The El Niño-Southern Oscillation (ENSO) is one of the most striking manifestations of interannual variability in the tropical Pacific, and it has been studied for several decades. Understanding the changes in its characteristics is still an important issue for worldwide environmental and socioeconomic interests ( McPhaden et al., 2006; Ashok and Yamagata, 2009). Our ability to predict ENSO has improved significantly over the last several decades ( Latif et al., 1998; Jin et al., 2008) due to better observations ( McPhaden et al., 1998) and modeling studies ( Delecluse et al., 1998). However, our ability to simulate and predict ENSO is still far from perfect ( Barnston et al., 2012). Due to the importance of tropical air-sea interactions on the global climate, the interannual variability in the tropical Pacific has become one of the most critical metrics for evaluating the performance of coupled general circulation models (CGCMs). A number of studies have been carried out to evaluate the abilities of CGCMs to simulate the ENSO variability (e.g., Latif et al., 2001; Guilyardi, 2004; Yu et al., 2013). In general, simulated ENSO events can be qualitatively explained by the main or classic ENSO mechanism: the Bjerknes positive feedback between the sea surface temperature (SST) and surface wind ( Bjerknes, 1969), and the ocean dynamics that provide the delayed negative feedbacks ( Battisti and Hirst, 1989).

However, most coupled models fail to realistically reproduce the interannual variability over the equatorial Pacific, and the ENSO events simulated by most coupled models show a higher frequency than observed, as well as a false ENSO phase-locking ( Yu et al., 2013). The main purpose of the present study is to evaluate the basic performance of the Max Planck Institute (MPI) climate model (ECHAM5/MPI-OM), the atmospheric model ECHAM5 (which fuses the EC for European Center and HAM for Hamburg) coupled to the MPI ocean model (MPI-OM), in terms of its ability to reproduce the seasonal cycle and interannual variability in the tropical oceans. A 400-year control simulation with pre-industrial greenhouse gases is analyzed, and the simulation of key oceanic features, such as SST, is compared with observations.

In this study, the MPI coupled model ECHAM5/MPI-OM is employed. The atmosphere model (ECHAM5.4) is run at T63 spectral resolution (1.875° × 1.875°) with 31 vertical (hybrid) levels. The ocean model (MPI-OM) has an average horizontal grid spacing of 1.5° with 40 unevenly spaced vertical levels. Technical details of the ocean model, MPI-OM, the embedded sea ice model, and the parameterizations that have been implemented during the transition from the Hamburg Ocean Primitive Equation (HOPE) model ( Wolff et al., 1997) to the MPI-OM model can be found in Marsland et al. (2003). The atmosphere and ocean are coupled by means of the Ocean-Atmosp-here-Sea Ice-Soil (OASIS) coupler ( Valcke et al., 2003). The model does not require flux adjustment to maintain a stable climate; it simulates the mean state. Mean deviation from the observation in SST is less than 1 K over much of the tropical Pacific ( Jungclaus et al., 2006). The model has been used in climate predictability and prediction studies (e.g., Keenlyside et al., 2008).

A 400-year pre-industrial climate control simulation is presented. The analysis concentrates on key climate variables, such as SST over the tropical Pacific, which is compared with observations. The SST data are taken from the Hadley Center Sea Ice and Sea Surface Temperature dataset version 1.1 (HadISST 1.1), which is an EOF- based reconstruction of observations extending from 1871 to 2010 ( Rayner et al., 2003), and are provided by the Met Office Hadley Centre (http://www.metoffice.gov.uk/hadobs/ hadisst/). The HadISST 1.1 dataset was also used to vali-date the performance of ECHAM5/MPI-OM by Jungclaus et al. (2006).

