ExtremeCast: Boosting Extreme Value Prediction for Global Weather Forecast
This addresses a critical limitation in ML-based weather forecasting for meteorologists and disaster preparedness agencies, though it's an incremental improvement on existing forecast models.
The paper tackles the problem of inaccurate extreme weather prediction in data-driven global weather forecast models by proving symmetric losses cause biased predictions and introducing Exloss (asymmetric loss) and ExBooster (training-free enhancement module). The solution achieves state-of-the-art performance in extreme weather prediction while maintaining overall forecast accuracy comparable to top models.
Data-driven weather forecast based on machine learning (ML) has experienced rapid development and demonstrated superior performance in the global medium-range forecast compared to traditional physics-based dynamical models. However, most of these ML models struggle with accurately predicting extreme weather, which is related to training loss and the uncertainty of weather systems. Through mathematical analysis, we prove that the use of symmetric losses, such as the Mean Squared Error (MSE), leads to biased predictions and underestimation of extreme values. To address this issue, we introduce Exloss, a novel loss function that performs asymmetric optimization and highlights extreme values to obtain accurate extreme weather forecast. Beyond the evolution in training loss, we introduce a training-free extreme value enhancement module named ExBooster, which captures the uncertainty in prediction outcomes by employing multiple random samples, thereby increasing the hit rate of low-probability extreme events. Combined with an advanced global weather forecast model, extensive experiments show that our solution can achieve state-of-the-art performance in extreme weather prediction, while maintaining the overall forecast accuracy comparable to the top medium-range forecast models.