Guest Post by Bob Tisdale
In a comment in the March 2009 SST Anomaly Update thread, Blogger DB reminded me of the Bratcher and Giese (2002) paper “Tropical Pacific Decadal Variability and Global Warming” [GEOPHYSICAL RESEARCH LETTERS, VOL. 29, NO. 19, 1918, doi:10.1029/2002GL015191, 2002].
“An analysis of ocean surface temperature records show that low frequency changes of tropical Pacific temperature lead global surface air temperature changes by about 4 years. Anomalies of tropical Pacific surface temperature are in turn preceded by subsurface temperature anomalies in the southern tropical Pacific by approximately 7 years. The results suggest that much of the decade to decade variations in global air temperature may be attributed to tropical Pacific decadal variability. The results also suggest that subsurface temperature anomalies in the southern tropical Pacific can be used as a predictor for decadal variations of global surface air temperature. Since the southern tropical Pacific temperature shows a distinct cooling over the last 8 years, the possibility exists that the warming trend in global surface air temperature observed since the late 1970’s may soon weaken.”
Link to GRL Abstract:
Also refer to the copy of the Bratcher and Giese slide presentation:http://www.decvar.org/documents/CCR_workshop/bratcher.htm?PHPSESSID=df81bed52419c895efe9135099fb26e9
And theWorldClimateReport post on the study here:http://www.worldclimatereport.com/archive/previous_issues/vol8/v8n04/feature1.htm
And CO2Science did a write up here:http://www.co2science.org/articles/V6/N20/C1.php
As illustrated in Figure 1 (Figure 1 of Bratcher and Giese 2002), their comparison of Global Temperature Anomaly (GISTEMP) and NINO3 SST Anomaly (Simple Ocean Data Assimilation-SODA) ran from 1948 to 2000. The graph appears to be of annual (not monthly) data, with 5-year smoothing as discussed in the paper. The two questions that struck me were: How far back in time do the two datasets coincide and what would the updated graph look like? (The latter question was also part of DB’s comments.)
The problem: the easily available SODA data through the KNMI Climate Explorer website only includes the years of 1958 to 2004. So I’ve substituted ERSST.v3b data in place of the SODA data. I’ve also used monthly instead of annual data.
UPDATING THE GLOBAL TEMPERATURE AND NINO3 SST ANOMALY COMPARISON
Figure 2 illustrates a reasonable facsimile of Cell A of the Bratcher and Giese comparison of Global Surface Temperature and NINO3 SST anomalies. It has been extended in time through February 2009. The global surface temperature data is GISTEMP and the NINO3 SST anomaly data is ERSST.v3b. Bratcher and Giese appear to use different base years than the ones used here (1971-2000), and, of course, the variability appears greater with the monthly data.
Figure 3 updates Cell B of the Bratcher and Giese Figure 1. Without accounting for volcanic aerosols, any attempt to determine the 4-year lag between NINO3 and Global Surface Temperatures as claimed by Bratcher and Giese would be difficult with these time-series graphs, including Figure 1. However, Global Temperatures do appear to respond gradually over time to the shift in NINO3 SST anomaly. Note the impact of the Pacific Climate Shift of 1976 on the smoothed NINO3 SST anomalies. It stands out in Figures 1, 2, and 3.
Based on the averages of the smoothed NINO3 SST anomalies for the periods of 1950 through 1975 and of 1978 to present, Figure 4, the magnitude of the 1976 shift in NINO3 SST anomalies is more than 0.4 Deg C. If one considers NINO3 SST anomalies as a forcing, then the rise of global temperatures from 1976 to the early 2000s would then appear to be a natural response to a natural variation.
EXTENDING THE COMPARISON BACK IN TIME
Figure 5 illustrates the NINO3 SST and global surface temperature anomalies from 1880 to present. Both datasets are smoothed with 61-month (5-year), running-average filters. Two things to consider when looking at the data before 1948: the discontinuity in the SST anomalies at 1945 would have impacted both datasets, and prior to 1914 and the opening of the Panama Canal, there were very few SST samples in the NINO regions. With those in mind, the fact the two datasets do seem to “track” is quite remarkable.
The NINO3 data was still noisy with the 61-month smoothing, so I changed to a 121-month filter in Figure 6. This seems to aid in illustrating the influence of NINO3 SST anomalies on global surface temperature. ENSO appears to dictate whether global surface temperatures rise or fall over decadal periods. It also illustrates a gradual “ramp up” required to overcome global thermal inertia.
There’s a mismatch with those two datasets. The NINO3 SST anomalies are based on ERSST.v3b data, while GISS uses HADSST data prior to November 1981 for their global surface temperature product. So let’s look at matching data.