The coupled model has been run with no flux correction for 400 years, and the model‘s climate remains stable. The details of the model‘s global climate simulation are further described by Jungclaus et al. (2006). In this section, the performance of ECHAM5/MPI-OM in presenting the tropical variability in a 400-year simulation is validated by the observed SST field. The spatial pattern of interannual SST variability is portrayed by the correlation between the anomalous Niño3.4 (averaged over (5°S- 5°N, 170°-120°W)) SST and the global SST field (Fig. 1). The pattern of SST variability is in good agreement with observations (HadISST), both in structure and strength. Notable discrepancies are that SST variability in the Pacific is too equatorially confined and extends across the warm pool. The latter is consistent with the extension of the model‘s equatorial SST cold bias into the warm pool. The variability in the Indian and Atlantic oceans is quite realistic, except for the positive correlations (above 0.3) in the equatorial Atlantic region; a weak negative correlation is seen in observations ( Slutz et al., 1985). Experiments with other CGCMs indicate that atmospheric resolution is important in shaping tropical Pacific interannual SST variability ( Gualdi et al., 2003). The intercomparison study by van Oldenborgh et al. (2005) of 20 IPCC (Intergovernmental Panel on Climate Change) models confirmed that the ECHAM5/MPI-OM coupled model pro-duces relatively realistic ENSO variability. Given the ocean‘s coarse meridional resolution in the tropics, these results support the idea that the atmosphere plays a dominant role in determining important characteristics of interannual SST variability ( Guilyardi et al., 2004).

At the same time, the model realistically simulates the ENSO variability, and the strength of the tropical Pacific SST variability compares well to observations. Figure 2 compares the standard deviations of the SST anomalies from ECHAM5/MPI-OM and from observations. The standard deviation is calculated using monthly data. The amplitude of the SST anomaly produced by ECHAM5/ MPI-OM compares well with the observation in the eastern equatorial Pacific, but its extent into the western Pacific is a little too far. ECHAM5/MPI-OM underestimates the standard deviation of SST anomalies along the eastern coastline. In general, the structure over the tropical Pacific is quite realistic. It is apparent from the Niño3.4 time series (not shown) that the standard deviation of the 400-year simulated Niño3.4 SST anomalies is 1.1°C, whereas the standard deviation of the observed SST anomalies is between 0.8°C and 0.9°C, depending on the period considered. The spectrum of the simulated Niño3.4 SST variability has two peaks at 4.9 years and 3.3 years, comparing well with the observed spectrum (Fig. 3).

As shown above, the ENSO variability is well represented and realistically simulated by the ECHAM5/ MPI-OM model. However, when adopting the coupled model to make a realistic ENSO prediction, there are two issues that should be further examined to check whether the coupled model is suited for realistic forecasting. One issue is the annual cycle of equatorial SST, and the other is the phase-locking of ENSO events. As demonstrated in Latif et al. (2001), a realistic simulation of the mean climate and annual cycle of the tropical Pacific should be a prerequisite for a good ENSO simulation. The model‘s comparatively good simulation of ENSO variability indicates that the model realistically captures air-sea interactions in the tropical Pacific. The model‘s annual cycle of equatorial SST is characterized by a weak semi-annual cycle in the west, and a strong westward-propagating annual cycle in the east (Fig. 4). In the west, the phase and strength match observations well. In the east, the

Figure 1

	Figure Option ViewDownloadNew Window
	Figure 1 Correlations between the SST averaged over the Niño3.4 region and the global SST for (a) the HadISST data and (b) the ECHAM5/MPI-OM simulation. Regions where the correlation exceeds 0.3 or -0.3 are shaded in color.

Figure 2

	Figure Option ViewDownloadNew Window
	Figure 2 The spatial patterns of standard deviations of the interannual SST anomalies over the tropical Pacific for (a) the HadISST data and (b) the ECHAM5/MPI-OM simulation. The contour interval is 0.2°C.

Figure 3

	Figure Option ViewDownloadNew Window
	Figure 3 Power spectra from observed and simulated Niño 3.4 SST anomalies. Observations are from HadISST for the period 1871 to 2010. Simulated data are from the total 400-yr period. All data are detrended and normalized by their standard deviation.

Figure 4

	Figure Option ViewDownloadNew Window
	Figure 4 Annual cycle (°C) along the equator (2°N-2°S); deviations from the annual means for (a) the HadISST data and (b) the ECHAM5/MPI-OM simulation.

strength is well simulated but the positive phase lags observations by one to two months, and the negative phase slightly leads observations. This is an improvement on an earlier version of the model ( Keenlyside et al., 2005) and resembles the better-performing models of the El Niño simulation intercomparison project ( ENSIP; Latif et al., 2001). However, this improvement is not due to the shear correction and is more likely related to changes in the mean state, which were much larger between the old and new versions of the model. The link between the mean state and the annual cycle has been demonstrated in earlier modeling studies (e.g., Li and Philander, 1996).

Also, the phase-locking of ENSO events simulated by the coupled model is similar to the pronounced observation. Niño3.4 indices of the SST anomaly for the ECHAM5/ MPI-OM simulation results and 78 consecutive model El Niño events from a total 400-year simulation are shown in Fig. 5. The El Niño events are selected according to the criteria that warm amplitudes of three consecutive months centered on the month of maximum warming exceeds 0.5°C. The composite analysis of El Niño events shows the type of El Niño phase-locking: locked to boreal winter, with the warming starting from the previous September and reaching its peak during the previous December to the next January, and then terminating in June. The simulated El Niño events, with the peak phase-locked to boreal winter, resemble the observations well ( Guilyardi et al., 2004).

We have discussed the main ocean-related results from a 400-year-long control integration with the MPI coupled climate model ECHAM5/MPI-OM. The simulation of SST over the tropical Pacific is shown to be stable and realistic. The pattern of SST variability is in good agree-ment with observations (HadISST), in both structure and strength. Similarly, the model can realistically simulate the ENSO variability, and the strength of the tropical Pacific SST variability compares well with observations. The ENSO period is also realistically simulated, with two dominant periods of three and five years. Furthermore, two phenomena (the annual cycle of the SST and the phase-locking of ENSO), which are significant in affecting the model‘s performance of realistic ENSO prediction, are further validated to be well reproduced by ECHAM5/MPI-OM.

In this paper, the performance of the ECHAM5/MPI-OM climate model in ENSO variability is only validated in a long-term control run, and the model is approved to be suitable for making a realistic ENSO forecast ( Keenlyside et al., 2005; Jungclaus et al., 2006). However, the ability of ECHAM5/MPI-OM to produce a realistic ENSO prediction should be further validated and explored, and many studies are still taking place. For example, the realistic atmospheric boundary forcing field since the 1950s should be adopted when making a realistic ENSO forecast. Furthermore, a more comprehensive initialization scheme and more available observations can improve the ability of the coupled model to produce a realistic ENSO prediction ( Zheng et al., 2006, 2007, 2009; Zheng and Zhu, 2010). These can also can minimize the initial errors and alleviate the ‘spring predictability barrier‘ of the model ( Duan et al., 2009; Duan and Wei, 2012).

Figure 5

Figure Option ViewDownloadNew Window

Figure 5 Niño3.4 indices of SST anomaly for the ECHAM5/MPI-OM simulation results, and 78 consecutive model El Niño events from a total 400-yr simulation are included here. Dashed black curves are aligned based on the year of the peak warm phase, and the thick red curve represents the average of these warm events. Warm events are selected according to the criteria that warm amplitudes of three consecutive months centered on the month of maximum warming exceeds 0.5°C.

Acknowledgments. This work was supported by the National Program for Support of Top-notch Young Professionals, the National Basic Research Program of China (Grant Nos. 2012CB955202 and 2012CB417404), ‘Western Pacific Ocean System: Structure, Dynamics, and Consequences‘ of the Chinese Academy Sciences (WPOS; Grant No. XDA10010405), and the National Natural Science Foundation of China (Grant No. 41176014).