New Research Explores Role of Trade Winds, Volcanoes in Warming ‘Hiatus’


Two new peer-reviewed studies further explore possible explanations for the slow-down in global warming over more than a decade, as scientists continue their quest to better understand the pause or ‘hiatus.’

Climate scientists are continuing to work hard to better understand why Earth has warmed more slowly over the past decade than predicted by most climate models. In February alone, two new important papers have been published in Nature journals that address different aspects of the problem.

The first, by Matthew England and colleagues, argues that the strengthening of trade winds over the Pacific Ocean have increased ocean mixing and heat uptake, and that including these wind changes in climate models explains much of the apparent pause.

The second, by Ben Santer and colleagues, suggests that 17 small volcanic eruptions since 1999 have had a modest cooling effect, and that incorporating them into models helps explain divergences in temperatures of the lower part of the atmosphere. Both papers suggest that the factors behind the hiatus are temporary, and do not undermine longer-term projected warming.

Blowing in the Wind

The Earth’s oceans contain the vast majority of the heat absorbed from the Sun. Earth’s atmosphere doesn’t hold much heat, and air temperatures are sensitive to changes in the rate at which the oceans transfer heat to the air. Ocean heat uptake can be impacted by winds; the more windy it is, the more turbulence occurs in the upper layers of the ocean, and the more heat can be transferred quickly from the ocean surface to deeper layers.

England and his co-authors looked at trends in Pacific trade winds, and found a dramatic increase in strength since the late 1990s. The figure below, from their paper, shows global temperatures at the top and trade winds at the bottom, with negative values indicating the acceleration of Pacific trade winds.

When they compared changes in wind speed and temperature over the Pacific region in particular, the researchers found that the correlations are quite strong, and they further found that increasing winds seem to have contributed to a large portion of the observed cooling. The figure below, adapted from their paper, shows changes in trade winds (the black arrows in the top panel) and surface air temperatures over the last two decades (in the bottom panel).

Much of the hiatus in temperatures over the past decade or so has been concentrated in the part of the Pacific where the trade winds have increased, as the rest of the world has largely continued to warm. England and colleagues concluded that “a pronounced strengthening in Pacific trade winds over the past two decades — unprecedented in observations/reanalysis data and not captured by climate models — is sufficient to account for the cooling of the tropical Pacific and a substantial slowdown in surface warming through increased subsurface ocean heat uptake.” That led them to their conclusion that “the net effect of these anomalous winds is a cooling in the 2012 global average surface air temperature of 0.1–0.2 degree C, which can account for much of the hiatus in surface warming observed since 2001.”

They also pointed out that the changes in both wind speed and temperatures seem similar to what had occurred in the period from 1940-1975, when global temperatures also had plateaued. Both that period and the current hiatus are associated with a negative phase of what the authors referred to as the Interdecadal Pacific Oscillation (IPO), which is shown in the first figure. This phenomenon has also been referred to as the Pacific Decadal Oscillation, and other modeling work has suggested that hiatus periods are linked to negative phases of the IPO/PDO. The multidecadal pattern of cooling produced is quite similar to that of more common La Niña events, but with a much longer persistence.

England and his co-authors used a number of climate model runs with observed trade wind changes as inputs to see if they could replicate the hiatus and predict what might happen next. They found that “the wind-induced cooling can account for approximately 50 percent of the observed hiatus when comparing the observed and model SAT projections to 2012.”

“Much of the rest of the hiatus can be accounted for when the wind trends are further prescribed in a fully coupled ocean–atmosphere model,” they wrote. The results of their analysis are illustrated in the figure below, which shows the standard IPCC model projections in red and the new trade-wind-forced models in green.

These results suggest that the current hiatus could persist for much of the current decade if trade winds remain strong, with temperatures again rising to levels predicted in most models once more normal conditions return.  However, the researchers did note that models seemed not to capture the types of changes in trade winds observed: “None of 48 climate model experiments examined in detail captures the magnitude of the recent acceleration in Pacific trade winds.” they wrote. “In fact the most extreme acceleration seen in the models is generally less than half that observed.”

Small Volcanoes — Big Effects?

