New Data, Old Claims About Volcanoes

Guest Post by Willis Eschenbach

Richard Muller and the good folks over at the Berkeley Earth Surface Temperature (BEST) project have released their temperature analysis back to 1750, and are making their usual unsupportable claims. I don’t mean his risible statements that the temperature changes are due to CO2 because the curves look alike—that joke has been widely discussed and discounted, even by anthropogenic global warming (AGW) supporters. Heck, even Michael Mann jumped on him for that one, saying

It seems, in the end–quite sadly–that this is all really about Richard Muller’s self-aggrandizement 🙁

And if anyone should know about “self-aggrandizement”, it’s Michael Mann … but I’m not talking about Muller’s claim that humans caused the warming. No, I mean the following statement:

The historic temperature pattern we observe has abrupt dips that match the emissions of known explosive volcanic eruptions; the particulates from such events reflect sunlight and cool the Earth’s surface for a few years.

In support of this statement, Richard Muller offers up the following chart:

Figure 1. BEST claims about temperature and volcanoes. SOURCE

So what’s not to like?

Well, first it appears he has included and excluded volcanoes depending on whether they show up in his temperature record. If we look at big eruptions, eruptions with a “volcanic explosively index” (VEI) of 6 or above, since 1750 we have the following volcanoes:

Mount Pinatubo, 1991

Novarupta, 1912

Santa María, 1902

Krakatoa, 1883

Mount Tambora, 1815

Grímsvötn and Laki, 1783

So Muller has left off Santa Maria and Novarupta, and included El Chichon and Cosiguina. But that’s not the real problem. The real problem is that many of these occurred after or during the temperature drop that they are supposed to have caused … here’s the BEST data including all relevant volcanoes, without the style of overlay that they have used that obscures the actual timing:

Figure 2. BEST temperature data and dates of volcanoes. Red line is a four-year centered Gaussian average of the temperature data. Photo shows Mt. Redoubt in Alaska.

So let’s look at the volcanoes, one by one:

LAKI, 1783: Occurred near the end of the fall in temperature that it is supposed to have caused.

TAMBORA, 1815: Occurred at the end of the fall in temperature that it is supposed to have caused.

COSIGUINA, 1835: Occurred near the middle of the fall in temperature that it is supposed to have caused.

KRAKATOA, 1883: Occurred at the end of the fall in temperature that it is supposed to have caused.

SANTA MARIA, 1902: Occurred in the middle of the fall in temperature that it is supposed to have caused.

NOVARUPTA, 1912: I can see why Muller omitted this eruption, which occurred just before a rise in temperature …

EL CHICHON, 1982: Occurred during the fall in temperature that it is supposed to have caused.

PINATUBO, 1991: This is arguably the only one of the eight volcanoes that could legitimately be claimed to cause a detectable fall in temperature … a whopping fall of 0.15°C or so.

So while volcanoes certainly may cause a minor drop in global temperature, the claim of Richard Muller and the BEST folks that there are “abrupt dips that match the emissions of known explosive volcanic eruptions” is simply not true. There are abrupt dips, but they don’t match up with the volcanic eruptions.


[Update] Further reading:

Prediction is hard, especially of the future discusses the GISS analysis of Pinatubo.

Missing the Missing Summer is about the eruption of Tambora.

Dronning Maud Meets the Little Ice Age investigates a claim that the Little Ice Age was triggered by vulcanism.

Volcanic Disruptions plays the game “Spot the Volcano”

[Update] Another way to investigate the question is to look at the average temperature anomaly during the two years before and the two years after the eruption. Figure 3 shows that result.

Figure 3. Average temperature anomaly two years before and two years after the eruptions. Black lines show the standard error of the mean.

After some eruptions it cooled a bit, after some it warmed a bit, and after some there was no change … go figure.


newest oldest most voted
Notify of

Wonderful forensic dissection of the usual warmist cack.
Why do these people keep lying and keep thinking they can get away with it?


So what we really might have is a possible relation with falling temps causing large volcanic eruptions…. /sarc

If you think up a theory get the facts to correlate otherwise theory goes down.

Steve Crook

I continually seem to suffer cognitive dissonance with this sort of thing. Willis presents figures with which it is hard to argue *and* they’re easy to find. A group of reputable scientists have access to the same figures and draw an opposite conclusion. One bit of me wants to believe the scientists, but the other just looks at the graph. Is it any wonder I’m sceptical and my head hurts?

Ian H

Don’t forget that a four year centered Gaussian average would start reacting to an abrupt drop two years ahead of time. Even so this doesn’t rescue the claim.

