The impact of ENSO periodicity on North Pacific SST variability

Abstract

The periodicity of ENSO in nature varies. Here we examine how changes in the frequency of ENSO impacts remote teleconnections in the North Pacific. The numerical experiments presented here are designed to simulate perfectly periodic ENSO in the tropical Pacific, and to enable the air–sea interaction in other regions (i.e., the North Pacific) via a simple mixed layer ocean model. The temporal evolution and spatial structure of the North Pacific SST teleconnection patterns are relatively insensitive to the frequency of ENSO, but the amplitude of the variability is sensitive. Specifically, the 2-year period ENSO experiment (P2) shows weak event-by-event consistency in the ENSO response mature pattern. This is because there is not enough time to damp the previously forced ENSO teleconnections (i.e., 1 year earlier). The 4-year period ENSO experiment (P4) has 1 year damping time before a successive ENSO event matures, so the structure of the response pattern is stably repeated. However, the event-by-event variance of anomaly magnitude, specifically responding to El Niño, is still larger than that in the 6-year ENSO experiment (P6), which has 2-year damping time between consecutive ENSO events. In addition, we tested whether the variability due to tropical remote forcing is linearly independent of the extratropical local variability. Statistical tests indicate that tropical remote forcing can constructively or destructively interfere with local variability in the North Pacific. Lastly, there is a non-linear rectification of the ENSO events that can be detected in the climatology.

Appendix

1.1 Statistical significance of spatial correlation

Here we present an additional calculation addressing the statistical significance of correlation coefficients shown in Table 3. In Table 3, the spatial correlation coefficient of the seasonal ENSO composite is calculated for entire 60-year record. The ENSO composite pattern is made from the average of 30 El Niño and La Niña events in the P2 experiment, 15 ENSO events in the P4 experiment, and 10 ENSO events in the P6 experiment. In order to support the results shown in Table 3 and exclude any sampling issues, we calculated the correlation coefficients based on a sub sampled, 36-year periods. The 36 year ENSO composite means 18 ENSO events in the P2 experiment, 9 ENSO events in the P4 experiment, and 6 ENSO events in the P6 experiment. We chose three different sub-periods for the calculation. If we assume that the entire 60 years period is from year 1 to year 60, the first sub-period (SP1) is from year 1 to year 36, the second sub-period (SP2) is from year 13 to year 48, and the third sub-period (SP3) is from year 25 to year 60. Table 5 displays the correlation results for the sub-periods. Comparing Table 3 with Table 5, the general trend is similar each other. In the El Niño composite, the spatial correlation between the P2 and P4 experiment is typically the highest, specifically from boreal winter to summer. In the La Niña composite, the correlation between the P4 and P6 experiment is also typically higher than the other columns. However, the details are not always similar between Tables 3 and 5. For example, Table 5 shows that the correlation in SP3 is different from those in SP1 or SP2 (e.g. P2 vs. P4 in boreal winter in the El Niño composite, P4 vs. P6 in boreal spring and summer in the La Niña composite). In summary, sub-sampled calculation of spatial correlation is generally consistent with entire period correlation although there are a few exceptions. Hence, this result strengthens the significance of Table 3.

Table 5 Spatial correlation between the ENSO experiments calculated in 145°E–125°W, 30–60°N. Compared to Table 3 where correlation was calculated for entire 60 years, Table 5 shows correlation coefficients for 36 years sub-period. In detail, “SP1” represents sub-period of year 1 to year 36, “SP2” represents sub-period of year 13 to year 48, and “SP3” represents sub-period of year 25 to year 60