Climate scientists long have understood that volcanoes have a cooling effect on Earth’s atmosphere in the years after they erupt. Large volcanoes tend to inject lots of sulfur particles into the high atmosphere (the stratosphere), where it reflects away incoming radiation. It’s easy to see large dips in global temperatures associated with volcanic eruptions like Pinatubo, and the massive eruption of Mount Tambora in 1815 is thought to have caused the famous “year without a summer” in 1816.

Volcanoes cause cooling because sulfur particles are quite reflective. In the high atmosphere, they scatter the Sun’s incoming rays, resulting in less solar radiation reaching Earth and a general dimming of the atmosphere.

With satellites in orbit around the Earth, scientists can now accurately measure just how much sunlight is being reflected away (referred to as stratospheric aerosol optical depth). This capability allows better measurements of the impacts of modern volcanic events. While large volcanic events have by far the largest effect on stratospheric aerosol optical depth, there is also a modest increase over the last decade even in the absence of major volcanoes, with global mean stratospheric aerosol optical depth increasing by 4-7 percent annually from 2000-2009.

Santer and colleagues, in their new Nature Geosciences paper, examined the impact of increased stratospheric aerosol optical depth over the past decade on atmospheric temperatures. This is a factor not well represented by climate models, which generally include no volcanic events in the period from 2000 through 2014. Instead, they have focused primarily on lower tropospheric temperatures (called temperature lower troposphere — TLT), where the divergence between modeled and observed warming has been particularly pronounced.

The figure below, from their paper, shows the range of climate models (in grey) and the multi-model mean (in black) compared to satellite observations (red and blue). The top panel shows the raw data; the middle panel shows the data once the effects of periodic El Niño and La Niña (ENSO) events are removed; and the bottom panel shows ENSO and major volcanoes (Pinatubo and El Chichon) removed.

Volcanoes tend to have a larger effect on TLT than surface temperatures, so the impact of Pinatubo and El Chichon are relatively easy to see, especially once ENSO effects have been removed. Even once all these external factors have been accounted for, observations are still on the low side of the range predicted by models. As Santer and colleagues remarked in the paper, “after 1999, however, a ‘warming hiatus’ is still apparent in the observed residual TLT time series [even after removing ENSO], but the lower troposphere continues to warm in the CMIP-5 multi-model average.” (CMIP-5 refers to the set of models presented in the latest IPCC report.)

CMIP-5 models have no volcanoes over the last decade, even though 17 relatively modest eruptions have occurred since 1999. Santer and colleagues suggested that the cumulative effect of many small volcanoes over the last decade might help explain a small part of the global temperature discrepancy.

They looked at two different statistical tests. The first test examines each volcanic eruption in turn to look for discernible signals in the TLT response. This method finds that three of the 17 volcanic eruptions since 1999 have effects that are significant, but only at the 10 percent level (in contrast to the 5 percent usually used as the cutoff). The second test looked at the correlation between stratospheric aerosol optical depth and the ENSO-removed TLT data, and led to the detection of stronger, more statistically significant results.

The authors concluded that their tests “suggest that internally generated variability could plausibly explain some of the observed tropical TLT changes after individual ‘small’ eruptions,” but they acknowledged that the volcanoes alone cannot explain the divergence between models and observations, with the discrepancy only reduced between 2 percent and 15 percent depending on the model used.

What Comes Next?

While temperatures have warmed modestly over the past decade, they have not warmed as much as had been projected by climate models. There are a number of factors that are not well-captured in climate models, like changes in trade winds, volcanoes, solar output, and ocean heat uptake. Collectively, they likely explain much of the discrepancy. It is also clear that there are still many unknowns surrounding how the climate behaves on a multi-decadal timeframe, and the drivers and impacts of types of natural variability are still poorly understood.

The vast majority of scientific articles examining the hiatus predict that it will be temporary and that the various factors contributing to it will abate in the next decade or so. As University of Colorado science writer Tom Yulsman is fond of saying, the climate bats last, and one benefit of studies that make short-term predictions is that we don’t have to wait that long to see if they are proven correct.

Photo: Eruption column at Mt. Pinatubo, June 15, 1991. Source: U.S. Geological Survey.

Zeke Hausfather

Zeke Hausfather, a data scientist with extensive experience with clean technology interests in Silicon Valley, is currently a Senior Researcher with Berkeley Earth. He is a regular contributor to Yale Climate Connections (E-mail:, Twitter: @hausfath).
Bookmark the permalink.

