deterministic neural network forecast system into an ensemble system.
Results
Ensemble mean forecasts from all four approaches beat unperturbed neural network forecasts; retraining gave the highest improvement (no numeric values stated in abstract).
Impact
Demonstrated that ensemble techniques can add skill to neural network weather forecasts.
Record
| Venue | Journal of Advances in Modeling Earth Systems |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Sebastian Scher, Gabriele Messori |
| Citations | 87 (OpenAlex, at import time) |
| References | 40 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1029/2020ms002331 |
| Open access | gold |
| Keywords | ensemble forecasting, neural networks, weather prediction, uncertainty quantification, singular vector decomposition |
Importance rating
significant — thousands to millions affected
Uncertainty estimates are essential for reliable weather-forecast decisions affecting millions; this work is a methods demonstration, with NN skill still below NWP models.
More in Climate & Environment
| # | Paper | Year | Citations | Importance |
|---|---|---|---|---|
| 1 | Learning skillful medium-range global weather forecasting | 2023 | 1,416 |
8
|
| 2 | Accurate medium-range global weather forecasting with 3D neural networks | 2023 | 1,631 |
8
|
| 3 | FuXi: a cascade machine learning forecasting system for 15-day global weather forecast | 2023 | 355 |
8
|
| 4 | Skilful nowcasting of extreme precipitation with NowcastNet | 2023 | 451 |
8
|
| 5 | Forecasting Renewable Energy Generation with Machine Learning and Deep Learning: Current Advances and Future Prospects | 2023 | 378 |
8
|