A cautionary tale about accuracy in science journalism, and jumping the gun on official climate reports
Portrait of a warming planet. (Animation of false-color images acquired by the Himawari-8 satellite on Feb. 18, 2016. Note Cyclone Winston swirling on the right. Source: NOAA/RAMMB/RAMSDIS) This past February was plenty warm, and the Arctic was particularly so. In fact, measurements taken at Earth's surface may ultimately show that last month was warmer than any previous February in records stretching back more than a hundred years. But unlike what you may have read in recent days, we don't really know that yet. And claims that this shows global warming has gone into overdrive are overblown.
Source: Roy Spencer/University of Alabama True, February apparently was the warmest such month ever in the satellite record. (See thumbnail at right.) And as my colleague Andrew Freedman points out at Mashable, this deals a setback to people, including some presidential candidates, "who frequently cite the satellite record of atmospheric temperatures as evidence that human-caused global warming either doesn't exist or is far smaller than scientists claim." But this record stretches back only to 1979. And satellites monitor temperatures up in the troposphere, not on the surface where we live. Official analyses of data collected during February at thesurface of the Earth — traditionally the scientific gold standard for measuring global warming — have not yet been completed. That did not stop Slate from claiming in an article widely circulated on social media that surface warming in February broke all records, showing that "global warming is going into overdrive." Maybe. But maybe not. To make his case, Eric Holthaus, the author of the story, chose analyses that turn out to be based on an inherently unreliable dataset for evaluating long-term climate change. Moreover, the story inaccurately reports that El Niño likely gave only a small boost to warming last month. Actually, in all likelihood it gave quite a bit more of a boost than Holthaus claims.
During years with conditions similar to El Niño (red), global temperatures tend to get a boost. At the same time, El Niño episodes are tending to spike higher as the long-term warming of Earth's climate continues. (Source: NOAA/NCEI) I originally didn't intend this to be a criticism of that one story. That's because as someone who has been reporting on climate change since 1983, I'm fully aware how challenging this beat can be. And I'm sure my own reporting sometimes has left much to be desired. I carry on in this beat because I believe my job as a journalist is to bear witness to events, to help explain them, and to put them in proper context. I'm sure that this is also what motivates Eric Holthaus, a journalist who has done excellent work. But if in bearing witness as journalists our testimony becomes exaggerated or even inaccurate, then the credibility of journalism in general suffers. In addition to being a journalist, I'm also a professor of journalism. So I think I have a particular responsibility to focus on issues like this. That's why I decided to take a detailed look at this one story, and to try to set the record straight — as a cautionary tale for all journalists who struggle to cover this complex topic. For his story, Holthaus relied on unofficial data and two equally unofficial analyses using something called the Climate Forecast System Reanalysis, or CFSR. Here's a key excerpt from his story:
Perhaps. But aside from one quote in the story, there's no evidence of independent, in-depth reporting to substantiate the use of words like "shocking," "shatters," "overwhelming," and "overdrive." Remember, these are two unofficial analyses. Should we trust them? The answer I've come up with through my own reporting — which included consultation with climate scientists — is an emphatic "no." To begin with, I contacted Kevin Trenberth at the National Center for Atmospheric Research. Concerning the CFSR data and methodology Holthaus relied on, Trenberth said this:
The CFSR is one of the worst to track climate. It is an operational product and very unreliable.
That's because there have been times when it went "unstable," and these issues have not been corrected. Moreover:
[CFSR] has "grid point" storms. It has discontinuities over time associated with changes in the observing system (satellites in particular). And the surface temperatures are not real. They do not use surface data to estimate surface temperatures. Instead [temperatures come] from the model based on an energy balance which depends on model clouds that are not realistic.
I don't think it matters much if you didn't fully follow that. But you should get this one key point: At least one prominent scientist believes this is not how a determination about global temperatures should be made.
Gavin Schmidt, head of NASA's Goddard Institute for Space Studies — one of two U.S. agencies that produce monthly climate analyses (the other is NOAA) — also has reservations about using CFSR in the way that Holthaus did for his story.
There are quite a lot of data online that allow people "to make pretty good inferences as to monthly or annual or even decadal anomalies," Schmidt said in an email. But there are pitfalls to having so many different sources of data and methods for analyzing them:
. . . what we are starting to see is a lot of ad hoc reporting based on one dataset and one methodology for one month, and then something completely different a month later, and something else again for month three — all with the commonality that all are records of some sort. This I think can give a misleading impression over time.
