Qiang Wang

Reliability is central to probabilistic forecast skill, and linking it to nonprobabilistic diagnostics is essential for interpreting its behavior and deficits. Although recent studies have suggested a possible relationship between reliability in dynamical seasonal prediction and the mismatch between model–observation correlation () and model–model correlation (), evidence remains largely qualitative and limited in scope and generality, with relatively little attention to finite‐ensemble effects. Here, we systematically investigate this relationship by conducting a Monte Carlo (MC) study that generates and analyzes synthetic forecast–observation data within a more general statistical framework than in previous studies and by analyzing global hindcasts from three dynamical prediction systems. Our MC investigation demonstrates that, in the infinite‐ensemble limit, reliability depends intrinsically on the magnitude and sign of the correlation mismatch: it deteriorates with increasing , with corresponding to probabilistic overconfidence and —associated with the signal‐to‐noise paradox—corresponding to probabilistic underconfidence. However, for modest ensemble sizes typical of current prediction systems, finite‐ensemble effects can materially confound the manifestation of this intrinsic relationship, in particular by substantially masking the magnitude of the underconfidence expected when . Our hindcast analysis shows that the diagnosed reliability–mismatch relationship closely follows the MC benchmarks that incorporate finite‐ensemble effects, thereby corroborating the MC findings. These results provide a compact diagnostic lens for interpreting probabilistic reliability in dynamical seasonal prediction, highlight the need to account for ensemble‐size limitations when interpreting reliability diagnostics, and have implications for estimating intrinsic reliability in practical settings and for improving reliability in practice.

The Northeast China Cold Vortex (NECV) is a key circulation system affecting weather patterns over North China, frequently triggering thunderstorms, hail, and other severe convective weather. Accurate prediction of NECVs is therefore of great importance. However, substantial forecast errors still remain, largely due to uncertainties in the initial conditions. To improve NECV forecast skills, we investigate the optimal initial errors and targeted observation sensitive areas using a sampling-based approximation of the conditional nonlinear optimal perturbation (CNOP) method together with the Pangu-Weather deep learning model. We first evaluate the model’s performance over Northeast China and find that Pangu-Weather exhibits forecast skill generally comparable to the ECMWF Integrated Forecasting System (IFS) during the May–August 2022 period over Northeast China. Then the CNOP-based approach is used to capture the optimal initial errors with the greatest impact on NECV forecasts. The largest error amplitudes are primarily located upstream of the vortex and near upper-level jet-entrance regions, which are identified as the targeted observation sensitive areas. Perturbation kinetic-energy diagnostics further indicate that baroclinic conversion is the dominant mechanism for error growth. Observing system simulation experiments suggest that, under an idealized assumption of completely eliminating errors in a given region, targeted observations over the sensitive area can produce the largest forecast improvement, with an average error reduction of approximately 13% relative to other areas. This study contributes to a deeper understanding of NECV predictability and may help improve forecasting capability.

In this study, the air‐sea carbon dioxide flux (FCO₂) in the subarctic North Pacific is investigated using three data products from 1985 to 2016. Extreme CO₂ release occurred during the winters of 1999–2001 with an average FCO₂ anomaly of 1.09 mol C m⁻² y⁻¹ across the three data products, which is remarkably higher than that in other years (−0.11 mol C m⁻² y⁻¹). Empirical analysis reveals that this event is primarily driven by increased carbon dioxide partial pressure (ΔpCO₂), whose contribution to the event is greater than the wind speed at 10 m, followed by sea surface temperature (SST) and salinity impacts. Specifically, the intensification of pCO_2sea (which contributes 58% of the FCO₂ anomaly) is induced by the upwelling of dissolved inorganic carbon (DIC) due to the increase in Ekman pumping caused by the positive anomaly of wind stress curl associated with the Victoria mode of the SST. Moreover, the weakening of pCO_2atm is induced by the negative anomaly of sea level pressure (contributing 9% to the FCO₂ anomaly), which is also related to the mode and the reduction in the mole fraction of CO₂ (contributing 11% to the FCO₂ anomaly), which is related to fossil fuel emissions. Ultimately, the sea surface pCO₂ is significantly oversaturated relative to the atmosphere, triggering extreme CO₂ release in the subarctic North Pacific. This study enhances our understanding of the natural variability of carbon fluxes in the North Pacific.