NINO3 AND GLOBAL SST ANOMALIES USING ERSST.v3b DATA
In Figure 7, the GISS Global Surface Temperature data has been replaced by ERSST.v3b Global SST Anomaly data. The most significant difference between the GISTEMP Global Surface Temperature and the ERSST.v3b SST data can be found between ~1880 and 1900. Note how the drop in the temperature from 1880 to 1900 is exaggerated in the ERSST.v3b Global SST anomaly data.
Note also how the effect of NINO3 SST anomalies on Global SST anomalies is still clear after 1914. This is especially true following the shift in NINO3 SST anomalies in 1976.
THE SECOND CONCLUSION OF BRATCHER AND GIESE 2002
As noted earlier, Bratcher and Giese wrote in the Abstract, “Anomalies of tropical Pacific surface temperature are in turn preceded by subsurface temperature anomalies in the southern tropical Pacific by approximately 7 years.” They illustrated this lag in their Figure 3, my Figure 8.
In the body of the paper, Bratcher and Giese discuss and illustrate the lag between the subsurface Tropical Pacific temperature and NINO3 SST anomalies. I am not trying to undermine that in any way. But the second problem I encountered while trying to update the Bratcher and Giese 2002 paper was the availability of Subsurface Temperature data for the Tropical Pacific Ocean. Simple Ocean Data Assimilation (SODA) data through KNMI does not include the subsurface temperature data.
It was my original intent to end the comparisons here, which is why the graphs are titled “Revisiting Part of Bratcher & Giese 2002”, but then it struck me that this would be a good time to illustrate a possible influence of the Southern Ocean on ENSO.
COMPARING SST ANOMALIES FOR THE SOUTHEAST PACIFIC ACC AND NINO3
Figure 9 shows the locations of NINO3 region and the portion of the Antarctic Circumpolar Current (ACC) in the extreme Southeast Pacific used in the following comparison. Note how the Humboldt Current carries waters from the ACC along the coasts of Chile and Peru and up to the eastern equatorial Pacific. The SST anomalies of the Southeast Pacific ACC should have an influence on NINO3 SST anomalies.
Figure 10 is a comparison of NINO3 and Southeast Pacific ACC SST anomalies. Again, both datasets have been smoothed with 121-month filters. From 1940 to present, there is a reasonable agreement between the two datasets, indicating that the underlying SST for the equatorial Pacific is impacted by the ACC and Southern Ocean SST anomalies. There does not appear to be the 7-year lag suggested by Bratcher and Giese, though.
I found the correlation between the Southeast Pacific ACC and NINO3 SST anomalies interesting, but not conclusive. And I have no explanation for the divergence between the two datasets from ~1915 to 1945. Did the North Pacific have a greater influence during those times? I can’t say. I’ll have to investigate that and the SST anomalies along the Humboldt Current in a future post to try to determine the reason for the disagreement during that period.
YET ANOTHER POST ABOUT GLOBAL TEMPERATURE RESPONSE TO ENSO
In my series of posts “Can El Nino Events Explain All of the Warming Since 1976?” I illustrated the processes that cause step changes in the East Indian and West Pacific SST anomalies, which in turn result in increased global SST anomalies. Refer to:
I illustrated the similar impacts of significant ENSO events on the North Atlantic Ocean in There Are Also El Nino-Induced Step Changes In The North Atlantic.
I’ve shown how Global Surface Temperature time-series data can be replicated using natural variables in Reproducing Global Temperature Anomalies With Natural Forcings. In that post, a running total of NINO3.4 SST anomalies establishes the underlying curve.
Based on the findings of Bratcher and Giese 2002, this post presents yet another way to illustrate that ENSO dictates long-term Global Surface Temperature Anomalies.
CLOSING REMARKS – THE BIG IFS
-IF the 5-year smoothing used by Bratcher and Giese (or the 61- and 121-month smoothing that I used) reflects the underlying NINO SST anomalies, and
-IF global temperatures do respond as implied by the correlation of the NINO3 SST anomalies and Global Surface Temperature anomalies as shown in the preceding and as discussed in Bratcher and Giese 2002, and
-IF NINO3 SST anomalies continue to follow the Southern Ocean and ACC SST anomalies,
-THEN Global Surface Temperatures should continue to decrease in response.
Will these natural variations overwhelm any anthropogenic sources of warming and drive global temperatures down, as opposed to only flattening the curve as it has recently and as it had from the 1940s to the late 1970s? Only time will tell.
The GISS Global Surface Temperature anomaly data and the ERSST.v3b SST anomaly data are available through the KNMI Climate Explorer website: