연구보고서
- 저자
- 정혜인 연구원
- 작성일
- 2016.01.23
- 조회
- 270
- 요약
- 목차
The impact of ocean initial conditions on the seasonal climate forecasts of two major types of El Niño, namely canonical (or cold tongue) and a new type of El Niño (El Niño Modoki, Warm pool El Niño, or Central-Pacific El Niño), using the APEC Climate Center (APCC) in-house coupled model is assessed for the boreal winter. We use the Community Climate System Model version 3 (CCSM3), developed by the National Center for Atmospheric Research (NCAR), as an APCC in-house coupled model. This APCC/CCSM3 coupled model is composed of the Community Atmospheric Model version 3 (CAM3), the Parallel Ocean Program (POP), with the Community Land Model (CLM) as the land surface model and the Community Sea Ice model version 4 (CSIM4) as the sea ice component.
In this study, two different ocean initialization approaches are involved in working toward improved forecast skill for the prediction of ENSO. The first method is to integrate the coupled GCM with Sea Surface Temperature (SST) nudging, which is an empirical method for data assimilation.The hindcast simulation with the initial conditions from the SST nudging method is hereafter referred to as the “Nudging experiment”. The second method is to use three dimensional ocean assimilated reanalysis data from the Global Ocean Data Assimilation System (GODAS), which is assimilated with a 3-dimensional variational method using the GFDL MOM3 ocean circulation model. The hindcast simulation with the initial conditions from GODAS will hereafter be referred to as the “GODAS experiment”. With these two different ocean initial conditions, we simulate a relatively long-term seasonal prediction for 7–months, starting from February, May, August and November of each year between 1983 and 2010 in the APCC/CCSM3 model. Each 7-month retrospective forecast consists of 5 ensemble members.
To more precisely understand the predictability of two types of El Niño according to two different ocean initial conditions, we examined the predicted performance of tropical Pacific SST anomalies. We used the Niño3 index (ENSO Modoki Index; EMI) to characterize the canonical El Niño (a new type of El Niño) event. Except for the August initial condition, the temporal correlation coefficients of the GODAS experiment for the Niño3 with a forecast lead-time perform better than those of the Nudging experiment. However, for the EMI, the Nudging experiment is slightly better than GODAS experiment. To assess the interannual predictability of the tropical Pacific SST anomalies, we calculated the pattern correlation coefficients over the tropical Pacific region in November. The results indicate that the skill of the Nudging experiment are higher than that of the GODAS experiment at each lead-time with different initial conditions. We also investigated the composite analysis for the canonical El Niño years (1986-87, 1997-98, and 2006-07) and the new types of El Niño years (1987-88, 1991-92, 1994-95, 2002-03, 2004-05, and 2009-10) within the study period. For the canonical El Niño years, the GODAS experiment represents the thermodynamic structure of the tropical ocean interior well. On the other hand, the intensities of SST, rainfall, and 500 hPa geopotential height anomalies in the GODAS experiment are relatively weak compared to the observational data or the Nudging experiment. For the new types of El Niño years, the Nudging experiment matches relatively well with the observed ocean dynamic features, demonstrating that the warm ocean temperature shifted westward. However, it is noteworthy that both the Nudging and GODAS experiments do not separate the distinct climate features in terms of the new types of El Niño years.
In this study, we demonstrate that the reason for the low predictability for ENSO prediction in the GODAS experiment is the initial shock due to the differences in climatology between the GODAS data and the CCSM3 model. Further, we also indicate that there are limitations in ENSO prediction influenced by a systematic model bias in the APCC/CCSM3 model through the EOF analysis of SST anomalies over the tropical Pacific, based on three experiments, namely the GODAS, Nudging and Control experiments. Nevertheless, the APCC in-house coupled model prediction has some strength as far as the canonical and new types of El Niño predictions when compared to other coupled model predictions collected by APCC for multi-model ensemble (MME) prediction. The temporal correlation coefficients of Niño3 and EMI at a forecast lead-time in the GODAS experiment show considerably high values for up to a 6-month lead time. The APCC/CCSM3 indicates better performance in terms of composite SST anomalies associated with the two types of El Niño, in comparison to other coupled models. We conclude that the APCC/CCSM3 has considerable usefulness once long-term ENSO prediction is implemented as a part of the APCC MME prediction system.