The pause. The hiatus. The global warming slowdown. It’s all the buzz in some circles.

Did global warming stop 10 years ago? Or 15 or 18? Did warming of Earth’s surface? Can carbon dioxide be taken out of the woodshed?

Here are some of the lessons learned — or that should be learned — since the onset of the warming slowdown.

6. Data Models Matter

British astronomer Sir Arthur Eddington, whose observation of an eclipse in 1919 confirmed Einstein’s prediction of the bending of starlight by the Sun, once said, “Experimentalists will be surprised to learn that we will not accept any evidence that is not confirmed by theory.”

Which is to say that when observations, or experiments, run counter to theory, the theory isn’t necessarily at fault. Sometimes the complex models that process the raw, observed numbers and turn them into useful data are wrong, inadequate, or insufficient. So scientists inspect both sides of the ledger.

One recent example involves the claim of faster-than-light neutrinos from the CERN particle laboratory in Geneva; turned out, a bad cable connection led to a spurious result, and theory remained intact. Another was the mid-1990’s cooling trend observed in tropospheric temperatures as measured by satellite microwave sensors, which contradicted signs of warming seen from surface measurements (and expected from climate models). After some corrections to the data model that translates microwave measurements into tropospheric temperatures, the observations came into much closer agreement with expectations.

There’s little doubt that the surface temperature data produced by NASS GISS, the Hadley Centre and NOAA show a slowdown of surface warming, with 15-year trends only a third of what they were in the middle of the previous decade. (It’s often forgotten that surface warming was then running ahead of expectations. Call it the “anti-hiatus.”)

All three of these data models make assumptions about how they treat areas without adequate temperature station coverage: regions in Africa, Antarctica and especially the Arctic. A surface dataset introduced by Kevin Cowtan and Robert G. Way in 2013 — in which they incorporate satellite data for the lower atmosphere and a different method of in-filling the regions devoid of stations, especially at high latitudes — finds a surface trend over the past 15 years to be about 50 percent higher than HadCRUT4 had shown.

But that still leaves half the slowdown unexplained. On the other hand, a just-published study by scientists from the Danish Meteorological Institute points to lower latitude trends as the dominant cause of the global temperature hiatuses (see Cowtan’s detailed response).

Either way, there’s no doubt these trends are below what climate models had projected for warming rates, which at this point in climate change is — absent natural variability — about 0.2 degrees C, a bit more than one-third of a degree F, per decade.

And there’s also uncertainty about satellite measurements of the lower troposphere. Data from a research group at the University of Alabama at Huntsville shows 0.15-0.20 degrees C of warming over the past 15 years, while the numbers from Remote Sensing Systems show no warming at all. John Christy, leader of the UAH research group, says:

While I’m fairly sure RSS overcorrects for the spurious warming drift of [satellite] NOAA-15 since 1998, it should be remembered that UAH applies no correction for that satellite (we will in version 6.0). So our trend since 1998 (when NOAA-15 was launched) is likely a little too warm. My suggestion would be to average the two together.

Ignoring this current mismatch by citing only the RSS results simply isn’t copacetic.

5. Natural Variability (Still) Matters

There have been warming slowdowns before. A study published last year in Nature by James Risbey and colleagues found five historical periods where the 15-year intervals saw negative trends and/or trends near the bottom of the range of climate model calculations: periods centered on 1890, 1905, 1945, 1970, and 2005.

But they also found 15-year periods where trends were near the top of model projections: 1925, 1935, and 1955. They wrote:

In other words, the recent “hiatus” centered about 2005 (1998-2012) is not exceptional in context. One expects the observed trend estimates…to bounce about within the model trend envelope in response to variations in the phase of processes governing ocean heat uptake rates, as they do.

It’s easy, and all-too-common, to draw fallacious conclusions from intervals that are overly influenced by natural factors, which is why scientists such as those with the World Meteorological Organization usually consider 30 years the minimum period over which climate trends can be reliably detected. Anything less is too dependent on natural internal variability in the climate system.

El Niños and La Niñas, which can cause temperature swings of 0.2 degrees C in a year or two, mask or amplify the warming from fossil fuel emissions. Since the monstrous El Niño of 1998, there have been seven La Niña seasons, two of them strong, and four El Niño seasons, none strong. The predominance of La Niña conditions has tended to cool Earth’s surface.

