Machine Learning Approaches for Predicting Extreme Weather Event Frequency
DOI:
https://doi.org/10.5281/zenodo.22815791Keywords:
machine learning, extreme weather prediction, random forest, gradient boosting, convolutional long short-term memory network, heatwaves, tropical cyclone genesis, ERA5 reanalysis, skill score, extreme precipitatioAbstract
Skilful sub-seasonal prediction of extreme weather event frequency remains a persistent challenge for operational
forecasting, with dynamical models often underperforming for localized or rapidly evolving extremes. Here we
evaluate three machine learning architectures, random forest (RF), gradient-boosted decision trees (XGBoost), and
a hybrid convolutional-LSTM (ConvLSTM) network, against climatological and persistence baselines for predicting
the weekly frequency of three extreme event types: heatwaves over western Europe, extreme daily precipitation
over the eastern United States, and tropical cyclone genesis over the North Atlantic basin. Models were trained on
1979-2015 ERA5 reanalysis predictor fields, including 500-hPa geopotential height, sea surface temperature
anomalies, integrated water vapour, soil moisture, and an atmospheric blocking index, and tested out-of-sample on
2016-2023 with a strict temporal train/test split to avoid leakage. Skill was assessed using the Brier skill score (BSS)
relative to climatology, area under the receiver operating characteristic curve (AUC), and root-mean-square error
(RMSE) of predicted weekly event counts. The ConvLSTM achieved the highest skill for heatwave frequency (BSS
= 0.34, AUC = 0.86), exploiting spatial persistence of blocking-associated circulation anomalies, while XGBoost
achieved the highest skill for extreme precipitation (BSS = 0.24, AUC = 0.81), for which localized convective triggers
limited gains from spatial context. All three ML models outperformed climatology and persistence for every event
type and every lead time tested (1 to 14 days), though skill for tropical cyclone genesis was considerably lower
across all models (best BSS = 0.13) and declined sharply beyond a 7-day lead time. Permutation feature
importance identified sea surface temperature anomalies and the blocking index as the dominant predictors for
heatwaves, and integrated water vapour and soil moisture for precipitation extremes. These results indicate that
machine learning approaches, and particularly architectures that preserve spatial structure, offer measurable and
event-type-dependent skill gains over conventional baselines for short- to medium-range extreme weather
frequency prediction.
References
Boukabara SA, Krasnopolsky V, Stewart JQ, Maddy ES, Shahroudi N, Hoffman RN (2019) Outlook for exploiting artificial intelligence in the Earth and environmental sciences. Bulletin of the American Meteorological Society 100(12):2473-2490. https://doi.org/10.1175/BAMS-D-18-0031.1
McGovern A, Elmore KL, Gagne DJ, Haupt SE, Karstens CD, Lagerquist R, Smith T, Williams JK (2017) Using artificial intelligence to improve real-time decision-making for high-impact weather. Bulletin of the American Meteorological Society 98(10):2073-2090. https://doi.org/10.1175/BAMS-D-16-0123.1
Chantry M, Christensen H, Dueben P, Palmer T (2021) Opportunities and challenges for machine learning in weather and climate modelling: hard, medium and soft AI. Philosophical Transactions of the Royal Society A 379(2194):20200083. https://doi.org/10.1098/rsta.2020.0083
Bi K, Xie L, Zhang H, Chen X, Gu X, Tian Q (2023) Accurate medium-range global weather forecasting with 3D neural networks. Nature 619(7970):533-538. https://doi.org/10.1038/s41586-023-06185-3
Lam R, Sanchez-Gonzalez A, Willson M, Wirnsberger P, Fortunato M, Alet F, Ravuri S, Ewalds T, Eaton-Rosen Z, Hu W, Merose A, Hoyer S, Holland G, Vinyals O, Stott J, Pritzel A, Mohamed S, Battaglia P (2023) Learning skillful medium-range global weather forecasting. Science 382(6677):1416-1421. https://doi.org/10.1126/science.adi2336
Ravuri S, Lenc K, Willson M, Kangin D, Lam R, Mirowski P, Fitzsimons M, Athanassiadou M, Kashem S, Madge S, Prudden R, Mandhane A, Clark A, Brock A, Simonyan K, Hadsell R, Robinson N, Clancy E, Arribas A, Mohamed S (2021) Skilful precipitation nowcasting using deep generative models of radar. Nature 597(7878):672-677. https://doi.org/10.1038/s41586-021-03854-z
Espeholt L, Agrawal S, Sønderby C, Kumar M, Heek J, Bromberg C, Gazen C, Carver R, Andrychowicz M, Hickey J, Bell A, Kalchbrenner N (2022) Deep learning for twelve hour precipitation forecasts. Nature Communications 13:5145. https://doi.org/10.1038/s41467-022-32483-x
Ham YG, Kim JH, Luo JJ (2019) Deep learning for multi-year ENSO forecasts. Nature 573(7775):568-572. https://doi.org/10.1038/s41586-019-1559-7
