The Comparison of Extreme Temperature Ensemble Forecasts Between Artificial Intelligence Models and Physics-based Numerical Weather Prediction

Extreme temperature events have significant impacts on human society and economic activities, but the prediction still involves considerable uncertainties, making the use of ensemble forecasting methods crucial. The integration of—Pangu-Weather (PGW)—with the perturbed initial conditions of the China Meteorological Administration Global Ensemble Prediction System (CMA-GEPS) to develop the Pangu-Weather Global Ensemble Prediction System (PGW-GEPS). Using the 2022 extreme heat wave event in Zhejiang and the 2024 cold wave event in Inner Mongolia as case studies, the forecasting performance of PGW-GEPS and CMA-GEPS for these two extreme temperature events was evaluated and compared based on multiple assessment metrics. The results indicate that, in both the Zhejiang heat wave and Inner Mongolia cold wave events, PGW-GEPS exhibited forecasting accuracy and uncertainty representation capabilities comparable to CMA-GEPS. Both systems effectively capture the increase in 2m temperature forecast uncertainty with longer lead times and its subsequent decrease as the forecast initialization approached the observation period. However, during the Zhejiang heat wave, PGW-GEPS showed deficiencies in forecasting the shear line and exhibited larger forecast errors in the medium range. A comparative analysis of the kinetic energy spectra of these two events further revealed that PGW-GEPS exhibited an attenuation phenomenon on sub-synoptic scales. In summary, the AI-based PGW-GEPS demonstrates forecasting capability for extreme temperature events.. Its forecast accuracy for 3 to 10-day extreme temperature predictions is comparable to that of CMA-GEPS, and it also has certain advantages in computational speed. However, PGW-GEPS still faces challenges in capturing meso-micro scale rapidly evolving weather systems, requiring further improvement in forecasting sub-synoptic systems. This study provides valuable insights into the application of artificial intelligence models in ensemble forecasting.