Modeling a mechanism of El Niño initiation

Droughts and forest fires rage in Australia and Indonesia. Heavy rains cause flooding and the collapse of fisheries in Peru and Ecuador. Hurricane and cyclone activity changes. Precipitation increases in the central Pacific and southern US.

These seemingly disparate global phenomena are closely related. Each is significantly shaped by the dominant mode of interannual tropical Pacific variability, the El Niño–Southern Oscillation (ENSO). Changes in the ENSO influence global atmospheric circulation, which in turn alters global temperature and precipitation patterns. Many El Niño (the warm phase of the ENSO) climate impacts are depicted in figure 1.

Given such widespread and often destructive impacts, it is critical that we understand ENSO behavior. One highly active area of ENSO research focuses on determining how and when ENSO events are generated. Characterizing the underlying processes of ENSO initiation will ultimately improve our ability to predict with greater lead times ENSO events and thus provide ample time to prepare for the resulting climate impacts. The skill of such predictions depends on our knowledge of the complicated interactions between the many processes that influence ENSO initiation and development.

During normal conditions, the easterly trade winds in the tropical Pacific result in a buildup of warm surface waters, known as the warm pool, in the western equatorial Pacific. The boundary between warm surface water and cold deep water, called the thermocline, is far below the ocean surface. Meanwhile, a “cold tongue” of cooler water forms in the eastern equatorial Pacific. During El Niño conditions, which occur approximately every two to seven years, weakened trade winds allow the warm pool to shift eastward, generating strong warm sea-surface temperature (SST) anomalies and a deeper thermocline in the eastern Pacific. La Niña conditions occur when stronger trade winds confine the warm pool to the far western Pacific and generate cold SST anomalies in the eastern Pacific. The peak of El Niño and La Niña events typically coincides with Northern Hemisphere winter. Figure 2 depicts the state of the tropical Pacific during El Niño and normal conditions.

The air–sea interactions and ocean dynamics of ENSO variability are well understood, and this knowledge affords some predictability of the magnitude and timing of the peak of ENSO events several months in advance. However, predictions are only skillful once the ENSO process has begun. Unfortunately, our grasp of how the process begins in the first place is limited.

The initiation of the ENSO is complicated by a multitude of processes, such as the Madden–Julian oscillation, trade-wind variability, and extratropical influences. Scientists use many techniques to study how those processes interact with ENSO variability. Along with my colleagues, I used a coupled climate model, which includes interacting atmospheric and ocean components, to study the extratropical factors that contribute to the ENSO.

Our research focused on the Pacific Meridional Mode (PMM), which is characterized by anomalously warm SST in a wide swath of the North Pacific. It’s initiated by a pattern of atmospheric variability known as the North Pacific Oscillation (NPO). The PMM can then generate ENSO variability in the equatorial Pacific. Both the PMM and the ENSO also contain negative phases that have important interactions and climate impacts.

It is challenging to characterize the interactions between the ENSO and the PMM, since it is impossible to separate their individual effects through observational analysis alone. However, models can provide some clarity. The National Center for Atmospheric Research Community Earth System Model (CESM) can be configured to have a fully active atmosphere coupled to a motionless ocean, which is sufficient to simulate the PMM response to the midlatitude wintertime NPO. That’s a good start, but we need a second model that includes ocean dynamics to simulate the El Niño response to the PMM. With the help of both models, we can isolate the PMM from the ENSO and cleanly identify the mechanisms through which the PMM initiates El Niño events.

An ensemble of 40 individual CESM simulations, in which each ensemble member contains slightly perturbed initial conditions, was forced with the surface heat flux of the NPO (figure 3a). The NPO forcing generates the PMM (figure 3b), which has maximum amplitude from March through May. The PMM SST anomalies generate an atmospheric response near the equator, which can be seen in the wind vectors in figure 3b. The westerly wind anomalies are capable of exciting waves that propagate energy eastward or westward along the thermocline and influence the generation of the ENSO.

The PMM wind is then used to force the intermediate coupled model to generate ENSO events (figure 3c). Through the careful selection of specific dynamical mechanisms within the intermediate coupled model, we are able to identify the type of waves that generate ENSO events due to extratropical forcing. Our results show that extratropical forcing generates ENSO events through Kelvin waves, which oscillate up and down along the thermocline under the influence of gravity. Rossby waves, which oscillate north and south along the thermocline, have very little influence on ENSO initiation.

When we expanded the analysis to examine ENSO responses to the PMM of each individual ensemble member, the most notable feature is the very large diversity of ENSO responses. Although there is a clear tendency for the positive phase of the PMM to generate El Niño events, our research suggests the ENSO response is sensitive to natural variability, which includes the perturbations between the initial conditions of each simulation as well as the chaotic noise that develops as the system evolves. Therefore, we must further explore the impact that natural variability has on ENSO development. Research to characterize the roles of the initial conditions and chaotic noise on ENSO initiation is ongoing.

Our modeling experiments highlight a few of the challenges associated with identifying underlying ENSO dynamics. In particular, the results show how difficult it is to isolate interrelated processes and to identify how they influence ENSO initiation. Overcoming those obstacles through the creative use of models and analysis techniques will hopefully improve the skill of future ENSO predictability as well as our ability to forecast major global climate events and their associated human and economic impacts.

Erin Thomas is a PhD student working with Dan Vimont at the University of Wisconsin–Madison. She studies ENSO dynamics through the use of coupled climate models and statistical techniques. She is currently working to identify the role of stochastic forcing on ENSO characteristics.