Potential Predictability of the Tropical Cyclone Frequency Over the Western North Pacific With 50-km AGCM Ensemble Experiments

The potential predictability determined by sea surface temperature (SST) of the frequency distribution of tropical cyclones (TCs) is studied using ensemble simulations with a 50-km resolution atmospheric general circulation model (AGCM). In this experiment, the interannual variability of the TC frequency over the western North Pacific (WNP) is primarily determined by SST, as the simulated ensemble mean signal is dominant over the ensemble spread and effectively follows the observed TC frequency in June-July-August (JJA). In contrast, the correlation between the observed and simulated TC frequency variability is less significant in September-October-November (SON). To explore the TC frequency-SST linkage, the relationship between the TC frequency variability over the WNP and climate modes determined by SST indices is investigated. Through a correlation analysis, we found that El Niño Modoki is the major driver of the TC frequency variability over the WNP in JJA. Partial correlation analysis further reveals that internal atmospheric dynamics are important for the TC frequency variability over the WNP in SON. These results suggest that the tropical climate mode is responsible for the seasonal predictability of the TC frequency variability over the WNP and that the prediction skill is seasonally dependent with a high (low) skill in JJA (SON).