Executive Summary 

APCC has contributed to the long-range forecast operated in KMA for the last several years since 2017. We have continuously improved technical affairs and decision making process necessary for long-range forecast. Over the last 3 years, we have tried to improve long-range forecast by combining both observed monitoring factors and its model predictability. The objective seasonal prediction includes the selection of monitoring factors and combination of those factors in the seasonal forecast models. Therefore, we choose various significant monitoring factors for the boreal winter temperature forecast.

To improve the information related to the ENSO in 3-month long range forecast, we investigated the performance of ENSO and its related monthly response in the seasonal forecast models. Although the performance of ENSO index was highly correlated with observation, monthly variation of ENSO related tropical precipitation and atmospheric circulation in East Asia had lower performance in models. The results of this study suggest that it is need to understanding the ENSO response in model as well as information from the other predictors to improve predictability in late winter.

This study investigated the impact of the Madden-Julian Oscillation (MJO) as a tropical oceanic climate driver for weekly winter temperature forecasts and proposed its application for 1-month outlooks. By analyzing the convective and tropical-midlatitude teleconnection responses of the MJO under ENSO mean background states, representative influence phases were identified. Through experiments using the Linear Baroclinic Model (LBM), we investigated the impact of mid-latitude wave propagation caused by tropical convection activities associated with the MJO. The experiment supports the influence of the MJO on the East Asian region analyzed from observation data. Additionally, the predictive performance of ECMWF model for ENSO-MJO influence phases was evaluated, leading to the development of practical guidelines for +3-week temperature forecasts.

In this study, we investigated the performance of North Pacific Oscillation (NPO) and its delayed impact for seasonal forecast models. Models reproduced the December NPO pattern and its one-month delayed impact processes as in observation. However, the predictability of NPO index had low performance except lead-0. Also, models simulated the stronger influence of ENSO than NPO, which was different from observation. It is positively correlated that the December NPO with January East Asia geopotential height, SST and temperature in Korea, and they had more tight tendency when NPO and ENSO has same phase. Models also simulated similar tendency when NPO and ENSO in model has same phase. We suggest that it could be useful in the January temperature forecast when models predict a strong NPO consistently with same phase of ENSO.

We focused on the Western Pacific (WP) pattern, a key atmospheric teleconnection influencing winter temperature over South Korea. Our analysis revealed that the December WP pattern consistently exhibits a strong correlation with South Korean temperature, whereas its influence on January and February temperatures has weakened since the 1990s. Models effectively reproduced the December WP pattern and its associated atmospheric mechanisms, including the northward shift of the upper-level jet, warm southerly wind anomalies, and a rise in air temperature during its positive phase. As the models better capture the interannual variability of the WP index than air temperature, the phase of the WP plays a crucial role in monitoring and predicting temperature variation. However, limitations remain in fully simulating the independent effect of the WP teleconnection without the influence of the ENSO. These findings highlight the significance of the December WP pattern for temperature forecasting and emphasize the need for improved models to better capture atmospheric teleconnection dynamics.

Snow cover in the Eurasian region in the fall (October) is a major factor in explaining winter temperature variability in East Asia. In early winter, radiative cooling caused by snow cover leads to the development of continental high pressure and a decrease in temperature, which is well reproduced in seasonal prediction models. On the other hand, in late winter, the relationship between snow cover-Arctic Oscillation-East Asia temperature has clear non-stationarity depending on the periods and model reproduciblity for this relationship is also low. Therefore, in early winter, Eurasian snow cover can be used as a monitoring factor for predicting winter tempeature, but development additional factors is necessary to improve predictability in late winter.

The autumnal Arctic sea ice and Eurasian snow cover have been considered as an major predictors for predicting winter temperature in South Korea. However, since these two factors share a common mechanism via the stratosphere, their effects may sometimes be double-counted in actual predictions. This study examined the individual and combined effects of these two predictors and evaluated how well these effects are represented in operational seasonal prediction model. The combined influence of the two predictors maintains a similar pattern throughout the winter months, with a particularly pronounced impacts on warm and cold anomalies over South Korea in January. The model simulates the atmospheric response to snow cover more effectively than to sea ice, and it shows higher predictability for the combined effect of the two predictors when sea ice in October is low, and snow cover in November is extensive over East Siberia.