Short-term climate prediction for East Asia and the western North Pacific (EA–WNP) remains challenging due to systematic errors in current general circulation models. The first version of the integrated climate model (ICM.V1) already demonstrated high skill in reproducing climatology and interannual variability, but retained notable errors: an overly cold tongue in the tropical Pacific, a double intertropical convergence zone, excessive westward extension of El Niño anomalies, and underestimated strength of key feedback mechanisms. These deficiencies were linked to the low horizontal resolution of the atmospheric component (T31, about 3.75°). This work continues the model development line for seasonal prediction started with ICM.V1 and aligns with efforts to improve the simulation of monsoon circulation and El Niño in climate models.
ICM.V2 retains the architecture of ICM.V1: the ECHAM5 atmospheric model, the NEMO 2.3 ocean model, and the OASIS3 coupler. The key change is increasing the horizontal resolution of the atmospheric model from T31 (3.75° × 3.75°) to T63 (1.875° × 1.875°), while the vertical resolution remains at 19 levels. The atmospheric model's time step is reduced from 2400 to 1200 seconds, the ocean model's time step is unchanged (2400 seconds), and the coupling frequency between components is preserved (once every 4 hours). The ocean model retains its previous resolution: about 2° at high latitudes and about 0.5° meridionally near the equator, with 31 vertical levels. The model was integrated for 1000 years, and the last 300 years (model years 701–1000) were used for analysis. Comparisons were made with observational data: HadISST for sea surface temperature, NCEP/NCAR reanalysis for atmospheric fields, and GPCP for precipitation over the period 1981–2010. To assess interannual variability, the 300-year series was divided into ten 30-year segments and the results were averaged.
ICM.V2 markedly reduced systematic sea surface temperature errors: in the 60°S–60°N band, deviations from observations mostly do not exceed 1°C, whereas in ICM.V1 areas with errors of 1–2°C were considerably broader. The simulation of the cold tongue in the tropical Pacific and warm anomalies off the western coasts of South Africa and the Americas improved. The climatology of the EA–WNP summer monsoon became more realistic: the excessively strong anticyclonic anomaly over the western North Pacific disappeared, and the mei-yu rainband is no longer shifted northward; its intensity and position are close to observed. El Niño variability improved: the dominant period of the Niño3 index in ICM.V2 is 3–4 years, closer to the observed approximately 4 years than in ICM.V1 (about 3 years). The spatial structure of sea surface temperature anomalies during El Niño became less elongated westward, although some excessive westward extent remains. The impact of El Niño on the summer climate of EA–WNP is reproduced much better: the amplitude of the anomalous anticyclone over the western North Pacific and the associated dipole precipitation pattern approached observations. The key mechanisms, the tropical Indian Ocean capacitor effect and the local air–sea feedback, became more realistic: the Indian Ocean warming after El Niño persists into summer with the correct strength, and wind and temperature anomalies over the western North Pacific are sustained throughout the summer season. However, a new error emerged: in June–August, negative sea surface temperature anomalies appear over the region 110°–150°E, whereas observations show positive anomalies.
This work demonstrates that increasing the horizontal resolution of the atmospheric component from T31 to T63 substantially improves the simulation of key processes governing short-term climate variability in East Asia and the western North Pacific. Unlike many models, ICM.V2 realistically reproduces both major mechanisms of El Niño's influence on the region's summer climate: the tropical Indian Ocean capacitor effect and the local air–sea feedback. This places ICM.V2 among the most successful models for seasonal prediction in this region. The improvement is not solely due to resolution but also to a more accurate sea surface temperature climatology, highlighting the importance of the mean state for correctly simulating anomalies.
Further model development may include increasing the resolution of the ocean component, which remains unchanged in ICM.V2, as well as improving physical parameterizations, especially boundary layer and cloud processes. Tuning of atmosphere–ocean interactions in the tropics may be needed to eliminate the new error in the western North Pacific. Also promising is the use of ICM.V2 for ensemble forecasts and studies of decadal climate variability.
The improved model could enhance the quality of operational seasonal forecasts for East Asia, including predictions of summer precipitation and circulation anomalies. The results are also useful for studying the mechanisms of El Niño's influence on regional climate and for assessing future changes under global warming.
The next step could be applying ICM.V2 to operational forecasting and comparing its forecast skill with other models within international projects. It is also important to investigate the causes of the new error in the western North Pacific and attempt to eliminate it.
This work is directly related to the fundamental problem of reproducing El Niño and its global teleconnections in climate models. Despite progress, open questions remain: why models systematically shift El Niño anomalies westward, how to improve the simulation of interannual and interdecadal variability in the tropical Pacific, and which physical processes are responsible for errors in air–sea feedbacks. ICM.V2 contributes to solving these problems but does not fully close them.
The ICM.V2 model was integrated for 1000 model years, equivalent to a thousand-year climate history, yet it uses an atmospheric resolution of T63, which corresponds to a grid of about 210 km at the equator, enough to distinguish individual monsoon systems but still too coarse to reproduce mesoscale phenomena such as tropical cyclones.