Operation and Improvement of APCC Climate Prediction System

Yoojin Kim, A-Young Lim, Changmook Lim, Daeun Jung, Jaewon Choi, Youngmi Min, Yoobin Yhang
  • 요약
  • 목차

Executive Summary


Since its establishment in 2005, APCC has made great efforts to develop and improve the long-term seasonal forecast technology using the Multi-Model Ensemble (MME) and as a result, it has established a climate prediction system using various global models of the world’s leading climate forecast operating and research institutes to provide climate monitoring information and long-term forecast and verification information on its website and
platform every month.

The climate prediction system, a foundation of APCC, consists of seasonal forecast system and intraseasonal forecast system. The Seasonal forecast system consists of APCC MME, Fire and haze early warning system, and SCoPS (Seamless Coupled Prediction System), which is an in-house seasonal forecast model of APCC. Also, BSISO (Boreal Summer IntraSeasonal Oscillation) forecast system is operated as the intraseasonal forecast system.

The AFS (Automated Forecast System), APCC MME monitoring/forecast/verification system, was build with the establishment of APCC. It has been improved and advanced persistently but it needs more being efficient and systematic at that time, so this project was set at 2017. The AFS is progressed since then; the monitoring/verification system is improved in 2019, the monitoring/forecast/verification system is unified in 2020. Continuously in 2021, an unification of source code, improvement of efficiency, temporal and spacial expansion are performed to improve the AFS. Thourgh this improvement, hindcast period of APCC MME was shifted from 1983-2003 to 1991-2010 then relatively newest climate forecast models with good-performance participated in MME and influence MME’s performance positively.

The participating models of APCC MME have changed steadily. A new model is added and for models are upgraded in 2021, so MME’s forecast skill is improved. When an individual model was changed or newly added, the forecast skills of new/old version of individual model and influence of the change to MME were analyzed every time but it would be consumptive and analysis range would be subjective by an analyst. Thus, more systematic analysis and management is necessary to run, so the metrics of forecast skill to compare new/upgraded individual models are configured and systematized. It is an automated system for calculation and also documentation, so expected to increase the efficiency of running the MME operating system.

For users in South Korea, new forecast and monitoring contents of East Asia region were launched at Korean homepage in 2019. But lack of forecast contents and probabilistic forecast information in Korean for domestic users were pointed out, so a new system is developed to display probabilistic spread of East Asian forecast data and added a new content to the Korean homepage in 2021.

In 2020, APCC monitoring system moved the climatology period to 1991-2010 consistent tothe forecast hindcast period. But the climatology period does not contain the recent climatological changes and the gap would grow by time goes so the climatology period shifted to recent 30 years, 1991-2020, which is recommended by WMO. The new climatology period is adapted to the climate monitoring system including ‘Current Climate’, and ‘Climate Indices’ contents in APCC homepage this year.

The intraseasonal forecast system is operated since 2013 and BSISO index forecast/verification/monitoring information is displayed and serviced through APCC’s homepage and platforms. in 2020, the new ‘BSISO impact anomaly’ content is developed but the interior source code is not changed so an improvement is needed at this time; an extension of forecast lead time, changes to the new climatological period, etc. Therefore, BSISO system is diagnosed to run the system efficiently in a long-term period and direction of problems and posibility of improvement is presented in this year.

In 2020, the users of APCC MME demanded the high-resolution forecast infromation and early issuing of seasonal forecast information to lead the area of seasonal prediction by APCC MME. To meet the demand, the temporally and spaciously expanded system is established and base research is performed this year. Many forecast institutes are running a higher resolution model than the current APCC MME’s resolution (2.5°x2.5°). To make a common resolution of participating models, lower resolution is adapted to APCC MME; and yields a loss of forecast information of original model and usability in a part. Thus a new system is established to make the high-resolution (1°x1°) MME forecast data to overcome the loss of original forecast data. Also comparison and analysis contents of forecast skills depand of resolution are included in this report to deliver the related objective information to users.


In 2019, APCC MME release date is advanced from 25 to 20 through an improvement of AFS. But many operating institutes for climate forecast producing/utilizing are still need earlier forecast data release to meet their schedule, for example, Korean Meteorological Agency (KMA) produces the model data and analyze around 15th to release the seasonal forecast outlook every 20th. Climate forecast institutes in the world are running similar schedule with KMA, so the early release of APCC MME data would increase the usability of these institutes. Therefore the possibility of earlier issue around 15th are investigated in the sense of efficiency and forecast performance in this report. Also the possibility of producing the 0.5-month leading forecast data around 10th is also investigated.

Though three years, 2019-2021, APCC climate prediction system is improved its efficiency, usability, and expand the system temporally and spaciously. Mary researches and improvements through this project will be the foundation to the next level climate prediction system and drive up the APCC MME’s competitiveness.