연구보고서
- 저자
- 신선희 박사
- 작성일
- 2019.06.03
- 조회
- 523
- 요약
- 목차
This project was designed to improve the stability, predictability, and utilization of the APEC Climate Center (APCC) seasonal forecasting system. The structure of the current prediction system has been reorganized and its efficiency improved to maintain stable and efficient operation. An additional East Asian Winter Monsoon (EAWM) prediction system has also been constructed using a newly developed prediction technology. To improve real-time forecast information, the APCC real-time forecast was evaluated and a hindcast sensitivity test performed to improve the utilization of the individual models contributing to the APCC MME prediction.
The APCC Sea Surface Temperature/El Niño–Southern Oscillation (SST/ENSO) prediction system was systematically integrated with the improved APCC Multi-Model Ensemble (MME) prediction system, and the APCC SST/ENSO prediction and graphics system and MME graphics system were developed and improved to ensure the stable and efficient operation of the APCC Automatic Forecast System (AFS). The ruby-, NCL-, and shell script- based run scripts were converted to the python programming language and the folder structure of SST/ENSO and MME graphics systems were improved. In addition, the running time of the SST/ENSO graphics system was improved using parallel processing and developing MME and SST/ENSO automatic run scrips. It is expected that the enhanced MME system in this project will operated more efficiently and systematically than the existing systems.
In 2017, a project to develop forecasting technology for the East Asian winter monsoon was carried out to improve the predictability of the winter season in East Asia, which is a weak part of the APCC seasonal forecast information. We developed a dynamical-statistical hybrid prediction model for the East Asian Winter Monsoon (EAWM) based on APCC MME seasonal forecast and have confirmed the former is stable even during the four month lead time. In this project, we established an automatic system to systematically operate the EAWM prediction model and wanted to provide improved winter forecast information for East Asia by commercializing previously developed prediction technology. It is expected to provide additional information on the intensity of the EAWM that dominates the East Asian winter climate, along with the APCC MME forecast information.
Assessing the skill of the seasonal forecast is essential to investigating the current levels and limitations of the performance of the APCC MME prediction system. This study was assessed the predictability of the individual models and MME forecasts that were disseminated to APEC member economies every month in 2017 (2017JFM –2017/18DJF) via our website, following the project from last year. The year 2017 was characterized by warmer to much warmer than average conditions across much of the globe; it was the warmest year without an El Niño present in the tropical Pacific Ocean. The skill of the real-time MME forecast for temperature is generally higher than that of hindcast (1983–2005) and the recent nine-year (2008–2016) period. This may be due to global warming but it also shows that there is a limit to the predictability for precipitation in ENSO-neutral or weak La Niña conditions.
To improve the usability of models that could not participate in the APCC MME prediction because of the constraints of the hindcast period, we conducted a sensitivity test using a different hindcast period, statistical significance test between climatologies from different hindcast periods, and comparing the skill of MMEs using different model sets. Results indicate that the 23-year common period for individual models used in the APCC MME predictions can be extended to 28 years with a relatively low sensitivity according to the different hindcast period. For these reasons, we also confirmed the possibility of including several models, including at least GLOSEA5 and UKMO, that did not participate in the APCC MME predictions