I would title the post differently now, too.
Figure 1 is a graph of the mid-latitude South Pacific SST anomaly, smoothed with an 85-month filter. Due to the sheer size of the South Pacific, and the filtering, the 82/83 and 97/98 El Nino events appear as tiny bumps on the curve.
Dividing the basin into smaller segments (Figure 2) and graphing those areas will hopefully provide a clue. The area highlighted in blue in Figure 2 is actually a part of the South Indian Ocean.
The segmented mid-latitude South Pacific SST anomalies are illustrated as a spaghetti graph in Figure 3, where the data is color coded with the areas highlighted in Figure 2.
Many of the SST anomaly curves follow a general pattern illustrated in the basin-wide graph. Refer to Figures 4 and 5, which cover the western and central South Pacific.
Then there is the outlier. In Figure 6, the SST anomaly for the area adjacent to South America is illustrated in black. Note the four major rises in temperature, peaking in 1879, 1897, 1940, 1779-81, and 1995. Recall that this graph is smoothed with a 7-year+ filter. I wonder if predictions of major El Nino events could come from major rises in Eastern Pacific SST.
But what about the overall shape of the eastern mid-latitude South Pacific SST anomaly? Comparing it to the filtered Southern Ocean SST anomaly, Figure 7, there is a very basic similarity, very basic. But could the differences be the result of the areas involved (the Southern Ocean is larger in area and its data would, therefore, be dampened) and the result of the freezing temperature of sea water, which would limit the extent of the negative anomalies.
Figure 8 represents the same SST anomalies (Southern Ocean and Eastern Mid-Latitude South Pacific), without smoothing. The Eastern Mid-Latitude South Pacific SST anomaly data has been scaled by a factor of 0.4. The similarities between the two data sets become more apparent.
A very much smoothed ENSO signal (NINO3.4 anomaly) is compared to Eastern Mid-Latitude South Pacific SST anomaly in Figure 9. As expected, the peaks and troughs are correlated, though the overall shapes of the curves vary. Are NINO3.4 temperature trends limited by the opposing Northern and Southern Hemisphere equatorial currents?
Figure 10 again compares NINO3.4 SST and Eastern Mid-Latitude South Pacific SST anomalies, this time unsmoothed. Neither data set has been scaled. You’ll definitely need the TinyPic link to view that graph. Again, they correlate well--not perfectly, but well.
The last graph, Figure 11, compares Southern Ocean SST and NINO3.4 SST anomalies. Detailed comparisons of the Southern Ocean with other data sets are impeded by the data availability for the Southern Ocean. But in general, it’s not a bad correlation.
I did try to carry the comparisons one step farther, by comparing a portion of the Southern Ocean SST anomalies, those west of the Antarctic Peninsula, with the SST anomalies of the Eastern Mid-Latitude Southern Pacific. Unfortunately, the two anomaly data sets turned out to be identical. (SSTs were different; SST anomalies were the same.) Subtracting one anomaly set from the other resulted in a zero difference, exactly, over the entire term of the data. This indicates, at least to me, that the Southern Ocean data for that region are a calculated function of the South Pacific, due to the limitations of available SST data for the Southern Ocean. When describing their newer SST data set, ERSST.v3, Smith and Reynolds mention that the newer data employs satellite data in later years to help with the Southern Ocean calculations. It’s unfortunate; it would’ve been interesting to carry the data comparisons one more step.
Sea Surface Temperature Data is Smith and Reynolds Extended Reconstructed SST (ERSST.v2) available through the NOAA National Operational Model Archive & Distribution System (NOMADS).