I’ve moved to WordPress: http://bobtisdale.wordpress.com/

Sunday, January 9, 2011

Can Most Of The Rise In The Satellite-Era Surface Temperatures Be Explained Without Anthropogenic Greenhouse Gases?

In this post, I divide the globe (60S-60N) into two subsets and remove the linear effects of ENSO and volcanic eruptions from GISS Land-Ocean Temperature Index data since 1982. This is done using common methods. I further adjust the data to account for secondary ENSO-related processes. The Sea Surface Temperature subsets used for these adjustments are identified. The processes are briefly discussed, supported by links to past posts, and the data are presented that support the existence of these secondary effects. An additional volcanic aerosol refinement that increases the global trend is made. The bottom line is, the GISS LOTI and Reynolds OI.v2 SST data indicates that natural variables could be responsible for approximately 85% of the rise in global surface temperature since 1982. I’ll be the first to point out that I qualified my last sentence with the word “could”. This post illustrates a story presented by the data, nothing more. But this basic evaluation indicates these secondary effects of ENSO require further research.

This post continues with the two-year series of posts that basically illustrate that the effects of El Niño-Southern Oscillation (ENSO) cannot be accounted for using a single index like a commonly used SST-based dataset such as NINO3.4, or CTI, or MEI. These indices represent only the sea surface temperature of the central and eastern equatorial Pacific (that’s modified in the case of the MEI). They do not represent the process of ENSO. They do not account for the warm water that is returned to the western Pacific and redistributed during the La Niña. This post provides further evidence of those effects.

This post is long but I elected not to divide it in two. It’s 6,000 words or 13 single-spaced pages in length. It includes 32 Figures, a gif animation, and a video. So there’s a lot to digest. I tried to anticipate questions and answer them.

Many papers and blog posts that attempt to prove the existence of anthropogenic global warming remove the obvious linear effects of El Niño-Southern Oscillation (ENSO) events and of stratospheric aerosols discharged by explosive volcanic eruptions. An example is Thompson et al (2009) “Identifying Signatures of Natural Climate Variability in Time Series of Global-Mean Surface Temperature: Methodology and Insights”… http://www.atmos.colostate.edu/ao/ThompsonPapers/ThompsonWallaceJonesKennedy_JClimate2009.pdf
…and its companion paper Fyfe et al (2010), “Comparing Variability and Trends in Observed and Modelled Global-Mean Surface Temperature.”

Let’s run through the process using GISS Land-Ocean Temperature (LOTI) data. That’s their global temperature anomaly dataset with the 1200km radius smoothing. A known problem with that dataset is that GISS Deletes Arctic And Southern Ocean Sea Surface Temperature (SST) Data. Since that creates a bias, we’ll delete the GISS LOTI data where they extend land surface data (with its higher variability) out over the oceans. That is, we’ll confine the data used in this post to 60S-60N.

Someone is bound to complain that I’ve deleted the Arctic data from the GISS LOTI data and that the Arctic is warming much faster than lower latitudes. Keep in mind that the Arctic is amplifying the effects of the rise in temperature at lower latitudes. This is the basis of the concept of polar amplification. If the vast majority of the change in temperature at the lower latitudes is natural, the same would hold true for the Arctic. Regardless, these latitudes were also chosen because the effects I want to illustrate with this post are relatively easy to display using them.

Back to the data: since GISS switches sources for their Sea Surface Temperature data from HADISST to Reynolds OI.v2 data in December 1981, we’ll look at the LOTI data starting in 1982. Smith and Reynolds (2004) Improved Extended Reconstruction of SST (1854-1997)] states the following about the OI.v2 SST data: “Although the NOAA OI analysis contains some noise due to its use of different data types and bias corrections for satellite data, it is dominated by satellite data and gives a good estimate of the truth.”

The truth is a nice place to start.

And we’ll smooth the monthly data with a 13-month running-average filter to lessen noise and season variations.

Figure 1 shows monthly GISS LOTI data (60S-60N), from March 1982 to November 2010, compared to NINO3.4 SST anomalies. The NINO3.4 data represent the Sea Surface Temperature of a region in the central equatorial Pacific bound by the coordinates of 5S-5N, 170W-120W. NINO3.4 SST anomalies are a commonly used proxy for the strength and frequency of El Niño and La Niña events, also known as ENSO. (And for those new to ENSO, refer to An Introduction To ENSO, AMO, and PDO – Part 1.) Note also that the NINO3.4 data has been scaled (multiplied by a factor of 0.16) so that the rises of the two datasets are about the same during the evolution of the 1997/98 El Niño. The NINO3.4 SST anomalies have also been shifted down 0.01 deg C and moved back in time by 3 months (lagged) to align the leading edges of the two datasets at that time. (The data in the graph starts in March 1982 because of the 3-month lag in the NINO3.4 data.) I chose the 1997/98 El Niño because that event wasn’t opposed by a volcanic eruption and it was large enough to overwhelm the background noise. As you can see, the wiggles of lesser El Niño events after 2000 don’t match as well.

Figure 1

Many of the large year-to-year changes in global temperatures are removed when we subtract the scaled NINO3.4 data from the GISS Global (60S-60N) LOTI data. Refer to the “olive drab” curve in Figure 2. Since the NINO3.4 data has a negative trend since 1982, we increase the trend in the GISS LOTI data by subtracting it. Also note how the ENSO-adjusted GISS LOTI data has “flattened” after 1998. Without the volcano-related dip and rebound starting in 1991, the period from 1988 to 1998 would also be relatively flat. It appears as though the ENSO-adjusted GISS LOTI rose in two steps since 1982. Let’s remove the cooling effects of the El Chichon and Mount Pinatubo eruptions to see if that holds true. We’ll use a GISS dataset that represents Stratospheric Aerosols. (ASCII data) Like the ENSO Proxy, we’ll scale the data and lag it. The estimated range of the impact of Mount Pinatubo on Global Temperatures varies from 0.2 to 0.5 deg C, depending on the study, so we’ll use approximately 0.35 deg C to account for its effect. Visually, that scaling appears right.
Figure 2

Figure 3 illustrates the GISS Land-Ocean Temperature Index (LOTI) anomaly data with the linear effects of ENSO events and the effects of large volcanic eruptions removed. Also illustrated is the linear trend. I’ve included the linear trend line to illustrate the effect the straight line has on the appearance of the data. The trend line gives the misleading impression that there has been a constant but noisy rise in global temperatures.
Figure 3

During the discussion of Figure 2, I noted that the data appeared to flatten after 1998. The upward steps in the data can be illustrated if we present the period average temperature anomalies for the three periods of 1982 to 1987, 1988 to 1997, and 1998 to 2010.
Figure 4

The timings of those upward steps coincide with the transitions from the large El Niño to La Niña events that took place in 1988 and 1998. This can be seen in Figure 5, which includes the adjusted GISS LOTI data. The other dataset is scaled NINO3.4 SST anomalies that have been inverted (multiplied by a negative number). Figure 5 is a gif animation, and in it, the NINO3.4 data shifts up and down. That was done to show how precisely the upward steps in the adjusted GISS data coincide with ENSO transitions. The adjusted GISS data trails the NINO3.4 data by a month or two. And the scales are correct for both upward steps.
Figure 5

But how can ENSO be impacting global temperature data if we’ve subtracted the scaled NINO3.4 anomaly data from the Global (60S-60N) GISS LOTI anomalies?

The assumption made when we removed the linear effects of ENSO (discussion of Figure 1) was that La Niña events were the opposite of El Niño events. But they are not. (This is the same incorrect assumption made by papers like Thompson et al 2009). This post is very long and to adequately describe how La Niña events are not the opposite of El Niño events would make it much longer. So it will be best to provide links to earlier detailed discussions on this topic.

Refer to:
More Detail On The Multiyear Aftereffects Of ENSO – Part 1 – El Nino Events Warm The Oceans
More Detail On The Multiyear Aftereffects Of ENSO - Part 2 – La Nina Events Recharge The Heat Released By El Nino Events AND...During Major Traditional ENSO Events, Warm Water Is Redistributed Via Ocean Currents.
More Detail On The Multiyear Aftereffects Of ENSO - Part 3 – East Indian & West Pacific Oceans Can Warm In Response To Both El Nino & La Nina Events

I provide a relatively brief description in the following section.

ENSO is a process, and ENSO indices such as NINO3.4 SST anomalies, the Cold Tongue Index (CTI), or the Multivariate ENSO Index (MEI) do not account for that process.

El Niño description: A reduction in the strength of the Pacific trade winds triggers an El Niño. A number of interrelated events then take place. Huge amounts of warm water from the surface and, more importantly, from below the surface of the western tropical Pacific (the Pacific Warm Pool) slosh east during an El Niño and are spread across the surface of the central and eastern equatorial Pacific. The increased area of warm water on the surface allows the tropical Pacific Ocean to discharge more heat than normal into the atmosphere through evaporation. That, combined with the change in location of the convection, cause drastic changes in global atmospheric circulation patterns. As a result, global temperatures vary. And most parts of the globe outside of the central and eastern tropical Pacific warm during an El Niño. The changes in atmospheric circulation work their way eastward--over the Americas, the Atlantic, Europe and Africa, the Indian Ocean and Asia. Eventually, the changes reach the western Pacific, but by that time, the El Niño is transitioning to a La Niña.

Refer again to the NINO3.4 SST anomaly data in Figure 1. A La Niña event, based on the temperature values on a graph, appears to be an El Niño of the opposite sign, and for some regional responses in temperature and precipitation that is true. But as noted before the use of NINO3.4 SST anomalies or other ENSO indices does not capture the fact that ENSO is a process. Those indices fail to account for the relocation and redistribution of huge amounts of warm water.

In the description of the El Niño, I noted that huge amounts of warm water from the surface and below the surface of the West Pacific Warm Pool had sloshed east during an El Niño. What happens to all of that warm water from below the surface of the Pacific Warm Pool that had been spread across the surface of the central and eastern tropical Pacific during the El Niño? Before the El Niño, it was below the surface and not included in the measured global surface temperature anomalies. During the El Niño, some of the warm water that had been below the surface is now on the surface of the central and eastern tropical Pacific and included in the measured global temperature. In response, surface temperatures there rose. The ENSO index captures that part of the process and only that part.

But during the La Niña, what happens to the warm water? It wasn’t all “used up” by the El Niño. And what happens to all of the subsurface warm water that had shifted east during the El Niño and had remained below the surface. It doesn’t simply disappear during the La Niña. Answering those questions explains why La Niña events are not the opposite of El Niño events, and why an ENSO index does not capture the aftereffects of an ENSO event.

The leftover warm water returns to the western Pacific. This is accomplished in a few ways. One is through a phenomenon called a slow-moving Rossby Wave. This can be seen in Video 1. It illustrates global Sea Level Residuals from January 1998 to June 2001 and captures the 1998/99/00/01 La Niña in its entirety. The video was taken from the JPL video “tpglobal.mpeg”. The slow moving Rossby wave is shown as the westward moving band of elevated sea level at about 10N. Watch the effect it has on western Pacific Sea Level Residuals when it reaches there.

Video 1Link to Video 1:

The second way that the leftover warm water is carried to the Western Pacific is through a strengthening of the trade winds. During a La Niña event, trade winds strengthen above their “normal” levels and the ocean currents carry the warm water back to the west and then poleward.

Animation 1 is taken from the videos in the post La Niña Is Not The Opposite Of El Niño – The Videos. It presents the 1997/98 El Niño followed by the 1998 through 2001 La Niña. Each map represents the average Sea Surface Temperature anomalies for a 12-month period and is followed by the next 12-month period in sequence. Using 12-month averages eliminates the seasonal and weather noise. The effect is similar to smoothing data in a time-series graph with a 12-month running-average filter.

There are a number of things to note in Animation 1. First, the El Niño and La Niña events cause changes in the sea surface temperatures in the central and eastern tropical Pacific. The NINO3.4 SST anomalies used in this post are a measure of that variation in the central equatorial Pacific, and only that variation. Second, during the El Niño, note how the sea surface temperatures warm first in the Atlantic, then in the Indian Ocean, and then in the western Pacific. The warming is caused by changes in atmospheric circulation. And by the time these changes in atmospheric circulation make their way east to the western Pacific and it starts to warm there, the El Niño is transitioning to La Niña. Third, note how the sea surface temperature anomalies in the Western Pacific (and East Indian Ocean) continue to rise as the La Niña event strengthens. Fourth, note how the SST anomalies remain elevated in the East Indian and West Pacific Oceans during the entire term of the 1998/99/00/01 La Niña.
Animation 1

The Sea Surface Temperatures of the East Indian and West Pacific Oceans remain elevated during the La Nina because the stronger trade winds reduce cloud cover. The reduction in cloud cover allows more Shortwave Radiation (visible light) to provide additional warming of the tropical Pacific waters east of the Pacific Warm Pool. The ocean currents carry this sunlight-warmed water to the west and then poleward.

To help illustrate the reasons for the upward shifts in the ENSO- and Volcano-adjusted GISS LOTI data (Figure 4), let’s divide the data into two subsets split at 20N. Refer to Figure 6.

Figure 6

First we’ll look at the Northern Hemisphere GISS LOTI anomaly data, north of 20N. It has a relatively high linear trend since 1982, about 2.8 deg C/Century. Part is due to the additional variability of the North Atlantic. To compound that, these latitudes have a relatively high land surface area, and land surface temperatures vary much more than sea surface temperatures. The land surface area of the Northern Hemisphere latitudes of 20N-60N is about 45% of the total surface area, but the land surface in the tropical and Southern Hemisphere latitudes of 60S-20N is only 17%.
Figure 7

The dataset shown in Figure 7 has not been adjusted for ENSO or volcanic eruptions. Let’s correct first for ENSO, then for the volcanic eruptions, using the same methods we did for the Global (60S-60N) data. Figures 8 and 9 illustrate the interim steps and the required scaling factors, and Figure 10 illustrates the result.
Figure 8
Figure 9
Figure 10

The Northern Hemisphere data still has a relatively high trend, approximately 2.2 deg C/Century. But what causes the additional variability if we’ve removed the effects of ENSO and volcanic eruptions? The additional variations are often described as noise, but they have sources.

There is a strong ENSO-related warming of the Kuroshio-Oyashio Extension that occurs during La Niña events. This was discussed and illustrated in my recent post The ENSO-Related Variations In Kuroshio-Oyashio Extension (KOE) SST Anomalies And Their Impact On Northern Hemisphere Temperatures. That secondary warming can be used to explain a major portion of the year-to-year variability in Northern Hemisphere land and sea surface temperature. And, along with ENSO, it helps to explain nearly all of the variations in the Northern Hemisphere (20N-60N) GISS LOTI data, including the rising trend. Figure 11 illustrates the location of the KOE dataset used in this post (30N-45N, 150E-150W).
Figure 11

The GISS LOTI anomalies for much of the Northern Hemisphere warm (cool) when the Kuroshio-Oyashio Extension SST anomalies warm (cool). This can be seen in the correlation map of annual (January to December) Kuroshio-Oyashio Extension SST anomalies and annual Northern Hemisphere (0-90N) GISS LOTI data, Figure 12. Also note the correlation with the North Atlantic.
Figure 12

As mentioned above, the secondary warming of the Kuroshio-Oyashio Extension was discussed in detail in my recent post The ENSO-Related Variations In Kuroshio-Oyashio Extension (KOE) SST Anomalies And Their Impact On Northern Hemisphere Temperatures. A quick description of the process: During a La Niña event, leftover warm water from the El Niño is returned to the Western Pacific and spun poleward by the North and South Pacific gyres. Much of that warm water finds its way to the Kuroshio-Oyashio Extension, where it apparently impacts atmospheric circulation.

The agreement between the variations in KOE SST anomalies and the adjusted Northern Hemisphere GISS LOTI anomalies is shown in Figure 13. I find that match quite remarkable. The additional spike (highlighted in blue) in the KOE data that starts in 1990 is out of place. It will make itself known later in this post. The other thing to note is the scaling factor required to align the two datasets in Figure 13. The scaling factor of 0.7 is very high. We’ll discuss this later in the post.
Figure 13

Some might think the agreement between those datasets is a lucky coincidence. Of course, the agreement between the adjusted LOTI data and the unadjusted KOE data in Figure 13 is based solely on the lags and scaling factors I used. But the scaling and lags were established logically. Eyeballing the data, the scaling factors appear to be correct. And as we shall see, using the same methods, the results are very similar for the data that covers the Tropics and Southern Hemisphere.

The Southern Hemisphere and Tropics dataset includes the GISS LOTI data from 60S-20N, Figure 14. This subset has a relatively low trend, approximately 1 deg C/Century. Some of this is related to the amount of continental land mass. For these latitudes, land represents only about 17% of the surface area. The Southern Ocean (90S-60S), which is outside of the latitudes portrayed in the post, also impacts the Southern Hemisphere data. And since the Southern Ocean SST anomaly trend over this period is negative, its interaction with the Southern Hemisphere oceans lowers the trend of the dataset.
Figure 14

And again, using the same methods, we’ll adjust for ENSO, then volcanic eruptions, Figures 15 and 16, and present the results, Figure 17. Refer to Figures 15 and 16 for the scaling factors.
Figure 15
Figure 16
Figure 17

As shown in Figure 17, removing the effects of the volcanoes has once again lowered the trend, and removing the ENSO data reduced the year-to-year variations.

Now we need a dataset for these latitudes to illustrate the secondary warming due to the leftover warm water from El Niño events and use it to account for the adjusted GISS LOTI data for the latitudes of 60S-20N.

The KOE was used in the discussion of the Northern Hemisphere data, so it seems logical that a similar area exists in the South Pacific. And for this discussion, we’ll designate that area as the South Pacific Convergence Zone (SPCZ) Extension. The SPCZ Extension data will be the SST anomalies of the area east of Australia (35S-20S, 160E-150W). As shown in Figure 18, it had a relatively high SST anomaly at the peak of the 1998/99 portion of the 1998 through 2001 La Niña.
Figure 18

The SST anomalies for SPCZ Extension are shown in Figure 19.
Figure 19

Like the KOE Extension data, the SST anomalies of the SPCZ Extension warm greatly during transitions from El Niño to La Niña events and appear to shift upward at those times. Refer to Figure 20.
Figure 20

Creating the correlation map of annual (January to December) SPCZ Extension SST anomalies and annual Tropical and Southern Hemisphere (90S-20N) GISS LOTI data was eye-opening. It appears the SPCZ data is a good proxy for those areas in the western tropical Pacific and southwest Pacific that warm during La Niña events. It would also appear to show the effects those western Pacific areas have on the rest of the globe. As we can see in Figure 21, when the SPCZ Extension warms (cools) many areas throughout the tropics and Southern Hemisphere warm (cool). But as illustrated in Figure 20, the warming that occurs during La Niña events is not counteracted by the cooling during El Niño events. This causes the data to rise in steps during the La Niña events.
Figure 21

Does the correlation map indicate that the upward shifts in the SPCZ Extension data also exist in the tropical and Southern Hemisphere GISS LOTI data? My understanding of correlation maps is that they emphasize the larger events in the data, and if we refer again to Figure 20, the larger events are those that occur during these upward shifts. We can also confirm this by comparing the respective time-series graphs.

Figure 22 illustrates the adjusted GISS LOTI data for the Tropics and Southern Hemisphere north of 60S. Also shown are scaled (0.25) SPCZ Extension SST anomalies. There are minor divergences from time to time, but in general the two curves agree surprisingly well.
Figure 22

What do the curves and linear trends of the adjusted GISS LOTI data look like if the KOE and SPCZ Extension data are removed? And what happens when you combine the two results to form a global dataset with all of the adjustments? Let’s take a look. The Northern Hemisphere GISS LOTI data (20N-60N) that’s been adjusted for ENSO and volcanic aerosols and the KOE SST anomalies is shown in Figure 23. Recall the divergence circled in blue in Figure 13; that’s the cause of the significant additional dip in 1990. Other than that, this was not a bad first attempt with scaling factors. But notice how small the trend is, 0.13 deg C/Century. If that dip was removed, the trend would be even lower.
Figure 23

The Tropical and Southern Hemisphere GISS LOTI data (60S-20N) with the ENSO, Volcano, and SPCZ Extension adjustments is shown in Figure 24. The trend is basically flat. This dataset appears noisy, but look at the temperature scale. The range is only one-quarter of one used in Figure 23.
Figure 24

We can combine the Northern Hemisphere data (20N-60N) with the Tropical and Southern Hemisphere data (60S-20N) using a weighted average. (The latitudes of 20N-60N represent approximately 29% of the surface area between 60S-60N.) Figure 25 shows the result. The linear trend is basically flat at 0.06 deg C/Century. The saw-tooth pattern is interesting, but…
Figure 25

Due to the timing, the saw-tooth pattern appears to indicate that there was a lagged (repeated) volcano signal in the data. Refer to Figure 26. The reason I say repeated is that originally when the volcanic signal was removed, the Aerosol Optical Depth data was lagged 3 months and the leading edges of the data aligned well in Figures 9 and 16. The volcano signals in Figures 25 and 26, assuming those spikes are volcano signals, are lagged 9 months. The additional signal may also simply mean the Sato Mean Optical Thickness data doesn’t account perfectly for the decay of the volcano signal and that an additional adjustment is required.
Figure 26

So let’s make the secondary volcano correction, refer to Figure 27. That will raise the linear trend of the adjusted GISS LOTI data.
Figure 27

After all of the adjustments are made, there is a small trend, about 0.24 deg C/Century. Compared to the original, unadjusted data, Figure 28, the trend of the adjusted data is only about 15% of the original GISS LOTI data for 60S-60N.
Figure 28

This makes perfect sense since there is little to no evidence of an anthropogenic global warming effect on global Ocean Heat Content (OHC) data. All one needs to do is divide the global oceans into tropical and extratropical subsets per ocean basin. Then it’s relatively easy to determine that ENSO, changes in Sea Level Pressure, and AMO/AMOC are responsible for that vast majority of the rise in OHC since 1955. Refer to:
A. ENSO Dominates NODC Ocean Heat Content (0-700 Meters) Data
B. North Pacific Ocean Heat Content Shift In The Late 1980s
C. North Atlantic Ocean Heat Content (0-700 Meters) Is Governed By Natural Variables

In this post and in The ENSO-Related Variations In Kuroshio-Oyashio Extension (KOE) SST Anomalies And Their Impact On Northern Hemisphere Temperatures, we illustrated that the Kuroshio-Oyashio Extension and South Pacific Convergence Zone Extension SST anomalies rise in steps during La Niña events. Since those upward steps are clearly responses to ENSO, there should be no need to detrend those datasets.


There is a natural variable I did not account for in this post, and it is the Atlantic Multidecadal Oscillation, or AMO. I did not remove its impacts on the Northern Hemisphere data. For those new to the AMO, refer to An Introduction To ENSO, AMO, and PDO -- Part 2.

As noted in that post, RealClimate defines the Atlantic Multidecadal Oscillation (“AMO”) as, “A multidecadal (50-80 year timescale) pattern of North Atlantic ocean-atmosphere variability whose existence has been argued for based on statistical analyses of observational and proxy climate data, and coupled Atmosphere-Ocean General Circulation Model (“AOGCM”) simulations. This pattern is believed to describe some of the observed early 20th century (1920s-1930s) high-latitude Northern Hemisphere warming and some, but not all, of the high-latitude warming observed in the late 20th century. The term was introduced in a summary by Kerr (2000) of a study by Delworth and Mann (2000).”

I could have accounted for the AMO before removing the impacts of ENSO and the volcanic eruptions. But I chose to leave it in so that I could include the impact of the KOE on the North Atlantic.

As shown in Figure 29, the trend of the North Atlantic SST anomalies between 20N-60N is 70% higher than the North Pacific SST anomalies trend. By accounting for that additional “some, but not all” trend from the AMO, the scaling factor required to align the KOE dataset with the North Hemisphere data would drop.
Figure 29

The scaling factor for the Kuroshio-Oyashio Extension data in Figure 13 was 0.7. To some, it would not seem likely that the secondary warming of the KOE could raise temperatures for the Northern Hemisphere (20N-60N) that high, especially when one considers the multiplier for the SPCZ Extension was 0.25 in Figure 22.

First: Let’s consider the known effects of an El Niño event. When surface temperatures around the globe warm in response to an El Niño, most of those areas warm due to changes in atmospheric circulation. That is, they do not rise because the heat released into the atmosphere is warming the land and sea surfaces. The following is an example I often use. During an El Niño, the tropical North Atlantic warms even though it is separated from the Pacific by the Americas. The tropical North Atlantic warms during the El Niño because the El Niño causes a weakening of the North Atlantic trade winds. With the decrease in Atlantic trade wind strength there is less evaporation, and if there is less evaporation, sea surface temperatures rise. There is also less upwelling of cool water from below the surface when the trade winds weaken. This also causes sea surface temperatures to rise.

Therefore, it is through teleconnections or atmospheric bridges, not the direct transfer of heat, that the KOE would impact the areas of the Northern Hemisphere.

Second: There is a second western boundary current extension in the Northern Hemisphere, and it is the Gulf Stream Extension in the North Atlantic. For this quick discussion, we’ll define the Gulf Stream Extension by the coordinates of 35N-45N, 75W-30W. The map in Figure 30 is a correlation map and it shows that when the Gulf Stream Extension warms (cools) there are many parts of the Northern Hemisphere that warm (cool). And note that the eastern tropical Pacific is negatively correlated, indicating that these areas warm during La Niña events.
Figure 30

Scroll back up to Animation 1. It also shows the parallel warming of the Gulf Stream Extension with the KOE.

But do the SST anomalies of the Gulf Stream Extension cool during El Niño events? As shown in Figure 31, the SST anomaly variations of the Gulf Stream Extension and the Kuroshio-Oyashio Extension are very similar. Both datasets can warm significantly during La Niña events but they do not drop proportionally during El Niño events. In an earlier linked post, I described the process that causes the KOE to warm, but I have not found a paper that describes the warming of the Gulf Stream Extension at those times. Why does the Gulf Stream Extension respond differently to El Niño and La Niña events? Like the KOE, is the warm water created during an El Niño also carried north by the Gulf Stream during the following La Niña? Do the changes in atmospheric circulation caused by the La Niña add to the warming? During the La Niña, does an increase in the strength of the North Atlantic trade winds also reduce cloud cover over the tropical North Atlantic? Does the warm water created by the decrease in cloud cover and resulting increase in sunlight then get transported to the Gulf Stream Extension? There are too many unanswered questions for me to use the Gulf Stream Extension data in this post.
Figure 31

But, the parallel warming of the KOE and the Gulf Stream Extension during the transitions from El Niño to La Niña events would help to reduce the KOE scaling factor required to explain the step changes in the adjusted GISS LOTI data.

If we scale sunspot numbers so that the variations from solar minimum to maximum represent about a 0.1 deg change in temperature, and if we lag the sunspot data 6 years, it compares well visually with the adjusted GISS LOTI data. Refer to Figure 32. Someone with additional data processing tools could duplicate the steps taken in this post and confirm how well the two curves align.
Figure 32

The warm water created during the previous La Niña(s) via the increase in Downward Shortwave Radiation (visible light) fuels El Niño events. This was discussed in More Detail On The Multiyear Aftereffects Of ENSO - Part 2 – La Nina Events Recharge The Heat Released By El Nino Events AND... ...During Major Traditional ENSO Events, Warm Water Is Redistributed Via Ocean Currents.

I would not expect that what was presented in this post could be extended back in time. The Pacific climate shifted in 1976/77. In the abstract of Trenberth et al (2002), they write, “The 1976/1977 climate shift and the effects of two major volcanic eruptions in the past 2 decades are reflected in different evolution of ENSO events. At the surface, for 1979–1998 the warming in the central equatorial Pacific develops from the west and progresses eastward, while for 1950–1978 the anomalous warming begins along the coast of South America and spreads westward. The eastern Pacific south of the equator warms 4–8 months later for 1979–1998 but cools from 1950 to 1978.”

The way ENSO events interacted with the Kuroshio-Oyashsio Extension and the SPCZ Extension also appear different before and after 1979 in the correlation and regression analyses presented in that paper. Link to Trenberth et al (2002):


Most of the data used in this post are available through the KNMI Climate Explorer Monthly observations webpage. GISS LOTI is identified there in the second field under “Temperature” as “1880-now anomalies: GISS”, with the “1200km” radius smoothing. The Reynolds OI.v2 is listed under SST as “1982-now: 1° Reynolds OI v2 SST”. The coordinates used are identified in the text and/or on the graphs.

And if you want to attempt to duplicate my results but have never used the KNMI Climate Explorer, refer to the post Very Basic Introduction To The KNMI Climate Explorer for a place to start.

The dataset used to simulate the impacts of the volcanic eruptions is available through GISS:

The Sunspot data is available through the KNMI Climate Explorer Monthly climate indices webpage. Refer to the Sunspots (1749-now, SIDC) field under the heading of “Sun”.


This was a very basic attempt to approximate the effects of natural variables on global temperatures, using scaling and lags that were eye-balled. Sometimes basic things work well, and in this case, they appear to have done that. The similarities between the adjusted GISS LOTI datasets and the respective KOE and SPCZ Extension data were remarkable. While those similarities and the correlation maps do not prove the KOE and SPCZ Extension SST anomalies cause those addition rises in surface temperature, they imply that natural factors are causing the upward steps in global temperatures illustrated in Figure 4.

After some preliminary discussions, I divided the global (60S-60N) GISS LOTI data into two sections. The linear impacts of ENSO and volcanic eruptions were then removed from those subsets. The processes that cause the Sea Surface Temperatures in two parts of the Pacific to warm greatly during La Niña events were discussed. The unadjusted SST anomalies of the KOE and the SPCZ Extension were then compared to their respective adjusted GISS LOTI anomalies. The related curves were surprisingly similar. After removing the impacts of the KOE and the SPCZ Extension from the related GISS LOTI data, the linear trends dropped significantly. When the two GISS LOTI datasets were again combined, we had removed approximately 85% of what some consider to be the “anthropogenic global warming signal.”

This post differs from studies such as Thompson et al (2009). Thompson et al assumed that the ENSO proxy accounts for all of the processes within the Pacific that take place during ENSO events. In reality, NINO3.4 SST anomalies (or the CTI SST anomalies they used) can only account for the linear responses to the changes in equatorial Pacific SST anomalies. NINO3.4 SST anomalies cannot be assumed to account for the ENSO processes that take place within the Pacific or the aftereffects of those processes. What I presented in this post was a simple way to view those aftereffects within the Pacific and the global responses to them.

In short, I presented a story told by the GISS Land-Ocean Temperature Index and Reynolds OI.v2 SST data between the latitudes of 60S to 60N.


Anonymous said...

Congratulation Bob to such a lengthy post.
One thing, should not be better to use some other dataset than GISS LOTI? I would create a dataset with the CRUTEM land component detrended to fit the UAH above land trend, combined with OI.v2. Now the remaining trend might be still of some UHI origin.

On the other side, since both types of ENSO events boost the warm pools elsewhere, what will change after recent climatic shift? Because "plenty of la Nina" period will just provide ground for next El Ninos.
Juraj V.

lucia said...

I have an predictive question: How might we know when another 0.2C-0.4C jump might occur. Are we due for one during this summer's La Nina?

Roger Andrews said...


Excellent post! I think you have come (are coming?) closer than anyone to explaining the recent increase in global temperatures in rational terms.

However, I do have a couple of (hopefully constructive) comments.

First, you are comparing a temperature series that's about 75% based on SST data with ocean indices that are derived directly or indirectly from SST data. So when you compare the two you are bound to see a relationship. I don't know how big a problem this really is, but you could get around it by using the GISS "met station only" time series, which doesn't use SST data. This way you would be comparing two independent rather than interdependent data sets. And the results should be essentially the same because GISS met is quite similar to GISS land and ocean.

Second, as you point out, ocean indices are commonly estimated by detrending the SST data, with the implicit assumption being that pockets of warm/cold water bubble periodically to the ocean surface and then go away again. If I understand you correctly you are now postulating that the warm pockets don't go away, they stay there, and this can happen only if they are being replenished. This is a critical strikes me Maybe you dealt with this in the text and I missed it. If so, apologies).

Roger Andrews

Roger Andrews said...

Whoops, my earlier post got garbled. Please delete "this is a critical strikes me" and insert a "(".

Bob Tisdale said...

Roger Andrews: Regarding the use of an interdependant dataset, the first reason for the post was to explain the rise in global temperatures. A land surface dataset doesn't fit the bill. I may in the future redo the graphs by using a land surface dataset, becuase that would only require changing a few inputs to the spreadsheet, changing the scaling factors, and the title blocks.

You wrote, "If I understand you correctly you are now postulating that the warm pockets don't go away, they stay there, and this can happen only if they are being replenished."

The La Nina replenishes the KOE and SPCZ Extension. I'll try to make that clearer in a follow-up post.

Bob Tisdale said...

lucia said, "How might we know when another 0.2C-0.4C jump might occur. Are we due for one during this summer's La Nina?"

Hi Lucia. Isn't this La Nina going to end in 2011?

Also I have a problem replying to your question. In order to see that there's been a shift, using the simple methods I use, I need a few years of data after the fact to establish it.

Bob Tisdale said...

Juraj V. asked, "On the other side, since both types of ENSO events boost the warm pools elsewhere, what will change after recent climatic shift?"

I hate to answer a question with a question, but Has there been a climate shift?

If we can have a string of El Nino events without a La Nina to recharge them, then the global temperatures should remain relatively flat as they had for most of the 2000s.

Also I used GISS LOTI for a number of reasons. The primary reason was for the Reynolds OI.v2 data.

lucia said...

Hi Lucia. Isn't this La Nina going to end in 2011?
I don't know. Some La Nina's last longer.

I know my question is challenging, and likely puzzling. I ask because from time to time, my readers will point to your blog as something that should make me question my notion that the earth will break a temperature record during next El Nino. (I still think it will.)

I see interesting discussions of what might of the ENSO process, and some explanation why the we see average temperatures exhibit warming or cooling at certain points in the process, I really don't know about the conclusions about the rate of warming.

If I understand you correctly, and you are correct about temperature rising in big jumps, then since 1985,, we've seen only 2 jumps. That's 2 cycles for the process -- which makes it difficult to have much confidence that data support a description of a fairly involved process.

Although I find the discussion of the motion of water fascinating, I'm not entirely sure why I should place more confidence in a trend computed over 2 jumps computed based on the "jump" notion relative to original OLS trend. I guess I'd be more convinced if the evidence showed more than 5 repeated clean jumps.

On this:

The La Nina replenishes the KOE and SPCZ Extension. I'll try to make that clearer in a follow-up post.
So, did this not get replenished during the 2008 La Nina? Do you anticipate that by monitoring this in 2011 you will be able to make a good stab at the size of the next El Nino? Or whether or not we will see a jump?

(Yes. I'm going back to trying to learn whether you think your explanation can be turned into a mid-range forecasting tool. Do you?)

Roger Andrews said...


Relative to your response to my earlier post, I guess I don't see why a land station time series shouldn't "fit the bill". Over the time period you are looking at there are enough stations on islands etc. to give reasonably good coverage over the oceans, and we don't see any large differences when we plot GISS met against the LOTI series (which I just did).

Another advantage of using a land station-based series is that land stations measure air temperatures directly while LOTI is dominantly a proxy reconstruction that uses SSTs as surface air temperature surrogates. Whether in fact SSTs are valid surface air temperature proxies is of course a separate issue, but by not using them to define air temperature trends you are at least eliminating another potential source of uncertainty.

Bob Tisdale said...

Lucia says, "So, did this not get replenished during the 2008 La Nina?"

Yes it did. It's apparent in the ENSO and Volcano adjusted data in Figure 3 or 4.

Lucia asked, "Do you anticipate that by monitoring this in 2011 you will be able to make a good stab at the size of the next El Nino?"

This is not a tool for predicting the strength of an El Nino.

You asked, "Or whether or not we will see a jump?"

One of the reasons the shifts were so apparent after 1988 and 1998 were that they were followed by a series of small El Nino events, but no La Nina events. The 2002/03 and 2004/05 El Ninos were not followed by La Nina events, and there wasn't a La Nina event from 1990 to 1994.

If we see a period like that after this La Nina (with a couple of small El Nino events with no La Nina between them) the upward step following the 2010/11 La Nina would be easier to picture.

You asked, "Yes. I'm going back to trying to learn whether you think your explanation can be turned into a mid-range forecasting tool. Do you?"

What I've presented is an attempt to explain why temperatures have risen. It also was intended to show that NINO3.4 SST anomalies do not capture the process of ENSO. It wasn't intended as a forecasting tool.

HR said...

Bob, really thought provoking.

There's a paper in press at GRL that suggests indices for two types of ENSO. I strikes me that the simple subtraction of a figure based on the tradnitional ENSO index might have no base. I'm not particularly criticising you here, I know many do the same sort of subtraction, I just wanted to get a feel about how useful this type of analysis is? I appreciate you already have identified some of the limits to reseachers understanding on the ENSO index.

Ren, H.-L., and F.-F. Jin (2011),
Niño Indices for Two Types of ENSO,
Geophys. Res. Lett., doi:10.1029/2010GL046031, in press.

A question on the scaling. Are the scaling values for KOE and SPCZ a measure of the amount of the warming(cooling signal) in the wider region? E.g. is 25% of the 60S-20N signal contained within the SPCZ. This sounds wrong but I have to ask.

The consensus would say that ENSO is an energy neutral process. It's not a sourse of energy so can't be responsible for a trend, it's just energy moving around the system. I wonder how you counter that? Maybe in a related way, ENSO has been going on forever. If you projected these step changes into the past (I see Lucia is already trying to project them into the future) then the world would be boiling now. So there must be something behind this making the step changes upwards (rather than downwards) at the moment and the possibility that future step changes downwards. And if that's natural then a negative or alternative process to allow steps down. So any idea what these other processes might be? Couldn't AGW still be correct just that the earth's system, thru' ENSO, is translating the smooth increase in forcing from CO2 into abrupt step changes?

Bob Tisdale said...

HR: I haven’t read Lin & Jin (2011) “Niño Indices for Two Types of ENSO”, and they’ve got a paywall on the preprint, so I can’t say if it would have an impact on these results. (The scaling of NINO3.4 SST anomalies [or CTI] and subtracting it has been a commonly used method for removing the linear effects of ENSO.)

You asked. “A question on the scaling. Are the scaling values for KOE and SPCZ a measure of the amount of the warming(cooling signal) in the wider region.”

The entire East Indian-West Pacific (60S-65N, 80E-180) dataset rises in similar upward steps, and like the KOE and SPCZ Extension, that’s before the effects of ENSO and volcanic eruptions have been removed. The KOE and SPCZ extension have the strong signals within that region.

You wrote, “The consensus would say that ENSO is an energy neutral process.”

And they would be wrong, as illustrated in this post. Their opinions are faulty as I’ve shown in this post. The use of an ENSO index does not account for the redistribution of warm water.

