Annual Nino3.4 Values = +2, -1.5, +1.0, -1.5
Annual Nino3.4 Running Total = +2, +0.5, +1.5, 0
But the intensity and frequency of positive and negative ENSO events are not equal and preparing the running total of the data created a curve that mimicked global temperature anomaly.
There were 8 El Ninos years, but only 3 La Ninas from 1976 to 1997. And from 1950 to 1975, there were 11 La Ninas and only 7 El Ninos. Reference:
Is this one of the reasons why global temperatures fell gradually from the 50s to the late 70s, then rose so quickly afterwards?
In Figure 1, I ran through the same process of creating running total graphs of the raw NINO3.4 data described in the prior post, but this time using monthly data.
Again, in order to make it comparable to global temperature anomaly, I needed to scale it, this time with a coefficient of 0.007. I also shifted the data by -1.7 deg C to create Figure 2.
Both data were smoothed with a 15-month running average filter, to reduce the noise in the global temperature anomaly data in the following Figure 3.
Note about the absence of year-to-year perturbations in the NINO3.4 running total: Recall that the monthly running total data has been scaled by a factor of less than 1%. What I’ve done is taken a small part of a noisy natural oscillation to simulate the global temperature anomaly curve. The raw monthly NINO3.4 data (Not a running total) versus global temperature anomaly is illustrated in the Figure 4.
ONI and MEI data only date back to 1950. I’ll use the monthly NINO3.4 running total data from 1950 to 2007 as reference. See Figure 5.
OCEANIC NINO INDEX (ONI) DATA
ONI is an ENSO Index maintained by the Climate Prediction Center. It is currently calculated from ERSST.v3 data from SST anomalies in the NINO 3.4 region. Data available here:
Figure 6 is a graph of monthly ONI data.
Applying a running total to the monthly ONI data creates a graph (Figure 7) that I originally anticipated when I used the NINO3.4 data from the Trenberth ENSO reconstruction. The negative trend from 1950 to 1976 results from the domination of La Nina episodes. The higher frequency and amplitude of El Nino episodes result in the positive trend from 1976 to 2007.
Now, if the anomaly base of the ONI data is shifted by 0.15139 deg C, the running total again mimics global temperature anomaly, as shown in Figure 8. And again, the scale is wrong.
Figure 9 illustrates the running total of monthly ONI data (scaled by a coefficient of 0.0055 and offset 1.46 deg C) compared to the monthly NINO3.4 data (scaled with a coefficient of 0.007). There is little difference between the curves.
MULTIVARIATE ENSO INDEX (MEI)
The MEI varies from other ENSO indices as it includes additional components, not just SST: sea level pressure, two surface wind components (zonal and meridional), surface air temperature, and cloudiness. These additional variables are said to do “a better job than other indices for the overall monitoring of the ENSO phenomenon, including, for instance, world-wide correlation with surface temperatures and rainfall. It is maintained by the NCDC.”
Figure 10 illustrates the raw MEI data.
Eyeballing it, it varies slightly from the other ENSO indices, but when compared with ONI and NINO3.4, MEI has a greater linear trend. Refer to Figure 11. The reason the other two indices have different trends has to be the differences in the SST data sets.
Second Order Polynomial Trends (Figure 12) also reveal that the MEI data has a higher positive amplitude during mid years than the ONI and NINO3.4 indices. This changes the curve enough to significantly alter the relationship of a running total.
Due the selected base year for the anomaly calculations, it too creates running trend with an expected two-trend curve, as illustrated in Figure 13.
The additional components of the MEI change the data to the point that it has to be shifted significantly (0.28 Deg C) to recreate a curve that resembles global temperature anomaly, as illustrated in Figure 14. And once more, the scale is wrong.
The magnitude of this shift requires such a small coefficient (0.0025) to make it correlate with the NINO3.4 curve (Figure 15) that the year-to-year ENSO perturbations are severely decreased in magnitude.
WHAT HAS ALL THIS PROVEN SO FAR?
Little, other than ENSO indices can be adjusted so that their running totals mimic global temperature anomaly.
WHAT DOES IT IMPLY?
That ENSO influences global temperature, or that global temperature influences ENSO, or a combination of both. Nothing new there. For now, it’s just a new way of looking at the interrelationship between ENSO and Global Temperature anomaly.
WHAT WAS I IN SEARCH OF WHEN I BEGAN THIS EXERCISE?
I was looking for long-term trends in NINO3.4 data. Then I stumbled on the correlation with global temperature anomaly. What’s wrong with that trend or the others the individual indices wish to divulge without my tinkering? Nothing.
Figure 16 (Same as Figure 1)
Figure 17 (Same as Figure 7)
Figure 18 (Same as Figure 13)
All three indicate that a shift occurred in 1976, which is consistent with current understanding. Following 1976, they indicate a contribution to global temperature. Between 1950 and 1976, all three indicate a deduction, to varying extents. And for the long-term analysis, it indicates an ENSO contribution between 1909 and 1941. Because the authors of the long-term index question the reliability of the data prior to 1909, we’ll exclude it from this summation.
All of that is reasonably consistent with ENSO relationship with the PDO, Figure 19, though the depiction differs.