Importance of mean state in simulating different types of El Niño revealed by SNU coupled GCMs

Highlights

• Mechanisms that control relative frequency of two-types of El Nino events are investigated.

• Mean equatorial thermocline structure is related to the occurrences of WP El Niño and CT El Niño.

• The eastern Pacific dryness results in weak CT El Niño and reduction in CT El Niño occurrence.

Abstract

Recent studies suggest that there are two types of El Niño events, which differ in terms of zonal distribution of sea-surface temperature (SST) anomalies. In this study, we investigate mechanisms in controlling simulation of two-types of El Nino using three different versions of the Seoul National University (SNU) air-sea coupled general circulation models.

The occurrences of two types of El Niño are related to the simulated climatological SST over the eastern Pacific. It is found that a model with relatively less canonical (or frequent Warm Pool or Central Pacific) El Niño occurrence has colder cold tongue in the equatorial central Pacific. Due to the cold mean SST and associated dryness over the eastern Pacific, positive El Niño-related SST anomalies over the eastern Pacific cannot trigger local convection effectively. The eastern Pacific dryness leads to the confinement of anomalous convective activity in the western Pacific, which results in weak canonical El Niño and reduction in canonical El Niño occurrence. Instead, the confined convective activity in the western Pacific can lead to strong SST anomalies over western-central Pacific through local air-sea interaction, which can increase Warm Pool (WP) El Niño occurrence. In addition, we found that mean equatorial thermocline structure is also related to the occurrences of WP El Niño and canonical El Niño that is, a model with deep thermocline depth simulates less WP El Niño occurrence, consistent with Yeh et al. (2009).

Introduction

Typical El Niño is referred to a phenomenon characterized by a positive sea-surface temperature (SST) anomaly over the central-eastern Pacific from its mean condition. The index for El Niño was often defined by SST anomalies over NINO3.4 (170°W–120°W, 5°S–5°N) or NINO3 (150°W–90°W, 5°S–5°N) regions, because it is found that maximum positive SST anomalies were located over the eastern Pacific. However, several recent studies have argued that there exist more than one type of El Niño, which has positive maximum SST anomalies over the central Pacific (Trenberth and Stepaniak, 2001, Larkin and Harrison, 2005a, Larkin and Harrison, 2005b, Ashok et al., 2007, Weng et al., 2007, Guan and Nigam, 2008, Kao and Yu, 2009, Kug et al., 2009a, Kug et al., 2010a, accepted for publication; Holland, 2009). Although there is no consensus in terms of terminology so far, it is commonly agreed that the “new type” El Niño has positive SST anomalies located in the central Pacific.

Kug et al. (2009a) showed that the new type of El Niño named “warm-pool (WP) El Niño” is different from the conventional El Niño named “cold-tongue (CT) El Niño”, not only in terms of the action center but also in terms of dominant mechanism. Among two important feedback processes associated with the El Niño dynamics, the thermocline feedback is a key process for the CT El Niño (Schopf and Suarez, 1988, Jin, 1997a, Jin, 1997b, Jin and An, 1999, Kang et al., 2001), whereas the zonal advective feedback is a dominant process for the WP El Niño (Kug et al., 2009a). Furthermore, due to the different SST centers, two types of El Niño have also differences in terms of teleconnection and global impact (Ashok et al., 2007, Weng et al., 2007, Weng et al., 2009, Kim et al., 2009, Kug et al., 2009b, Yeh et al., 2009, Kug et al., 2010b). For example, Kim et al. (2009) mentioned that two distinctly different forms of tropical Pacific Ocean warming are shown to have substantially different impacts on the frequency and tracks of North Atlantic tropical cyclones.

Several studies reported that the WP El Niño events occur more frequently in recent decades and they are associated with the global warming (Larkin and Harrison, 2005a, Larkin and Harrison, 2005b, Ashok et al., 2007, Yeh et al., 2009, Lee and McPhaden, 2010). Yeh et al., (2009) showed that new type of El Niño like WP El Niño occurs more frequently during recent decades using both observations and multi-model outputs under global warming scenarios from the Coupled Model Intercomparison Project phase-3 (CMIP3). Kim and Yu (2012) supported this argument using the CMIP5 models.

As the growth of the canonical El Niño is affected by the changes in basic states (An and Wang, 2000, Fedorov and Philander, 2000, Fedorov and Philander, 2001, Meehl et al., 2001; Vecchi and Wittenberg 2010), it is also believed that the occurrence of the WP El Niño events is determined by the mean states. Yeh et al. (2009) suggested that the shoaling thermocline under global warming would be associated with an increased frequency of the WP El Niño. Choi et al. (2010) found that the higher occurrence decades of the WP-type El Niño within a long integration of Geophysical Fluid Dynamics Laboratory (GFDL) CM2.1 is associated with a strong zonal gradient of mean surface temperature in the equatorial Pacific, and a deep mixed layer and deep thermocline depth in the western-central Pacific, along with a strong equatorial trade wind. It implies that the change of climatological mean state can modulate the occurrence frequency of the new type of El Niño.

The importance of basic state related to different type of El Niño was also emphasized using simple model framework (Bejarano and Jin, 2008). Bejarano and Jin (2008) found that two dominant modes related to ENSO using Cane-Zebiak-type model; One of them is referred to as the quasi-quadrennial (QQ) mode, and the other is referred to as the quasi-biennial (QB) mode. The characteristics of QQ (QB) mode are similar to the CT (WP) El Niño in terms of SST structures and underlying dynamics. For example, they showed that the positive peak of QQ-mode El Niño is over the far-eastern Pacific, while that of QB-mode El Niño is over the central Pacific. According to their study, the QB mode becomes dominant when the mean easterly winds are stronger, or when the mean upper-layer thickness of the ocean is shallower. In addition, they showed that the thermocline feedback is a crucial factor for the QQ mode, whereas the role of zonal advective feedback is dominant in the QB mode.

Therefore, the simulated occurrence of the two types of El Niño in climate models may be different from the observed, due to different climatological mean state of the model. This is an important issue to measure the ability of a model in simulating two types of El Niño, because it can directly affect a model’s ability to predict the different patterns of anomalous SST (Hendon et al., 2009, Jin, 2009). Jin (2009) pointed out that the independency between two types of the El Nino events in forecast results using NCEP CFS and FRCGC/SINTEX-F models is not robust as the observed, and this is related to the degradation of ability to predicting WP-type El Niño. However, Jin (2009) does not explain which factors can cause the single type of the El Nino events in forecast models. Similarly, Ham and Kug (2011) showed that only some models participated in the CMIP3 archive can simulate two types of El Niño events to some extent, while most climate models tend to simulate single type of the El Niño.

In this study, we investigate the occurrence of CT and WP El Niño events simulated by three different versions of the Seoul National University (SNU) coupled general circulation models (CGCMs) to better understand mechanisms that control relative frequency of one type or the other in CGCMs. Given the importance of the tropical mean state in the simulation of El Niño events, we focus on the tropical mean state in the three simulations and its relationship to two types of El Niño. Section 2 describes the observation data and the CGCMs used in this study. In Section 3, we show the CT and WP El Niño events simulated by different versions of SNU CGCMs. Basic states in the three versions of CGCM are depicted in Section 4. Summary and discussions are given in Section 5.