The Kuroshio Extension (KE) bimodality has important effects on the ocean environment, ecosystem and climate. Previous studies have revealed that the Kuroshio Extension (KE) bimodality is mainly determined by the westward-propagating Rossby wave triggered by the North Pacific decadal variability such as PDO or NPGO: the positive (negative) phase of NPGO corresponds to the stable (unstable) KE state. However, the KE state and the NPGO seem to be decoupled since 2017, during which the NPGO takes a negative phase but the KE is in a stable state. This study employs the Convergent Cross Mapping (CCM) method to investigate the causality between the KE bimodality and NPGO. Simultaneously, we divide the KE region into the upstream (west of 146°E) and downstream regions. It is found that the NPGO has a significant causal impact on the downstream KE state. But the effect on the upstream KE state significantly weakens around 2017. Further analysis indicates that the upstream KE state is mainly caused by eddy activity in the Kuroshio large meander region south of Japan. In particular, the changes in the eddy activity affect the downstream advection of eddies and induce changes in the Kuroshio position over the Izu ridge, which cause different states in the KE upstream region. Therefore, we should not only consider the NPGO change, but also the eddy activity change in the Kuroshio region south of Japan when understanding and predicting the KE low-frequency variability.

Given the essential implications of Kuroshio Extension (KE) bimodality on oceanic dynamical environment and climate, the present study investigates the targeted observation schemes, based on the conditional nonlinear optimal perturbation (CNOP) method and a reduced‐gravity shallow‐water model, to improve the forecast skills of transition processes of KE bimodal states. To obtain a suitable observing array, the observation schemes, with different numbers of observation sites and observation distances between two sites, are designed. Furthermore, to demonstrate the superiority of the observing networks in predicting KE transition processes, two existing observation schemes and six random observation schemes are compared with the CNOP‐determined observing array. Based on this, a relatively optimal observing array with three sites and observation distance of 90 km is established, which is mainly located between 31°N and 33°N in the south of Japan. This targeted observing network is universal for two KE transition processes. The removal of initial errors on this array results in the mean prediction improvements of about 9.2% and 22.5% for KE transition processes from the low‐ to the high‐energy state and from the high‐ to the low‐energy state, respectively.

The Kuroshio intrusion (KI) into the South China Sea (SCS) significantly affects the environment, ecology, and climate change of the SCS. However, due to the nonlinearity of KI, its numerical prediction often requires a large ensemble size to measure prediction uncertainty. The huge computational costs of large numbers of members and high-resolution numerical models pose significant challenges for KI prediction. We, therefore, construct a Kuroshio ensemble deep learning prediction system (KurNet) by taking different values of parameters to predict KI paths because the deep learning models have high prediction skills and low computational cost. The KurNet containing 64 ensemble members not only can output ensemble mean forecast results of the Kuroshio path but also can estimate probability density functions for the path types. The KurNet illustrates a high predictive ability for the KI with the mean classification accuracy of 71.9% and root-mean-square error of 0.913 on the testing set, which is superior to the single control prediction by ∼1.0%–2.9%, although the control prediction model is selected as one of the ensemble members with the best predictive ability on the validation set. Furthermore, the predictability analysis of 10 KI events indicates that when the lead time is 3 days, the most important areas are in the east of Luzon Island due to the upstream Kuroshio transport. As the lead time increases, the most important area is in the Luzon Strait due to the eddy activity. Observing system simulation experiments reveal that the KI forecast skill can be enhanced by ∼12%–18% when uncertainties of the input data in these important regions are removed.

Investigating latent heat flux (LHF) variations in the western boundary current region is crucial for understanding air–sea interactions. In this study, we examine the LHF trend in the East China Sea Kuroshio Region (ECSKR) from 1959 to 2021 using atmospheric and oceanic reanalysis datasets and find that the LHF has a significant strengthening trend. This strengthening can be attributed to sea surface warming resulting from the advection of sea surface temperatures. More importantly, the LHF trend has an apparent seasonal dependence: the most substantial increasing trend in LHF is observed in spring, while the trends are weak in other seasons. Further analysis illustrates that the anomaly of air–sea humidity difference plays a pivotal role in controlling the seasonal variations in LHF trends. Specifically, as a result of the different responses of the East Asian Trough to global warming across different seasons, during spring, the East Asian Trough significantly deepens, resulting in northerly winds that facilitate the intrusion of dry and cold air into the ECSKR region. This intensifies the humidity difference between the sea and air, promoting the release of oceanic latent heat. These findings can contribute to a better understanding of the surface heat budget balance within western boundary currents.