Can Machine Learning Models be a Suitable Tool for Predicting Central European Cold Winter Weather on Subseasonal to Seasonal Timescales?

Abstract

Skillful weather prediction on subseasonal to seasonal timescales is crucial for many socio-economic ventures. But forecasting, especially extremes, on these timescales is very challenging as the information from initial conditions is gradually lost. Therefore, data-driven methods are discussed as an alternative to numerical weather prediction models. Here, Quantile Regression Forests (QRFs) and Random Forest Classifiers (RFCs) are used for probabilistic forecasting of Central European wintertime mean 2-meter temperatures and cold wave days at lead times of 14, 21 and 28 days. ERA5-reanalysis meteorological predictors are used as input data for the machine learning models. The predictions are compared for the winters 2000/2001 to 2019/2020 to a climatological ensemble obtained from E-OBS observational data. The evaluation is performed as full distribution predictions for continuous values using the Continuous Ranked Probability Skill Score and as binary categorical forecasts using the Brier Skill Score. We find skill at lead times up to 28 days in the 20-winter mean and for individual winters. Case studies show that all used machine learning models are able to learn pattern in the data beyond climatology. A more detailed analysis using Shapley Additive Explanations suggest, that both Random-Forest (RF) based models are able to learn physically known relationships in the data. This underlines that RF-based data-driven models can be a suitable tool for forecasting Central European wintertime 2-meter temperatures and the occurrence of cold wave days.