I’ve moved to WordPress. This post can now be found at A Secondary (Repeated) ENSO Signal?##################
CORRECTION: In agreement with my post The Atlantic Multidecadal Oscillation - Correcting My Mistake, I edited the sentence in this post that read, “The disparity between the North Atlantic SST anomaly trend, Figure 2, and the rest of the subsets was striking, which is one of the reasons why North Atlantic SST is normally illustrated as a residual, the Atlantic Multidecadal Oscillation.” The AMO is not a residual. It is detrended North Atlantic SST anomalies, according to the NOAA ESRL. I therefore deleted, “which is one of the reasons why North Atlantic SST is normally illustrated as a residual, the Atlantic Multidecadal Oscillation.”
In the following, I’ve removed the Atlantic Ocean SST anomalies from the global SST anomalies, then removed the effects of volcanic aerosols and ENSO. The result was an unexpected secondary or repeated ENSO signal. This secondary ENSO signal would appear if the Atlantic Ocean SST anomalies remained, but the Atlantic skews the trend so significantly that it might be best to evaluate the secondary ENSO without it.
This post was originally intended to illustrate the influence of the North Atlantic on recent SST anomaly trends. It gets sidetracked with the new find. Because the new phenomenon subtracts from the original intent, I’ll repeat the first part of this post in a second one.
I offer no basis for this secondary ENSO signal? Is it a result of cloud feedback? Is it caused by some effect of lingering atmospheric and oceanic Rossby and Kelvin waves initiated by the El Nino? Or is it a result of the scaling factors I’ve used?
As with many of my posts, someone with a better grasp of statistical tools could select coefficients and scaling factors that are more refined and, with those tools, could perform a better analysis. I accept that and welcome additional research into what I present here. In that respect, consider this a preview. But I ask those who do follow-up on this to present your results with simple time-series graphs. Remember, also, to cite this blog and its author.
And for those who would like to research this but have never downloaded OI.v2 SST data, I’ve included instructions for downloading the data at the end of this post.
I prepared a post on SST trends for the globe, Figure 1, and for individual ocean subsets. (That post has now been put to the side.)
The disparity between the North Atlantic SST anomaly trend, Figure 2, and the rest of the subsets was striking. The North Atlantic SST anomaly trend for the period of November 1981 (the start of the OI.v2 SST dataset) and January 2009 is ~0.264 deg C/decade, while the global trend is ~0.0948 deg C/decade. The North Atlantic trend is approximately 2.8 times the global trend, driven by Atlantic Meridional Overturning Circulation and El Ninos, (yes, El Ninos).
NOTE: El Nino-induced step changes in the North Atlantic were illustrated in the post There Are Also El Nino-Induced Step Changes In The North Atlantic. Recall, also, that the Atlantic Meridional Overturning Circulation appears to be impacted by ENSO events as well. Refer to the post titled Atlantic Meridional Overturning Circulation Data.
REMOVING THE NORTH ATLANTIC DATA
So I decided to remove the North Atlantic SST anomaly data from the global. There is no simple way to do this with a coordinate-based system such as NOMADS or KNMI Climate Explorer, so I made an assumption. The Atlantic Ocean surface area is approximately 30% of the global ocean surface area, and I assumed the North Atlantic represented 50% of that. I then scaled the North Atlantic SST anomaly data by 0.15 and subtracted it from the global SST anomalies. The resulting dataset, noted as “Global SST Anomalies Without North Atlantic,” is illustrated in Figure 3. There isn’t that much difference between the peaks and troughs of this adjusted dataset and those of the global SST anomalies. In term of visual effect, the curve appears to have been rotated, decreasing the trend to 0.0546 deg C/decade.
REMOVING THE EFFECTS OF VOLCANIC AEROSOLS
Volcanic aerosols lower the SST anomalies at the time of the eruption and for a few years afterwards, so SSTs dropped as a result of the 1982 El Chichon and 1991 Mount Pinatubo eruptions. This increased the trend over the term of the dataset. Figure 4 is a comparative graph of the “Global SST Anomalies Without North Atlantic” and inverted Sato Index of Stratospheric mean optical thickness. I’ll use the Sato Index data to remove the impacts of the volcanic eruptions.
Eyeballing it, there didn’t appear to be any need to scale the Sato Index data, so I added the Sato Index data to the “Global SST Anomalies Without North Atlantic” data. The resulting changes in the “Global SST Anomalies Without North Atlantic” data is shown in Figure 5. Note how the trend has decreased to 0.036 deg C/decade.
A NOTE ABOUT THE TRENDS
Without the North Atlantic and without the impacts of two significant volcanic eruptions, the SST trend would extend out to a rise of only 0.36 Deg C over the next 100 years. If GCMs were “tuned” without considering these effects, the rise in global temperature over the next century is bound to fall short of their projections.
BACK TO THE ADJUSTED DATA
Note also that the curve resembles NINO3.4 SST anomalies. Refer to Figure 6, which is a comparative graph of scaled NINO3.4 SST anomalies and the “Global SST Anomalies Without North Atlantic” after the effects of volcanic aerosols have been removed. They correlate well and the timing of the rises and the peaks do not appear to have any significant lags.
