The North Pacific pacemaker effect on historical ENSO and its mechanisms
The following summarizes work published in Journal of Climate.
A recorded summary of the major results can be found under the Presentations tab.
Major Results:
We run a North Pacific sea surface temperature (SST) pacemaker experiment and show that North Pacific SSTs account for ~15% of wintertime ENSO variability in the historical record.
This connections stems from two physically distinct mechanisms associated with the Pacific Meridional Mode.
Among individual events, our pacemaker most closely reproduces the “failed” 2014-2015 El Niño and the extreme 2015-2016 El Niño.
Figure 1 North Pacific Ocean–Global Atmosphere (nPOGA) experimental design. SSTs are restored to the model climatology plus observed historical anomaly in the Pacific Ocean north of 15˚N (solid black line). There is a 5… linearly decreasing buffer zone from 15˚N to 10˚N (dashed black line). Colored shading represents HadISST SSTAs regressed on the PMM SST index. Shaded boxes indicate different Niño indices used in this study.
Past studies have shown that North Pacific ocean-atmosphere variability in the form of the Pacific Meridional Mode (PMM) can significantly modulate the timing and magnitude of ENSO events. However, few studies have investigated this connection in the observational record.
We investigate the influence of North Pacific SSTs on observed ENSO variability by producing a 10-member ensemble of North Pacific SST pacemaker experiments, referred to as North Pacific-Global Atmosphere (nPOGA) experiments (Fig. 1). Specifically, we force North Pacific SSTs in the model to follow observations from 1950-2016. Everywhere outside of the North Pacific is completely free to evolve according to the model physics.
The ensemble mean of our experiment represents the influence of North Pacific SSTs on the rest of the climate system.
We compare our nPOGA results to an ensemble of uncoupled runs where we force an atmospheric model with the same North Pacific SSTs and everywhere is set to a seasonal cycle.
Figure 2 (a) Nino3.4 index in observations (red) and the ensemble mean of nPOGA (black). (b) Seasonal correlation of observed and nPOGA Nino indices. X-axis denotes center month of three month seasonal average. Shading in (a) and errobars in (b) represent two standard errors.
Figure 3 Hovmöller diagram of SST anomalies averaged 3˚S-3˚N in the North Pacific in (a) the ensemble mean of nPOGA and (b) observations for January 2014-December 2016.
Our pacemaker simulation significantly reproduces observed ENSO variability in boreal fall and winter, with peak correlations of 0.4 (Fig. 1).
We are able to reproduce observed SST variability more accurately and over a longer portion of the year than in the Nino3 region, suggesting North Pacific SSTs may be a better predictor Central Pacific ENSO events than Eastern Pacific events (Fig. 1b).
Among individual ENSO events, nPOGA most accurately captures the onset, timing, and relative magnitude of the “failed” 2014-2015 El Niño and the extreme 2015-2016 El Niño.
Figure 4 Composite of (a) SST (shading) and precipitation (contours, green = wetter) anomalies and (b) SST, sea level pressure (SLP; contours; purple = negative), and surface wind anomalies (quivers) across 17 positive PMM events from 1950-2016 in the ensemble mean of our uncoupled atmospheric model experiments. These simulations consist of 10-members, so we have 170 independent samples. The SST anomalies are prescribed north of solid black line (15˚N), and everywhere else is a seasonal cycle (e.g., no anomalies), which is why it is white.
Our simulations reproduce observed ENSO as a result of two physically distinct mechanisms related to the PMM.
First, Wind-Evaporation feedback propagates surface anomalies southwestward from the North Pacific onto the equator, increasing the likelihood of westerly wind events around the dateline. See my review paper on the PMM and ENSO for more details on this mechanism.
Second, our uncoupled experiments reveal that the seasonally migrating ITCZ is sensitive to PMM-related SST anomalies in the North Pacific during boreal summer.
On average, positive PMM events tend to shift the ITCZ northward (Fig. 4a), which induced a large-scale atmospheric circulation response due to an increase in upper level latent heat (Fig. 4b).
This process, referred to as the Summer Deep Convection (SDC) response, significantly impacts the frequency and intensity of zonal wind anomalies near the dateline, which can then influence ENSO through equatorial ocean wave dynamics.
In particular, the extreme 2015-2016 El Niño was preceded in spring 2015 by the warmest PMM event on record. Our simulations suggest that this extreme PMM generated a large SDC response, which likely contributed to the intensity of the 2015-2016 El Niño.