APCC Multi-Model Ensemble Forecasting and an Empirical-Dynamical Approach for Summertime Temperature Prediction in South Korea

English Abstract:

The APEC Climate Center (APCC) is a leading climate information service provider and research institute that serves the 21 APEC member economies as well as the broader international community. Currently, 16 operational centers and research institutes from around the world participate in the APCC operational multi-model ensemble (MME) prediction system by routinely providing ensembles of global forecast fields. Various MME methods implemented in APCC’s MME system were compared with a simple averaged MME (with equal weightings) to assess one-month lead seasonal mean temperature and precipitation. The results showed that the simple averaged MME is generally superior to the empirically-weighted multiple regression MMEs if all aspects of the predictions were taken into account. On the other hand, the calibrated MME with a step-wise pattern projection resulted in reduced errors and improved forecast skills in a large proportion of cases. Two empirical models, a multivariate regression model and a simple empirical model, were established based on the tropical-extratropical teleconnection patterns and applied to the APCC MME prediction to investigate the predictability of summertime temperature in South Korea. The multivariate regression model outperformed the simple empirical model at all lead times ranging from one to six months. The notably different predictability arguably arose from the higher fidelity of the simulated tropical SST in the Central and Eastern Pacific than in the Western Pacific. In addition, the multivariate model outstripped the one-month lead MME prediction, suggesting that the reproducibility of tropical SST, despite the regional dependency, can pave a promising way for the use of an empirical-dynamical prediction for summertime temperature in South Korea.