7 Responses to New Research Explores Role of Trade Winds, Volcanoes in Warming ‘Hiatus’

  1. Paul Quigg says:

    The global mean SAT anomaly graph shows the temperature trends from 1910 to 2014. An overlay of a CO2 concentrations graph for the same period would show a fairly steady growth during this time period. The same is true for CO2 emissions which correlate closely with concentrations. The temperature graph has a poor correlation with the concentrations and emissions graphs, especially from 1940 to 1980 during the “tiny” Ice Age. We are probably just going through another of these tiny Ice Ages. Overlay these three graphs and see what you get. Remember climate has a minimum thirty year time frame and we are getting bogged down in weather events.

  2. Rob Painting says:

    A few nits:

    -The IPO & PDO are not the same. The authors mention the IPO because it is the Pacific-wide manifestation of sea surface temperatures, whereas the PDO relates to only the North Pacific.

    - global warming hasn’t paused when one considers the energy building up in the ocean – which amounts to around 93% of heat accumulation in recent decades. Surface warming has progressed at a slower rate than previously. These are not the same.

    - the atmosphere doesn’t warm the ocean by transferring heat to it. The oceans are warming due to the greenhouse gas-forced warming of the top of the cool-skin layer of the ocean – where conduction is the dominant mechanism of heat transfer. Warm the top of the cool-skin layer and the thermal gradient through it is reduced. The slows the normal loss of heat from the (typically) warmer ocean surface to the (cooler) atmosphere above. So when sunlight enters and warms the oceans, the reduced thermal gradient in the cool-skin layer traps more of the sun’s energy – hence the long-term warming with increasing greenhouse gases.

    • Hi Rob,

      Thanks for the clarification on ocean heat transfer mechanics. I’m updating the text accordingly.

      As far as I can tell, the PDO and IPO refer to the same phenomena, but the main difference is the region being measured to produce the index.

      I agree that ocean heating has continued unabated, and we’ve discussed that in depth in some of our previous articles on the “hiatus”. This article was focusing on two papers specifically examining reasons for the discrepancy between GCM results and observations for surface and TLT.

  3. Nullius in Verba says:

    So, looking at your last figure, would it be fair to say that this paper claims the residual rise in TLT after removing ENSO/volcanoes is about 0.2 C over 30 years? (0.7 C/century.) And that the models are still rising about 2-3 times faster?

    I note that the hiatus still exists post-adjustment, and it is proposed that the reason for the difference is that the models don’t include any eruptions after 1999. Rather than saying that the observations are lower than they should be because of the eruptions, wouldn’t it be more correct to say that the models are higher than they should be because of the lack of modeled eruptions? After all, smaller eruptions are something of a constant background, and to be expected.

    Or is it claiming that eruptions have increased?

    • Eruptions have increased slightly in the last decade relative to the late 1990s:

      I agree that it is probably more correct to say that models are higher than they should be because of the lack of modeled eruptions. Though as Santer and colleagues point out including recent aerosol loading in models only accounts for up to 15% of the divergence (and as low as 2-5% in the NASA GISS models).

  4. HR says:

    WRT the trade winds paper why not reverse the wording here and say that during the recent warming period (1070′s-2000) trade winds were lower than normal and so therefore a greater amount of the energy could leave the ocean and this contributed to the warming or at least acknowledge this alongside the relative cooling effect during the hiatus?

    I see the wind anomaly has wandered quite far from zero now but is there any real reason to think the present phase represents a deviation from the norm anymore than the persistent positive phase seen during the recent warming phase, I don’t see any other period in the wind record were winds were so persistently weak (I’m looking at the darker black line here).

  5. HR says:

    Another quick question on the England paper. the maps you show from the paper cover the period 1992-2011. When you restrict the analysis to the hiatus period (e.g. post 1998) then the earth looks alot more patchwork with cooling (and warming in many regions of the earth (both NH and SH). The sense that the cooling is focused over Pacific is lost.

    I see 1992 maybe when tradewinds may have started dropping but they were anomalously high through to the late 90′s.

    BTW I used this tool ( ) to generate trends from various date. Combos (I tried) of start and finish that lie within the period of the hiatus (AND strong tradewind) belie the idea that cooling was primarily focused over the Pacific.