Jimmy Haigh

But CO2 always rises after the temperature rises it causes (but only 400 years after), so it stands to reason that volcanic eruptions do the same: they happen after the temperature drops they cause. (Well, some of the time anyway.)
“Climate Science” is incredibly difficult. No wonder it’s only the geniuses of ‘the team’ and their acolytes who can understand it. We denialist scum don’t have a snowball’s chance. Even Einstein would have struggled to understand the temporal complexities involved…
(/sarc. Just in case…)


Ian H, that’s a good point. If we are going to look at the effects of point events then any form of spreading on the data needs to be avoided. Also is there anything odd about Novarupta that might explain why it seems to have a positive effect (though it could all be coincidence).

Ian H says: July 30, 2012 at 3:29 am
“Don’t forget that a four year centered Gaussian average would start reacting to an abrupt drop two years ahead of time. Even so this doesn’t rescue the claim.”

I would have thought more like four years (if sd=4), which does make the claim look better. Only Novaruptna looks to me to be totally out – some of the others seem to be a rather weak response.

Paul Matthews

It’s very interesting that in this paper they pointed out the effect of Pinatubo (1991, typo in post).
Previously, they forgot about Pinatubo when it suited their argument: Look at the FAQ on the BEST page, under “Has Global Warming Stopped?” and you’ll read “However, if you did that same exercise back in 1995, and drew a horizontal line through the data for 1980 to 1995, you might have falsely concluded that global warming had stopped back then.” No mention of Pinatubo. This was misleading of course because that pause was related to the eruptions, whereas the present one isn’t.


I’ve been trying to find Galeras (1993) in the lists, but nobody takes it into account, in spite of the victims. It occurred in january, (summer season in the souther hemisphere ), two years after Pinatubo. But in the chart, the temperature goes up. No visible dent.
Why is it so ?


Now this should be a simply one , the date a Volcano goes bang big style is easy to know , the temperatures over a period are easy to know , as the date is there taken on are recorded . So it is straight forward to see if there a link between them .
What we see here is a classic way in which statistics can be used to what is know to appear to be wrong because manipulating the numbers produces ‘accurate ‘ but wrong data . In other words its statistically created lie.


Ian H says:
July 30, 2012 at 3:29 am
Don’t forget that a four year centered Gaussian average would start reacting to an abrupt drop two years ahead of time. Even so this doesn’t rescue the claim.
Indeed. The smallest wiggles just before (two years is not much in this image!) and after Laki and Tambora may still be consistent with a cooling by the volcanoes, but not the drop in temperature several years in advance.
George says:
July 30, 2012 at 3:12 am
So what we really might have is a possible relation with falling temps causing large volcanic eruptions…. /sarc
I see that /sarc-tag, but I wouldn’t completely rule out a common cause…


Don’t forget there is statistical uncertainty over laid on the volcanic timings hence, the timings can be pushed forward or back depending on what cooling/warming trend was at the time.

Paul Matthews

Also, there’s an elementary error in their graph. El Chichon was 1982. The arrow on their graph points to early 1960s which was Agung.


I was looking at the Tambora eruption and not comparing it with the Berkley data – just trying to square it with the “Year Without a Summer” (1816) observations in the northern hemisphere. Does the data show some dramatic warming elsewhere on the globe or was the winter (1815 or 1816) not as cold as normal despite the summer conditions? I merely point it out because the “Year without a Summer” is after the Tambora event.

son of mulder

What happens to the actual temperature record after each eruption, forget about gaussian smoothing as we’re looking for patterns of cause and effect?


Muller threw Mann under the bus years ago, in an attempt to become the Golden Child of AGW, and now Mann is showing his professionalism in return.

Rather than ‘weak response’ IMO what you really mean is that for most of these volcanic eruptions there is more or less a non-existant response to the eruption when you fully allow for the year to year variability in the anomaly index.
IMO the correlation between ‘volcanic eruptions and cooling of the planet’ argument is very weak and is used as a ‘sticking plaster’ by warmists like your good self (i.e. that man-made or non man-made ‘dust’ causes cooling) to try and explain away the 1940 to 1970 global cooling period. The reality is that this well established cooling period has nothing to do with ‘particulate emissons’ from the post WW2 global industrial expansion and everyting to do with the approx. 60 year (most likely wholly) natural cyclic climatic variations we experience in the earth’s climate, the cause of which we are only just beginning to identify and understand.


As the song says: “Two outa three ain’t bad”, but one outa seven?


That reminds me of


And the 1980 Mount St Helen eruption, in mid may, after a couple months of earthquakes , and a really big one just the day before. That was a major eruption , but it is not shown on the graph, either.


Novarupta’s high latitude position meant its eruption had relatively little effect on global climate. Try Kravitz and Robock 2011.


Looking at the BEST curve in the media, it seems as if the only function of the pre-1900 volcano data is to make the curve match look good. The 20th century match to CO2 is just a general rise, in fact my age correlates better with the CO2 than the temp does.


Maybe the cooling CAUSES the volcanos – i.e. making the earth’s mantle more brittle or something?
/sarc off


can someone enlighten me as to what the units are for the solid (dark) line on the Berkley plot?

D. Cohen

Don’t forget that for temperature drops associated with eruptions several centuries ago, there might have been unrecorded volcanic eruptions just around the time of the recorded ones. So if there is a tendency for eruptions to cluster in time, and not all the eruptions have been recorded, then we might expect to see temperature drops correlated with eruptions, with unrecorded eruptions starting off the temperature drop and the few recorded eruptions occurring somewhere in the middle (or even toward the end) of the drop. Only in recent times, when all the big eruptions are known, would we see a relatively pure Pinatubo effect with the known eruption preceeding a subsequent temperature drop.
So to test this hypothesis, is there a tendency over the last, say, five decades — when we can assume very good record keeping — for volcanic eruptions to cluster in time?

Nick says:
July 30, 2012 at 4:59 am
Novarupta’s high latitude position meant its eruption had relatively little effect on global climate.

That could be one possibility. What’s your theory about none of the others having much (if any) effect?


[snip . . aw c’mon now, just think it don’t post it OK? . . mods]

Bill Illis

We have seen this before. The pro-AGWpeople like to point to volcanoes causing temp drops but when you get into the data, there is either nothing there or the temps had dropped long before the volcano or long after.
I also looked at Berkeley’s new numbers. There is little match to volcanoes once again.
Temps fall in 1781 and 1782 before Laki; Temps were already low by 1809 well before Tambora; Temps do nothing after Krakatoa in 1883, if anything they go up etc. etc.

It seems some people like the volcano inference, but it seems to me that Willis is correct. Volcanoes just don’t matter much (and we could do nothing about them if they did). Where is Mosh with some thorough defense, and Dr. Curry?

Don K

A couple of small notes:
1. The CO2 plus volcano line doesn’t really seem to track temperature very well — especially in the 1930s.
2. The Tambora eruption chanced to occur after several very cold years. e.g. the last frost fair on the frozen Thames in London was held in 1814 — a year prior to the Tambora eruption. Tambora may or may not have extended the cold spell.
3. Looking at satellite measurements that cover the time period of the Pinatubo eruption, it appears that El Nino/La Nina effects probably have at least as much influence over temperatures as Pinatubo did. Maybe more.

John Brookes

So I think Willis has shown that sudden dips in temperature can be used to predict volcanoes?


Just to add, you know, Michael Mann and his ilk deem it appropriate to label climate change skeptics as holocaust deniers and beetle larvae. Only seems fair that we recognize his contributions to climate change science by giving him his own category.
MANNIAN SCIENCE = hokey schtick science.

@Justice4Rinka says: July 30, 2012 at 3:12 am
“Why do these people keep lying and keep thinking they can get away with it?”
Well, the evidence so far is that they DO get away with it, and indeed, are encouraged to do so by all manner of vested interests, prime amongst which being the political classes and the MSM.

Wasn’t Mt. St. Helens a VEI 5 in 1980? Wouldn’t that help El Chichon show up? And where is Mt. St. Helens here ? I think there are more things to consider here. Also, what justification beyond models is there for tying CO2 to some presumed volcano parameter? Does the paper try to explain?


I seem to remember that Tambora 1815 was preceeded by another large volcanic eruption a few years earlier, but from memory nobody knows where it is. This would make Tambora ok because it was part of a broader temperature decline, partly related to an earlier eruption.
Also, some ‘eruptions’ are part of events that occur over several years or even longer, but may be recorded by history as a single ‘event’, when they aren’t. So for some eruptions, there may be earlier eruptions, from the same volcano or regional area, as the eruptions/episode builds over several years, although this would only be true for a few major eruptive events, not all.

Earlier this year, I did a very crude look at the 12 VEI 5 and 6 eruptions we saw last century. And, I highlighted the temp response, according to HadCrut. As Willis notes, the supposed responses aren’t consistent. If anyone is interested in taking a gander just go here. Some have already stated, ENSO seems to have a greater effect than any volcano.

Bob K.

