Machine Learning-Based Prediction of Extreme Temperature Events in Texas: Understanding the Role of Large-Scale Climate Modes

Extreme warming events in Texas have far-reaching environmental, economic, and societal consequences, including impacts on agriculture, energy demand, public health, and infrastructure. These events underscore the urgent need for reliable prediction systems that can anticipate their occurrence and inform mitigation and adaptation strategies. In this study, we develop machine-learning-based models to predict extreme temperature events across Texas by identifying and modeling the key drivers of these phenomena. The predictive framework incorporates the influences of large-scale climate modes and processes from both the Pacific and North Atlantic Oceans, including the El Niño–Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO), Warm Pool (WP), and Atlantic Multidecadal Oscillation (AMO). By integrating these climate indices with regional atmospheric and surface data, the model captures the complex interactions between large-scale climate variability and regional temperature extremes. The contributions of each climate mode are quantified and analyzed to determine their relative importance in driving warming events across different temporal and spatial scales. To ensure the robustness of the predictions, the model outputs are further validated against physical mechanisms linking large-scale climate modes to atmospheric circulation patterns. This validation process provides a mechanistic understanding of the statistical relationships uncovered by the machine-learning models, ensuring that the predictions align with established climate dynamics. The findings from this study enhance our understanding of regional climate dynamics in Texas and demonstrate the potential of machine-learning approaches for improving the predictability of extreme temperature events.