Prabhat, Kashinath K, Mudigonda M, Kim S, Kapp-Schwoerer L, Graubner A, Karaismailoglu E, von Kleist L, Kurth T, Greiner A, Mahesh A, Yang K, Lewis C, Chen J, Lou A, Chandran S, Toms B, Chapman W, Dagon K, Shields CA, O Brien T, Wehner M, Collins W (2021) ClimateNet: an expert-labeled open dataset and deep learning architecture for enabling high-precision analyses of extreme weather. Geoscientific Model Development 14(1):107-124. https://doi.org/10.5194/gmd-14-107-2021
Chattopadhyay A, Hassanzadeh P, Pasha S (2020) Predicting clustered weather patterns: a test case for applications of convolutional neural networks to spatio-temporal climate data. Scientific Reports 10:1317. https://doi.org/10.1038/s41598-020-57897-9
Gagne DJ, Haupt SE, Nychka DW, Thompson G (2019) Interpretable deep learning for spatial analysis of severe hailstorms. Monthly Weather Review 147(8):2827-2845. https://doi.org/10.1175/MWR-D-18-0316.1
Herman GR, Schumacher RS (2018) Money doesn t grow on trees, but forecasts do: forecasting extreme precipitation with random forests. Monthly Weather Review 146(5):1571-1600. https://doi.org/10.1175/MWR-D-17-0250.1
Schultz MG, Betancourt C, Gong B, Kleinert F, Langguth M, Leufen LH, Mozaffari A, Stadtler S (2021) Can deep learning beat numerical weather prediction? Philosophical Transactions of the Royal Society A 379(2194):20200097. https://doi.org/10.1098/rsta.2020.0097
Watson-Parris D (2021) Machine learning for weather and climate are worlds apart. Philosophical Transactions of the Royal Society A 379(2194):20200098. https://doi.org/10.1098/rsta.2020.0098
Toms BA, Barnes EA, Ebert-Uphoff I (2020) Physically interpretable neural networks for the geosciences: applications to Earth system variability. Journal of Advances in Modeling Earth Systems 12(9):e2019MS002002. https://doi.org/10.1029/2019MS002002
Barnes EA, Hurrell JW, Ebert-Uphoff I, Anderson C, Anderson D (2019) Viewing forced climate patterns through an AI lens. Geophysical Research Letters 46(22):13389-13398. https://doi.org/10.1029/2019GL084944
Giffard-Roisin S, Yang M, Charpiat G, Kumler Bonfanti C, Kegl B, Monteleoni C (2020) Tropical cyclone track forecasting using fused deep learning from aligned reanalysis data. Frontiers in Big Data 3:1. https://doi.org/10.3389/fdata.2020.00001
Reichstein M, Camps-Valls G, Stevens B, Jung M, Denzler J, Carvalhais N, Prabhat (2019) Deep learning and process understanding for data-driven Earth system science. Nature 566(7743):195-204. https://doi.org/10.1038/s41586-019-0912-1
Rasp S, Pritchard MS, Gentine P (2018) Deep learning to represent subgrid processes in climate models. Proceedings of the National Academy of Sciences 115(39):9684-9689. https://doi.org/10.1073/pnas.1810286115
Rasp S, Dueben PD, Scher S, Weyn JA, Mouatadid S, Thuerey N (2020) WeatherBench: a benchmark dataset for data-driven weather forecasting. Journal of Advances in Modeling Earth Systems 12(11):e2020MS002203. https://doi.org/10.1029/2020MS002203
Weyn JA, Durran DR, Caruana R (2019) Can machines learn to predict weather? Using deep learning to predict gridded 500-hPa geopotential height from historical weather data. Journal of Advances in Modeling Earth Systems 11(8):2680-2693. https://doi.org/10.1029/2019MS001705
Kurth T, Treichler S, Romero J, Mudigonda M, Luehr N, Phillips E, Mahesh A, Matheson M, Deslippe J, Fatica M, Houston M, Prabhat (2018) Exascale deep learning for climate analytics. Proceedings of the International Conference for High Performance Computing, Networking, Storage, and Analysis (SC18), pp. 649-660. https://doi.org/10.1109/SC.2018.00054
Huntingford C, Jeffers ES, Bonsall MB, Christensen HM, Lees T, Yang H (2019) Machine learning and artificial intelligence to aid climate change research and preparedness. Environmental Research Letters 14(12):124007. https://doi.org/10.1088/1748-9326/ab4e55
Scher S (2018) Toward data-driven weather and climate forecasting: approximating a simple general circulation model with deep learning. Geophysical Research Letters 45(22):12616-12622. https://doi.org/10.1029/2018GL080704
Davenport FV, Diffenbaugh NS (2021) Using machine learning to analyze physical causes of climate change: a case study of U.S. Midwest extreme precipitation. Geophysical Research Letters 48(15):e2021GL093787. https://doi.org/10.1029/2021GL093787
Vandal T, Kodra E, Ganguly S, Michaelis A, Nemani R, Ganguly AR (2017) DeepSD: generating high resolution climate change projections through single image super-resolution. Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 1663-1672. https://doi.org/10.1145/3097983.3098004
Karpatne A, Ebert-Uphoff I, Ravela S, Babaie HA, Kumar V (2019) Machine learning for the geosciences: challenges and opportunities. IEEE Transactions on Knowledge and Data Engineering 31(8):1544-1554. https://doi.org/10.1109/TKDE.2018.2861006
Camps-Valls G, Tuia D, Zhu XX, Reichstein M (eds) (2021) Deep learning for the Earth Sciences: a comprehensive approach to remote sensing, climate science and geosciences. Wiley. https://doi.org/10.1002/9781119646181