You wrote, “If you projected these step changes into the past (I see Lucia is already trying to project them into the future) then the world would be boiling now.”

The methods used in this post are likely dependent on the climate of the Pacific after the 1976/77 Climate Shift. Speculation: It would seem likely to me that similar conditions existed in the earlier warming period (1910 to 1945) but did not exist during the cooling epoch (1945 to 1975). Is it dependent on the AMO as well? Dunno.

You asked, “So any idea what these other processes might be?”

I’d be adding to the above speculation. The problem with trying to identify them is, there is less and less source data as we go back in time. Poorer coverage and more infilling. Many studies of ENSO start in 1950 because of the earlier data. Compound that with the need to use the SST data for two more relatively small regions in the Pacific for this analysis.

You asked, “Couldn't AGW still be correct just that the earth's system, thru' ENSO, is translating the smooth increase in forcing from CO2 into abrupt step changes?”

There’s no evidence of an anthropogenic signal in the source of fuel for ENSO, which is the tropical Pacific OHC.

HR said...

Thanks for the answers Bob, hope you don't mind me pushing a bit further.

On your point about no increase in OHC in the tropical Pacific I plotted the NODC OHC using KNMI Climate Explorer.


Using your years the NINO4.3 region does appear to have some increase in OHC. A lot of variability but a step change after the 1998 El Nino.

BTW I'm watching the news here in Australia while writing this. Watching the effect of La Nina rains in Queensland. The need to understand all this seems highly pertinent.

Bob Tisdale said...

HR: Sorry if I misled you with my earlier comment or that you misunderstood what I meant. Obviously, I should have been clearer. Based on your coordinates on your KNMI graph, it appears you’ve plotted the central and western tropical Pacific. The graph of tropical Pacific OHC I was referring to captures the entire Tropical Pacific (20S-20N, 120E-90W). Here’s the long-term version:

The Tropical Pacific OHC first rises in response to the 1973/74/75/76 La Niña, then decays until the 1995/96 La Niña, when it makes an upward shift. It drops for the 1997/98 El Niño, and is recharged during the 1998/99/00/01 La Niña.

If you were to keep the data you plotted and compare it to the eastern tropical Pacific, you see the effects of the warm water sloshing back and forth. But if you look at the entire dataset as I've shown you can see what causes the overall rises and falls.


HR said...

You didn't mislead me I just didn't know the proper area.

Look I graphed it again using your Lat/long numbers and for 1982-2010.


It still looks like OHC is going up. Put a linear trend and it's definitely sloping up. Use you're preference there seems to be a step change arournd 1998.

Bob Tisdale said...

HR: Trend lines don't indicate the cause of the rise, only the fact that there was a rise. That's the whole problem with the AGW position. Any positive trend in sea level, OHC or temperature is attributed to greenhouse gases.

The upward step in your graph can be seen in the longer-term graph:

There's even an upward trend in the long-term data, but the majority of the rise is caused by upward steps during La Nina events, which would be attributable to decreased cloud clover and the resulting increases in downward shortwave radiation.

HR said...

Thanks again,

I'm still not convinced on you're explanation on the the OHC data in tropical pacific because it appears you're speculation on what's driving the changes in OHC data is just that speculation. I don't see how that data either rules in or rules out an anthropogenic signal. But I'm happy to accept my own limitations on the subject isn't helping, a bit more reading is called for I think.

I was actually really interested in your simple (and maybe profound) observation "that the effects of El Niño-Southern Oscillation (ENSO) cannot be accounted for using a single index"

Just before I read you're article I also read one by Tamino


I'd found his post compelling until I read you one. I guess his post is doing a similar job to some of the published work you refer to such as Thompson et al (2009). I guess not only does he assume that La Nina is the reverse of El Nino but that there is a linear relationship between ENSO and global temperature going back in time.

Anyway I was digging around today, mainly promoted by what's going on in Queensland ATM, and came across this paper.

(this one's better that it's freely available)

While it focuses on East Asia it shows that the effects of ENSO are modulated by the phase of PDO in a quite dramatic way. While this is a regional analysis the introduction suggests that the phenomenon has been seen in other ENSO affected regions. It seems to agree with you on the fact that ENSO index alone is a poor tool for investigating it's effect on climate but maybe for different reasons.

A global analysis on the relation between ENSO, temperature and the PDO phase would be fascinating. Have you done anything along these lines? Maybe something to keep you busy for the next two years ;)

Bob Tisdale said...

HR: You wrote, "I don't see how that data either rules in or rules out an anthropogenic signal."

The fact that tropical Pacific OHC dropped from the late 1970s to the mid 1990s makes it difficult to argue for the existance of an anthropogenic signal. That and the fact the the only major rises occur during La Nina events, when DSR increases due to reduced cloud cover, makes the AGW argument even more difficult.

You asked, "A global analysis on the relation between ENSO, temperature and the PDO phase would be fascinating."

When we go back in time, there are less source SST measurements and more infilling. It would in interesting to see if the process I used in this post also worked during the early 20th century warming period, but I don't expect that it would.


Tips are now being accepted.

Comment Policy, SST Posts, and Notes

Comments that are political in nature or that have nothing to do with the post will be deleted.
The Smith and Reynolds SST Posts DOES NOT LIST ALL SST POSTS. I stopped using ERSST.v2 data for SST when NOAA deleted it from NOMADS early in 2009.

Please use the search feature in the upper left-hand corner of the page for posts on specific subjects.
NOTE: I’ve discovered that some of the links to older posts provide blank pages. While it’s possible to access that post by scrolling through the history, that’s time consuming. There’s a quick fix for the problem, so if you run into an absent post, please advise me. Thanks.
If you use the graphs, please cite or link to the address of the blog post or this website.