*** MODERATORS *** Shouldn’t Mueller be just Muller ?
[Thanks, Bob, fixed. -w.]

Bernie McCune

I have direct experience of measuring daily solar values at the surface over several months before and after El Chichon erupted. I was working at a solar furnace in southern New Mexico and during testing we needed to know what solar values were for each test and also to make sure values during each shot did not vary (due to stray clouds and jet contrails). We set the calibrated (I believe the spec for the instrument was within 1 watt/m^2) Eppley pyrheliometer each morning to point directly at the sun. It would track all day on a equatorial tracking mount so that we obtained normal incident energy values (watts/meter^2) throughout the day every working day that we operated the furnace (I worked there for two years). Early morning and late afternoon values were significantly lower due to the sun angles being low and the significant amount of atmosphere that incoming solar energy needed to pass through. Of course the energy level peaked at solar noon. Typically values of solar incoming energy at the surface in the morning and afternoon were many hundreds of watts/m^2 lower than at solar noon. Typical solar noon values during the year on clear days peaked during the fall days when atmospheric moisture content dropped to very low values here in the NM desert. During the “monsoon” season at this time of the year (July and August) when atmospheric moisture content is high (even on clear days with no clouds) the values were lower. Values varied throughout the year but peak surface values in the fall could reach over 1000 watts/m^2 at noon and be as low as 800 watts/m^2 at noon in July and August. I was able to observe a normal annual cycle before El Chichon. After El Chinchon erupted south of us we did not immediately see the results of the fine ash that was blown into the atmosphere but several weeks later as the cloud circled the globe and moved into the NH, we could see an on average steady decline in the solar values until for several months we noted values that were on average about 100 watts/m^2 lower than “normal”. This persisted for at least 6 months but slowly declined until about a year later there were no noticeable effects from the ash cloud. Apparently volcanoes in equatorial regions produce effects like this where volcanoes at higher latitudes or that are not violent enough do not necessarily create these same patterns. I was able to witness a very dramatic decrease in energy input over this fairly long period of time of at least one of the volcanoes noted above. As to the effect on temperature, it would seem that a decrease in normal incident solar radiation of 100 w/m^2 at the surface could easily have some short term effects. At least I can relate to you what the long term energy values were and I did watch the daily plots carefully enough to see overall daily, seasonal, and annual changes in surface energy levels during a “normal” year and in a “volcanic ash” year. It was interesting to see nature at its most dramatic and variable best. Most of us only noticed the beautiful sunsets caused by the ash cloud.

Thanks, Willis. I started to write a post titled “Do These Graphs Set Climate Science Back 3 or More Decades?” But I had a feeling you’d jump on this too. Your post covers most of what I wanted to present.


Laki is further north than Novarupta. The latitudinal defence is BS. Just another post hoc, hand-waving justification for withholding adverse data.


re: galeras eruption: the eruption was not strongly eruptive (VEI 2). For example Pinatubo was VEI 6. If I understand the nomenclature correctly that makes pinatubo 10000 times Galeras (in kg/s of erupted material).
In addition, mostly tropical eruptions have large global climate impacts. Therefore Mt. St. Helens doesn’t feature prominently (though VEI5). Of general importance is, at how much and how high sulfur compounds are injected into the stratosphere.

Novarupta and Laki have in common a very high latitude, in contrast with Pinatubo, which at 15N latitude, injected large amounts of ash into the stratospheric circulation of both hemispheres. This extreme northern location is expected to dramatically reduce their ability to cool global climate. I would not be surprised if they warmed the polar Winter by blocking the radiative losses in the long Winter night.

Jim Clarke

Don’t forget this phrase: “…all else being equal.” To understand the true impact of volcanoes, one must first determine the effects of all the other variables at the time, like ENSO and so on. Only then can we say “All else being equal, this is the impact major volcanic eruptions have on climate.”
It is remarkable how many climate scientists, like Mueller, have forgotten this basic grade school tenant to good science.

Justice4Rinka asks:-
“Why do these people keep lying and keep thinking they can get away with it?”
Because all socialists everywhere and all of those flimflam merchants who are wedded to Common Purpose will always lie to achieve their dastardly ends.


small mistake: PINATUBO, 1912: should read PINATUBO, 1991:
[Thanks, fixed. -w.]

Pamela Gray

The glitter of wriggle matching attracts many. Sun worshippers, CO2 devotees, soot believers, and other hypothesis adherence folks trying to grab the brass ring.

Ian W

George says:
July 30, 2012 at 3:12 am
So what we really might have is a possible relation with falling temps causing large volcanic eruptions…. /sarc
Espen says:
July 30, 2012 at 4:33 am

As Espen says if there is a relationship there is potentially a common cause. After all the Earth is a rotating blob of molten rock with a thin crust two thirds of which is covered in moving fluids being pulled by gravitational forces and stresses due to inertia from rotation rate Length of Day changes. It may take less than we think to form cracks and leaks in that thin mantle.


The quite separate issues here, vulcanism, cold spells, cycles of the sun, sun spots etc, need to be fully and impartially investigated. Piers Corbyn makes his living based on this stuff and has nearly nought resources. He may be not quite spot on, or fail the scientific rigour, but there is something going on. Time some cash was spent finding out what.


Something doesn’t make sense in the BEST plot. Why does the smoothed curve only show dips but no spikes, there are plenty in the data?