A 2011 study by Foster and Rahmstorf that controlled for El Niños and La Niñas found a continued, steady rise in surface and lower tropospheric temperatures of 0.14 to 0.18 degrees C per decade.

A string of relatively small volcanic eruptions — small in comparison to major cooling events like 1991’s Mt. Pinatubo eruption — account for up to 15 percent of the current warming slowdown. One recent study found that satellite readings miss a significant amount of volcanic aerosols, so their cooling effect was underestimated. In that study, David Ridley and colleagues found this translated into an estimated global cooling since 2000 of 0.05 to 0.12 degrees C — up to more than half a decade’s warming from greenhouse gases. A January 2015 study by Benjamin Santer and colleagues found fingerprints of late 20th and early 21st century volcanic activity in a host of climate variables, such as sea surface temperatures, tropospheric temperature, and atmospheric water vapor.

Very recent work provides insight into how the ocean cycles in the Atlantic and Pacific oceans that take decades to complete (and may even be partly related) seem to have aligned to create the current “false pause” or slowdown. Steinman, Mann and Miller found that when the temperature influences of greenhouse gases, volcanoes and changes in solar output are estimated from climate simulations, and then subtracted from the actual temperature trends in the northern hemisphere since 1880, what’s left — which should be only internal oscillations — aligns closely (see figure below) with the periods of overheated trends like were observed twice last century (from about 1910-1945 and 1975-2000), and the slowdowns of 1945-1975 and that of the last 15 years.

It’s more evidence that internal climate variability has been acting recently to partially offset anthropogenic global warming.

Estimates of the effect on surface temperatures of the variation of the Atlantic Multidecadal Oscillation (blue), the Pacific Decadal Oscillation (green), and the newly defined Northern Hemisphere Multidecadal Oscillation (NMO; black). Shading indicates uncertainties. The NMO aligns with recognized warming and slowdown periods in the temperature record. Used with permission from Michael Mann and Realclimate.org.

As our emissions of heat-trapping gases continue to rise, natural variability over time will matter less. One study found that at current greenhouse gas emissions trends, there is little chance of a decade-long hiatus after 2030. Another recent study in Nature Climate Change by researchers at the Met Office Hadley Centre in the U.K. used an ensemble of climate models to find the probability of a 10-year hiatus driven by natural variability to be about 10 percent, and only 1 percent for a 20-year hiatus. But given the 15-year hiatus that’s already here, the chance it could extend to 20 years is between 0 and 25 percent, with a best estimate of 15 percent.*

So we shouldn’t be surprised, they say, if the current slowdown extends until the end of this decade, and they advise that “there is an increased likelihood of accelerated global warming associated with release of the phase of decadal variability in the Pacific Ocean.” In fact, 2014 was the first year in the past eight that the Pacific Decadal Oscillation was above zero, suggesting this natural cycle may be flipping back to its positive, warming stage.

4. Climate Models Don’t Do Short-term Predictions

Climate modelers don’t know the future — if they did, there’d be no need for their models.

In particular, they don’t know what volcanoes will erupt in the next decade or two, or how the Sun’s irradiance will change, or what El Niños and La Niñas will occur. And models cannot yet predict when these ocean cycles — and longer cycles like the Pacific Decadal Oscillation and Atlantic Multidecadal Oscillation — will occur naturally.

And that’s not all. Aerosols, tiny particles that float in the air, have a major effect on climate. They are emitted by volcanoes, forest fires, microalgae; about 10 percent of them come from burning fossil fuels. In reflecting sunlight, their latitude matters a lot.

“The CO2 forcing and other greenhouse gases are quite predictable,” says Gavin Schmidt, director of NASA’s Goddard Institute for Space Sciences (GISS), “but aerosol forcings are insufficiently constrained.”

All these factors can influence long-term climate trends for a year, a decade, or longer. So climate models in reality can be wrong from the day they offer up their output. But climate models aren’t intended to calculate short-term variability — at least not yet. Instead, they use energy conservation over the long haul — many decades to centuries — over which these shorter-term effects average to, or decline to, approximately zero.

So climate models can only do “projections” — calculations based on a set of assumptions — and they can’t do predictions. They can’t plot the exact path climate change will follow.

