연구보고서
- 저자
- 함수련 박사
- 작성일
- 2019.06.12
- 조회
- 500
- 요약
- 목차
This study introduces processes to operate and improve the newly developed APEC Climate Center (APCC) in-house model (Seamless Coupled Prediction System; SCoPS) for the APCC multi-model ensemble (MME) system for predicting seasonal forecasts. The SCoPS is a state-of-the-art global prediction system for seasonal time scales, based on a fully coupled climate model with integrated initialization processes for the atmosphere, ocean, and sea ice. The SCoPS initialized data for 10-member ensembles was assimilated by NCEP CFS data and several observational subsurface profile data. Real-time forecast runs are started at fixed calendar dates, the 1st and 5th of each month, with five perturbed ensemble members by gaussian spread and integrated for up to seven months. In previous study, there was evaluation of prediction skill compared to the operation model (the APEC Climate Center Community Climate System Model version 3; APCC CCSM3) using 32 year (1982–2013) ensemble hindcast runs. As a result, we found that seasonal climate forecasts using SCoPS are useful for simulating climate variability, especially in East Asian monsoon system. Furthermore, prediction results from hindcasts and forecasts show perform well in current MME participant models. Based on these studies, we have provided newly developed SCoPS seasonal forecast data to the APCC multimodel ensemble (MME) system as the APCC operational model every month since November 2017.
The SCoPS system has been updated and the initial processes have been improved. In previous studies, we found that there were significant systematic biases in subsurface ocean temperature and that these lead to excessive or weak ENSO signals with long lead times. Now, some ocean initialization processes related to subsurface temperature has been corrected. However, systematic biases in subsurface ocean temperature are still significant, especially in terms of the excessive positive anomaly below the thermocline in the eastern Pacific. This study evaluates some variables related to the ocean temperature biases. The initial condition of SCoPS tends to have cold biases in surface heat fluxes and deepens the thermocline. To improve the ocean circulation feature in SCoPS, we have suggested modification of vertical diffusivity profile in the ocean model to change the thermal structure of the ocean. Increasing and decreasing the ocean diffusivity coefficient leads to reductions in the sea surface temperature biases and precipitation in seasonal forecasts even causes minor changes in the thermal structure of the ocean. However, because the spin-up time of the equatorial ocean is about 10–20 years and the timescale of subtropical-tropical exchange is about 20–30 years, the sensitivity experiment needs a longer integration time.
Meanwhile, because of the recent increase in the occurrence of extreme weather and climate events, data on the prediction of extreme events in various time frames is being requested. Therefore, by developing the SCoPS subseasonal forecast processes, the predictability is being evaluated. It can expect to expand the application for prediction data such as APCC BSISO. Discussions with APCC BSISO working group determine the configuration of subseasonal predictions based on information from the NMME group discussion, which is part of the S2S project. In one case study (MJJAS 2014–2015), we found that the result of predictions with a 1–3 week lead time shows a reasonable spatial distribution of large scale circulations and precipitation related to the East Asian monsoon system, although the TCC of the predicted variables is lower than those based on a lead time of 4–5 weeks. Furthermore, these subseasonal forecast data have been provided to the APCC BSISO forecast system. A figure that validates the prediction skill of the BSISO index is given in this study.