Executive Summary

The Asia–Pacific region experiences pronounced variability on annual-to-decadal (A2D) timescales, and this low-frequency climate noise strongly conditions the occurrence of high-impact extremes such as droughts, heat waves, cold surges, and frontal heavy rainfall. Yet most current prediction and climate-service systems focus either on long-term climate change (multi-decadal trends) or on short-range to seasonal forecasts, leaving a critical gap at the 1–10 year horizon where infrastructure planning, water-resource management, energy investment, and adaptation policies actually operate. Large-scale coupled modes such as ENSO, PDO/IPO, and NAO evolve and interact on A2D timescales, setting the background state that either amplifies or suppresses extremes in particular “regimes.” Understanding these memory pathways and regime-dependent teleconnections, and improving the A2D performance of prediction systems like CMIP6 DCPP, is therefore essential for providing robust, storylines-based climate information that can directly support medium- to long-term decision-making and climate services in a warming world.

The spring drought analysis reveals significant power at 2–3-year ENSO timescales and at 11 17-year quasi-decadal timescales, with clear non-stationarity over 1960–2015. It demonstrates that ENSO primarily controls year-to-year SPI-6 variability, while PDO/IPO increasingly dominate at longer periods, and that the strength of these teleconnections changes over time. ENSO–PDO phase diagrams and winter–spring composite circulation patterns further show strong asymmetry: the ENSO−/PDO+ regime is associated with the driest springs via a pronounced anticyclonic ridge and moisture divergence over East Asia, whereas ENSO+/PDO− tends to suppress drought. Low-frequency band analysis and interaction regressions (including ENSO×PDO) quantify PDO’s role as a decadal background setter and confirm that ENSO PDO combinations modulate spring drought risk nonlinearly. Together, these results argue for a regime-aware, interaction-based A2D prediction framework for East Asian spring drought, rather than relying on a single stationary ENSO–SPI relationship.

We also evaluates the multi-year prediction skill for spring (MAM) drought, defined by the 6-month Standardized Precipitation Index (SPI6), using hindcasts from five DCPP models (CanESM5, CMCC-CM2-SR5, HadGEM3-GC31-MM, MIROC6, and MPI-ESM1-2-HR) against CRU observations. Results showed the distinct differences in model performance for drought prediction. MIROC6 demonstrated the highest performance in deterministic verification, whereas CanESM5 excelled in probabilistic verification. These characteristics stem from differing error properties: CanESM5 showed high reliability but low resolution in its probability forecasts, while MIROC6 exhibited high discrimination ability between drought and non-drought events but low reliability. These findings imply that rather than applying a uniform bias correction method across all models, a tailored correction approach that accounts for the specific predictive characteristics of each model is necessary.

This study is conducted in two major topics to obtain climate predictors suitable for East Asian extreme temperature in summer. First, with dynamical analysis, we obtained the important components of the dynamical process and the atmospheric teleconnection patterns related to the variability of the extreme temperature. Second, we assessed the applicability of the climate predictors through multi-regression model construction designed to diagnose and validate their predictive skills. These two research tasks laid the groundwork for leveraging climate predictors to predict East Asian extreme high temperatures within the A2D time scale.

Using six heat-related indices, we evaluates summer heatwaves in five DCPP models over East Asia. Summer days and warm days generally showed positive biases, while warm days and warm nights exhibit increasing biases with longer lead times. Most indices display overestimated warming trends, with sharp rises in warm days and warm nights after the 2000s. Predictability is highest for warm days and warm nights, and maximum value of daily minimum temperature outperformed among intensity indices. CanESM5 shows overly strong trends in Tmin-derived indices, whereas MPI-EMS1-2-HR achieves the best overall accuracy across most indices.

We also defines rapid warming and cooling events in East Asia during winter and examines the associated ocean-atmosphere patterns. The frequencies of both rapid warming and cooling shows significant negative correlations with the NAO and AO. While the negative NAO acts as a slow forcing that shapes the background state over East Asia, the actual abrupt temperature changes are determined by the AO phase transitions. Given that the NAO is known to possess relatively high near-term predictability, it may serve as a valuable indicator for forecasting the frequency of rapid temperature variability over East Asia.

Summer precipitation in East Asia is governed by highly complex atmospheric systems. Understanding the characteristics of each system is essential for improving near future (A2D) climate predictions. This study identifies a significant northward migrations of the East Asian summer front in the recent decades. Such migration has intensified heatwaves and increased heavy frontal precipitation. These findings highlight that understanding frontal behavior is crucial for reducing climate-related risks and improving adaptation strategies.