They can tell the likelihood of where climate will end up — with uncertainties — but not how we’ll get there. Especially for the first few steps out the door.

3. Understanding Slowdown: Classic Illustration of How Science Works

Science advances by scientists’ designing and building models, calculating their consequences, keeping what goes right, and revisiting what goes wrong. It’s a self-correcting process; not even Einstein’s science came out fully formed, perfect from the beginning.

The response to the warming slowdown is a perfect example. It was not foreseen when it began around the start of this century. But once the data came in, climate scientists began reexamining their ideas and looked for explanations, examining lots of cracks and crevices where heat might have gone, or wondering if their models were projecting too much trapped heat from greenhouse gases. In turn, the IPCC’s Fifth Assessment Report reduced the lower limit of climate sensitivity — how much the surface will ultimately warm from a doubling of carbon dioxide in the atmosphere — from 2.0 degrees C to 1.5 degrees C (3.6 degrees F to 2.7 degrees F).

The slowdown has also highlighted the need for more observations, especially of aerosols and in the ocean depths below 2,000 meters, depths not currently visited by the robotic Argo buoys. That’s given rise to NOAA’s Deep Argo program, now being prototyped.

Kevin Trenberth, a climate scientist with the National Center for Atmospheric Research in Boulder, Co., in 2013 told Yale Climate Connections “One of the things emerging from several lines is that the IPCC has not paid enough attention to natural variability, on several time scales.” In response, climate modelers will do what they’ve always done: try to find if and where their models need improvements — better representations of the physics and better approximations — and incorporate these into their models.

2. Global Warming is Ocean Warming

“One could say that global warming is ocean warming,” wrote NOAA’s Gregory C. Johnson and John M. Lyman in an October commentary in Nature Climate Change. Average surface temperature is only one indicator of climate change — albeit traditional, because we live on the surface. But it’s becoming clearer that most of global warming is happening beneath the waves.

More than 93 percent of the heat trapped by greenhouse gases since 1970 has gone into the ocean, and only about 2 percent into the atmosphere. The surface and tropospheric temperature records are full of noise; in contrast, the ocean is much less noisy. And its top half has been warming steadily (see figure below).

Changes in global ocean heat content for the top 2,000 meters of the world ocean, relative to the 1955-2006 average; data from S. Levitus et al., “World Ocean heat content and thermosteric sea level change (0-2000 m),” 2012, 1955-2010. Geophys. Res. Lett., 39, L10603 (2012). Updated data from the National Oceanic and Atmospheric Administration/National Oceanographic Data Center. Credit: NOAA/NODC.

A study published in Science magazine last August examined millions of ocean temperature measurements made since 1970, and found that, over the past decade and a half, Atlantic Ocean water below 300 meters has stored more energy than the rest of the world’s oceans combined. “We found the missing heat,” said co-author Xianyao Chen.

“This situation cannot go on indefinitely,” says Andreas Schmittner, a climate scientist at Oregon State University. “At some point the subsurface warming will find its way back to the surface. It is possible that this is what’s happening right now and why the last five months have been warmer.”

This ocean warming in the upper 2,000 meters (1.2 miles) averages about a half-Watt per square-meter (think of a Christmas tree light), and shows the Earth still very likely has an energy imbalance — more energy coming in from the Sun than leaving outward through the top of the atmosphere. (One would need to know the heat content trends in the deep ocean to be sure.)

Global sea ice volume has decreased. Sea level continues to rise. But natural variability still plays its role. The slowdown in surface warming may continue for several (or many) more years, driven by wind shifts in the Pacific Ocean, volcanic eruptions large and small, a slightly weaker sun, and long-term ocean cycles.

1. Manmade Climate Change Hasn’t Gone Anywhere

Notably, 2014 was one of the warmest years on record, as measured at the planet’s surface. This reality is all the more impressive because no El Niño has fully formed. Sea surface temperatures smashed their old record, and upper ocean heat content continues to rise steadily each year.

Climate change is going to be a very long ride, and there will no doubt be more surprises along the way. Questions begging firm answers are major — Will a doubling of atmospheric carbon dioxide lead to 2 degrees C of warming or 4 degrees C? — but surface warming is not guaranteed to be monotonically upward. Perhaps someday even members of Congress will understand this.

*Correction made on 3/11/15.