REMOVING THE EFFECTS OF EL NINO PRESENTS A CURIOSITY
The next obvious step was to remove the NINO3.4 SST anomaly impact from the “Global SST Anomalies Without North Atlantic and With Volcanic Aerosols Removed” data. In this step, I simply subtracted the scaled NINO3.4 SST anomalies from the other dataset. Refer to Figure 7, which is a graph of the “Global SST Anomalies Without North Atlantic” in which both the volcanic aerosols AND ENSO impacts have been removed. A curiosity resulted. A “secondary El Nino” signal appears 8 months after the November 1997 peak of the 1997/98 El Nino.
THE SCALING FACTORS I’VE USED ARE “IN THE BALLPARK” OF REALITY, BUT THIS “SECONDARY EL NINO” SIGNAL MAY SIMPLY BE A PRODUCT OF THE SCALING FACTORS AND WEIGHTING I USED. IT DOES, HOWEVER, NEED TO BE INVESTIGATED FURTHER. INCREASING THE NINO3.4 SST ANOMALY SCALING FACTOR BEFORE REMOVING ITS IMPACT FROM THE “ADJUSTED GLOBAL” DATA WILL OBVIOUSLY DECREASE THE “SECONDARY EL NINO” SIGNAL, AND VICE VERSA.
Back to the post.
A QUICK LOOK FOR THE SOURCE OF THE “SECONDARY EL NINO” SIGNAL
The global temperature anomaly mapping feature at the GISS website can be modified so that it only presents SST anomalies. GISS uses OI.v2 SST data from December 1981 to present, the same source used in this post.
I changed the base period to 1971-2000, the same as the OI.v2 SST anomaly data, set the smoothing for 250km, and selected July 1998, 8 months after the November 1997 peak of the 1997/98 El Nino. Refer to Figure 8. A hot spot appears in the vicinity of the Kuroshio Extension (32N-38N, 142E-180E).
NOTES ABOUT THE KUROSHIO EXTENSION AND THE “SECONDARY EL NINO” SIGNAL
Figure 9 is a graph of the SST anomalies for the Kuroshio Extension. As you’ll note, while the Kuroshio Extension SST anomalies are elevated (they acquire an upward step after the 1997/98 El Nino) in July 1998, they do not drop as the “secondary El Nino” signal does. In fact, the Kuroshio Extension SST anomalies do not peak until October 1999 (the spike in 1999), two years after the peak of the 1997/98 El Nino. Also, Mercator projections, and similar maps, visually distort the importance of high latitudes. So the Kuroshio Extension anomaly is a contributor to the “secondary El Nino” signal, but not the major factor. There are other areas of noticeably elevated SSTs, the Pacific Warm Pool and the South Atlantic. The Pacific Warm Pool does increase in temperature during La Ninas, but I can’t say whether the same holds true for the South Atlantic.
One of the reasons is that the 1982/83 and 1997/98 El Nino had different characteristics. The decline of the 1982/83 El Nino is flatter, taking longer, than the 1997/98 El Nino. Refer to Figure 10. The 1982/83 El Nino was also suppressed by the El Chichon eruption. (I borrowed Figure 10 from a prior post. This is why the 1986/87/88 El Nino data appears in it.)
A FINAL COMPARISON
In Figure 11, the scaled NINO3.4 SST anomalies have been lagged 8 months. The NINO3.4 data correlates well with the “Global SST Anomalies Without North Atlantic With Volcanic Aerosols AND ENSO Impacts Removed”. The step changes in the “Adjusted Global” dataset after the 1986/87/88 and 1997/98 El Nino events also stand out.
Figure 12 illustrates the “Adjusted Global” SST anomalies in which the North Atlantic and the volcanic aerosols, ENSO, and “Secondary ENSO” impacts have been removed. Fine tuning the scaling factors would impact the shape of that curve, but it would be futile for me to adjust them without cause. That’s where you the reader take over this analysis.
AS AN AFTERTHOUGHT
What remains (Figure 12) would include the effects of the North Atlantic on the remainder of the oceans. The step changes in the East Indian and West Pacific Oceans would also have to be accounted for.
DOWNLOADING OI.v2 SST ANOMALY DATA
The source of the SST anomaly data is the NOAA NOMADS system. Refer to the following link.
Under “Control File”, select “monoiv2.ctl…Monthly OIv2 SST” and under “Plot Type”, select “time series”. Click on “Next Page”.
For “Field”, select “ssta *OIV2 SST monthly anomaly (C) rel to 1971-2000. The “Initial Time” and “Final Time” are automatically set for the first and last month of the dataset. Input the desired latitudes and longitudes. (South Latitudes and West Longitudes are negative numbers and the “From” must be less than “To”.) The default latitudes and longitudes are global, obviously. Click on “Plot” and a new window will appear.
Scroll down and click on the link for the text file. The first four global SST anomalies and header should be…
data from 00Z01NOV1981 to 00Z01JAN2009
Obviously, if you’ve selected different start and end months, the header and initial data will change.
The coordinates used in this post, in addition to global, are:
North Atlantic = 0 to 65N and 78W to 20E
NINO3.4 = 5S to 5N and 170W to 120W
Kuroshio Extension = 32N to 38N and 142E to 180E
SATO INDEX DATA SOURCE
The Sato Index Data is available from GISS at: