Before proceeding, credit where credit is due. The concept of a relationship between the sun and ENSO events isn’t new. It’s been discussed at least twice on this blog (here and here) and in detail by Theodor Landscheidt (here).The connection between ENSO events and warming isn’t new either. Three years ago Bob Tisdale (here) showed how ENSO events caused periodic upward shifts in the SST record that explained all of the recent global warming. Also not new, thanks to our host, is the theory that the oceans periodically release stored heat to the air (here). So a h/t to these gentlemen and to any others I may have omitted.
What follows is my attempt to condense these hypotheses into a narrative that uses observational data to illustrate how the solar cycle, ENSO events and the release of stored ocean heat, and not man-made greenhouse gases, combined to cause the recent global warming, which began, incidentally, in 1976.
ENSO Events and Solar Cycles
Figure 1 plots the Niño3.4 Index since 1960 (the Bivariate ENSO, Multivariate ENSO and Oceanic Niño Indices give essentially the same results). I’ve used the commonly-accepted +/-0.5C threshold to define individual Niño and Niña events and the zero crossover to define Niños and Niñas that transition directly into each other, and the duration of each event is shown by the red and blue vertical stripes.
Figure 1: ENSO Events Defined by Niño3.4 Index (Niños red, Niñas blue)
We’ll start by superimposing sunspot number on the ENSO stripes. There’s a broad periodicity in the ENSO events but no clear relationship with the sunspot cycle:
Figure 2: ENSO Events vs. Sunspot Number (Niños red, Niñas blue)
Next we’ll plot the El Niños against sunspot number. Now there’s even less of a relationship.
Figure 3: El Niños vs. Sunspot Number
Now we’ll plot the La Niñas, which I’ve numbered for reference:
Figure 4: La Niñas vs. Sunspot Number
We have a match. Niñas 1, 3, 4, 6 and 8 begin on the back side of the sunspot cycle and end almost exactly at sunspot minimum. Niñas 5 and 7 begin on the front side and end at or very close to sunspot maximum and ongoing Niña 9 is poised to follow suit. The only Niña that doesn’t fit the pattern is Niña 2, but when we include it another pattern emerges – two superimposed La Niña cycles (2, 4, 6, 8 and 3, 5, 7, 9) with 11-12 year periodicity that are shifted relative to each other by 3-4 years.
The nine Niños that transition directly into a La Niña automatically show the same sunspot cycle-linked periodicity, with Niños 5, 7 and 9 beginning at or just after sunspot minimum:
Figure 5: El Niños That Transition into La Niñas vs. Sunspot Number
The six El Niños that didn’t transition into La Niñas also broadly match the sunspot cycle, beginning at or after sunspot minimum in cycles 20 and 21 and at or after sunspot maximum in cycles 23 and 24.
Figure 6: Non-Transitional Niño Events vs. Sunspot Number
It is of course possible that these correlations are purely coincidental, but the odds are strongly against it. I think we can reasonably conclude that the sun controls ENSO events, although the mechanism is obviously complex.
The Warming Impacts of ENSO Events
Figure 7 plots the HadSST2 global sea surface temperature record since 1960. It shows that the recent SST warming began with the abrupt temperature increase in 1976 (the surface air temperature record shows the same thing), so to explain the recent warming we need only concern ourselves with the period since 1976. It also shows that the SST record is quite heavily distorted by ENSO impacts, such as the 1972/73 and 1997/98 El Niños, and the trend line approximates the gradual warming gradient we would expect to see if the warming had been caused by greenhouse gases and if the ENSO distortions weren’t there:
Figure 7: HadSST2 Global Sea Surface Temperatures, Degrees C
But I don’t see straight-line warming. I see the warming occurring in a series of stair-step upward shifts beginning in 1976 (with an earlier downward shift in 1964/65) with +/-10 year intervals of no warming between the shifts, again mirroring sunspot cycle periodicity. Here’s an eyeballed approximation:
Figure 8: HadSST2 Global Sea Surface Temperatures, Upward Shift Approximation
But the record is too heavily distorted by ENSO impacts to be absolutely certain the shifts are there, so the next step is to remove the ENSO impacts.
The accepted way of doing this is to take an ENSO Index – usually the Eastern Equatorial Pacific Cold Tongue Index – and use regression relationships to quantify the temperature impacts of variations in the Index and then subtract them from the SST record. However, this approach assumes that the global SST response is always directly proportional to the amplitude of ENSO events, which it isn’t, and also that ENSO events have only short-term impacts.
A more robust approach is simply to remove the El Niño and La Niña intervals from the global SST record altogether. I did this by deleting all the months where the Niño3.4 Index was greater than 0.5C or less than minus 0.5C, and I also deleted the periods of cooling after the 1963 Agung and 1991 Pinatubo eruptions – the 1982 El Chichón cooling, if any, gets removed anyway because it coincides with an ENSO event. (A few notes: First, the criteria admit data during some Niño-Niña transitions, so there will be points inside the pink and blue ENSO strips shown in the previous Figures. Second, the post-Agung cooling probably had nothing to do with the Agung eruption but I deleted it anyway because it cleans the plot up. Third, removing the ENSO data with a 6-month lag between Niño3.4 and global SST makes no significant difference to the results.)
Figure 9 shows what remained:
Figure 9: HadSST2 Sea Surface Temperatures, ENSO & Volcanic Cooling Events Removed
The shifts in the SST record are clearly visible, they occur during ENSO events and there’s no evidence for any upward shifts of comparable size in the intervals between the ENSO events. The trend lines in fact show net cooling rather than warming during these intervals.
Except for the 1964/65 shift, which isn’t obvious in the surface air record, the same shifts occur in the global air temperature records:
Figure 10: Surface Air (GISS Meteorological Station Only, Red) and Lower Troposphere
(UAH TLT, Green) Temperatures, ENSO & Volcanic Cooling Events Removed
There may be some residual questions as to the extent to which the sun controls ENSO events, but I don’t think these results leave much doubt that the post-1976 warming was caused by ENSO events and not by man-made greenhouse gases.
The Relationship Between Temperature Shifts and ENSO Events
Figue 11 shows that the three upward shifts in the SST record coincide with the 1972-76, 1986-1989 and 1997-2000 transitional Niño-Niña events, but the SST data are too scattered to tell us exactly when they occur, so in this section we will look into that more closely. We will also look into the question of why other transitional Niño-Niña events fall outside the shift windows.
Figure 11: HadSST2. ENSO Impacts Removed vs. Transitional Niño-Niña Events
Figure 12 shows how global SST changed during the 1981-1992 Niño-Niña-Niño-Niña transition.
Figure 12: Niño3.4 (left scale) vs. HadSST2 (right scale), 1982-1989 ENSO Events
SSTs rose during and after the 1982 Niño (swamping the cooling impacts, if any, of the El Chichón eruption), fell back to pre-1982 levels during the erratic but prolonged 1983-86 Niña, rose again during the 1987 Niño and then fell back once more to pre-1982 levels during the 1988 Niña. It was not until the recovery from the 1988 Niña that temperatures finally rose and stabilized at a higher level. (The sequence of events during the earlier 1969-1976 transition was the same.)
The same pattern is seen Figure 13, which shows how global SSTs, surface air temperatures and lower troposphere temperatures changes during the 1997-2000 Niño-Niña event. The upward shift in all three records occurred during the recovery from the La Niña:
Figure 13: HadSST2, 1997-2001 ENSO Events
Evidently the shifts don’t occur until SSTs stabilize, which doesn’t happen until after the final Niña regardless of the number of Niños and Niñas that precede it.
The fact that Figures 11 and 12 show lasting heating impacts from La Niñas but not from El Niños seems counterintuitive given that El Niños heat the sea surface while La Niñas cool it. But since 1960 there have been two strong El Niños that didn’t transition directly into or out of a Niña, and the impacts of these events on global SST ranged from minimal to zero:
Figure 14: Niño3.4 (left scale) vs. HadSST2 (right scale), 1991 and 2002 El Niños
The 1991 Niño generated a ~0.1C hump in the post-Pinatubo cooling trend but that was all. It can be argued that Pinatubo stifled this particular Niño, but the 2002 Niño had no impact whatever on global SST and there were no volcanic eruptions of any size in 2002.
- El Niños by themselves have no permanent impact on global temperatures (and in the case of the 2002 El Niño not even a temporary impact). The upward shifts in global temperature since 1976 are associated with La Niñas and occur at or around the end of the La Niña events.
- However, since 1960 there have been no La Niñas that weren’t preceded by an El Niño, so it’s reasonable to assume that the El Niños are what initiate the process.
Source of the Heat Introduced by the ENSO Events
The upward shifts in the global temperature records were caused by the release of excess heat from deeper in the ocean to the sea surface. We can be certain of this because a) there’s nowhere else the heat could have come from and b) the heat must have been excess or the ocean wouldn’t have released it.
In all likelihood this excess heat got into the ocean during the rapid increase in TSI between 1910 and 1960, which warmed not only the sea surface (Figure 15) but which would also have warmed the ocean layers beneath it:
Figure 15: ICOADS SST (left scale, degrees C) vs. Shapiro TSI (right scale, smoothed, W/m2)
The abrupt divergence between the TSI and SST records that begins in 1976 coincides with the year in which the stored ocean heat began to surface. The trigger was the series of strong ENSO events that ended in that year at sunspot minimum between cycles 20 and 21, but reduced solar activity during cycle 20, which was much weaker than the cycles that preceded it, might also have contributed..