North China, a region with climatologically temperate summers, has experienced a pronounced increase in extreme heat events under global warming, amplifying risks to societal resilience and public health. However, the predictability of these extremes, particularly sources of prediction errors, remains inadequately quantified. This study investigates prediction skill and potential sources of prediction errors for North China summer heat extremes using hindcasts and real-time forecasts from the Beijing Climate Centre (BCC) and ECMWF subseasonal to seasonal (S2S) prediction systems. Our analysis reveals that effective prediction for North China summer heat extremes is limited to approximately 2–3 pentads, with ECMWF consistently outperforming BCC in both deterministic and probabilistic forecasts. The primary constraint on prediction skill stems from model performance in forecasting Eurasian mid–high latitude atmospheric circulation patterns, which largely explains ECMWF’s superior performance. While both systems reproduce observed relationships between North China heat extremes and local anticyclonic anomalies, ECMWF maintains significant prediction skill for critical mid–high latitude circulation features at 3-pentad leads, whereas BCC’s skill becomes confined to low-latitude regions. Real-time forecasts of two distinct 2024 heat events further validate this circulation-dependent predictability. These findings underscore that advancing heat extreme prediction requires targeted improvements in large-scale circulation forecasting, providing prioritized development pathways for S2S models. Notably, probabilistic forecasts provided meaningful early risk indicators even when deterministic thresholds were unmet, offering actionable technical approaches to enhance operational early warning capabilities.

In May–June 2022, South China (SC) experienced record-breaking rainfall, resulting in severe flooding and significant socioeconomic impacts, which posed a challenge to operational seasonal forecasting efforts. Notably, most forecast systems and their multi-model ensemble severely underestimated the SC floods in early summer 2022, forecasting decreased rainfall instead. Observational analysis links the 2022 SC floods to an anomalous low-level anticyclone over the western North Pacific and a zonal wave train across the mid-to-high latitudes of Eurasia. The forecast systems, however, generally missed the Eurasian zonal wave train and predicted an anomalous low-level cyclone over the western North Pacific, which would typically result in decreased rainfall over SC. Further analysis suggests that the La Niña-related cold sea surface temperature (SST) anomalies in the tropical Pacific, accurately predicted by the forecast systems, contributed to the modeled anomalous cyclone over the western North Pacific. Excessive reliance on the La Niña-related Pacific SST anomalies appears to be the primary driver of the inaccurate prediction of the 2022 SC floods. The SST bias in the North Atlantic also contributed, albeit to a lesser extent, by influencing the anomalous western North Pacific anticyclone and the Eurasian mid-to-high latitude wave train. Additionally, certain ensemble members from the forecast systems did predict increased rainfall over SC in their forecasts, yet exhibited a spurious precipitation mechanism inconsistent with observations, primarily due to distinguished differences in the atmospheric circulation patterns over SC and Eurasian mid-to-high latitudes. This implies that factors beyond the Pacific La Niña might have played a more significant role in sustaining the 2022 SC floods. The study emphasizes the need to improve seasonal forecast systems, particularly in their representation of atmospheric internal variability and external forcing beyond the Pacific La Niña/El Niño.

Based on a combination of six Chinese climate models and three international operational models, the China multi-model ensemble (CMME) prediction system has been upgraded into its version 2 (CMMEv2.0) at the National Climate Centre (NCC) of the China Meteorological Administration (CMA) by including new model members and expanding prediction products. A comprehensive assessment of the performance of the upgraded CMME during its hindcast (1993–2016) and real-time prediction (2021–present) periods is conducted in this study. The results demonstrate that CMMEv2.0 outperforms all the individual models by capturing more realistic equatorial sea surface temperature (SST) variability. It exhibits better prediction skills for precipitation and 2-m temperature anomalies, and the improvements in prediction skill of CMMEv2.0 are significant over East Asia. The superiority of CMMEv2.0 can be attributed to its better projection of El Niño–Southern Oscillation (ENSO; with the temporal correlation coefficient score for Niño3.4 index reaching 0.87 at 6-month lead) and ENSO-related teleconnections. As for the real-time prediction in recent three years, CMMEv2.0 has also yielded relatively stable skills; it successfully predicted the primary rainbelt over northern China in summers of 2021–2023 and the warm conditions in winters of 2022/2023. Beyond that, ensemble sampling experiments indicate that the CMMEv2.0 skills become saturated after the ensemble model number increased to 5–6, indicating that selection of only an optimal subgroup of ensemble models could benefit the prediction performance, especially over the extratropics, yet the underlying reasons await future investigation.