Decoding sub-seasonal predictors of extreme heat with interpretable machine learning

Jagger Alexander, Zong-Liang Yang

As climate change accelerates, heat waves are becoming more frequent, intense, and deadly. Enhancing forecasting capabilities through a better understanding of sub-seasonal predictors of extreme heat is crucial for adaptation efforts. This study utilizes an interpretable machine learning model, implementing Extreme Gradient Boosting (XGBoost) with SHapley Additive exPlanations (SHAP), to evaluate the predictive strength of various climate factors—including local weather, global climate indices, geopotential heights, soil moisture, and sea surface temperatures—on extreme daily maximum temperatures. This model demonstrates strong predictive performance for extreme heat in Austin, TX, USA, on the sub-seasonal time scale, with soil moisture features emerging as more influential than atmospheric features. Notably, our analysis uncovers previously underexplored teleconnections between distant soil moisture anomalies and local extreme heat, warranting further investigation. It is also shown that the Madden-Julian Oscillation (MJO) has predictive value for extreme heat in Austin, underscoring its utility relative to other indices like ENSO and NAO. This method shows promise for application to other cities and for integration with dynamical modeling approaches, advancing sub-seasonal extreme heat forecasting more broadly.