One journalist at Slate has now chosen one unofficial and inappropriate dataset to support his conclusions. As a result, readers now have the impression that February was record-setting — even though we can't reach that conclusion yet.
And what will happen if the forthcoming official analyses by NASA and NOAA, carried out with reliable data and methods, show something different? People could conclude that scientists have no idea what they're talking about. Which is clearly untrue. Or that journalists who cover climate change are clueless. They'd be wrong about that too, but I wouldn't blame them for thinking it.
And that's why I felt motivated to write this piece.
In his story for Slate, Holthaus acknowledges that we're still in the midst of a record-tying El Niño, which ordinarily gives a boost to global temperatures. But among other things, he argues that since the Arctic experienced much warmer temperatures than normal in February — and that El Niño doesn't affect the Arctic much (which may be untrue) — the phenomenon likely played only a small role in the warming that occurred globally during the month.
The blue line shows the trend in the geographic extent of Arctic sea ice during the cold season so far. Lately it has been trending at a record low level. (Source: NSIDC) Here's the relevant portion of Holthaus's story:
But El Niño isn’t entirely responsible for the absurd numbers we’re seeing. El Niño’s influence on the Arctic still isn’t well-known and is likely small. In fact, El Niño’s influence on global temperatures as a whole is likely small—on the order of 0.1 degree Celsius or so.
What of that first claim, that El Niño's impact on the Arctic is likely small? The NASA article Holthaus refers to in the first link doesn't make that statement at all. The article quotes Richard Cullather, a NASA climate modeler as saying:
Although we haven’t been able to detect a strong El Niño impact on Arctic sea ice yet, now that the ice is thinner and more mobile, we should begin to see a larger response to atmospheric events from lower latitudes.
How does that support the contention that El Niño's impact on the Arctic "is likely small"? It doesn't. In fact, research detailed in a 2014 paper
published by Trenberth suggests that what happens in the tropics can indeed influence what happens in the high latitudes, including the Arctic. The research shows that when anomalies in sea surface temperatures arise in the tropical Pacific, there can be far reaching impacts on global atmospheric circulation. These anomalies can arise from a long-lived El Niño like phenomenon called the Pacific Decadal Oscillation
, as well as El Niño itself. And, in fact, these changes to atmospheric circulation can exert "strong influences into the Arctic," Trenberth told me in an email message. This could well be the subject of scientific debate. Which is why it's necessary to interview experts before claiming in a story that El Niño's impact on the Arctic is "likely small." Maybe yes. But there is credible research to suggest that El Niño might have played a significant role in what we've been seeing in the Arctic this winter. What about the claim that El Niño's influence on global temperatures was also modest? If you click on the link Holthaus uses to back up his claim, it takes you to a well-reported article at Carbon Brief
— a story that ultimately fails to support it. The Carbon Brief story does quote Adam Scaife, head of the long-range forecasting division of the U.K. Met Office, as saying that El Niño probably played only a small role — in 2015. But then Scaife goes on to say that El Niño's impact on global temperatures usually peaks in the second calendar year, which is now. Here are the relevant portions of Scaife's remarks:
We think El Niño made only a small contribution (a few hundredths of a degree) to the record global temperatures in 2015.
But . . .
The effects are very clear: there is a little warming in the period preceding the winter El Niño peak, but the big effect on global temperature comes in the following calendar year as it takes a few months for heat to increase in other ocean basins around the world.
Scaife concludes that about 25 percent of the warming expected during 2016 is likely to come from El Niño. That's pretty significant. Had Holthaus interviewed Scaife, rather than relying on someone else's story (and then misreading it), maybe he would have avoided this error in his story.
NCAR's Kevin Trenberth also addressed El Niño's influence on global temperature. He focused on what's known as the Niño 3.4 index.
"Niño 3.4" refers to a region that encompasses the central Pacific along the equator. And the index is a measure of warming in that region. When the index passes a certain threshold for a long enough period, NOAA considers an El Niño to be in progress. According to Trenberth, for every 1 degree Celsius change that occurs in the Niño 3.4 index, there is about a 0.1 degrees C change in global mean surface temperature. "But there is, on average, a three-month delay," Trenberth notes.
The Niño 3.4 index began to peak at 2.3 degrees C back in November. (It's now down to 2.2.) Given the three month delay, this means El Niño's peak effect on global average surface temperatures should have occurred in — you guessed it — February, or thereabouts.
Doing the math, we come up with an estimated El Niño boost to global temperatures of about .23 degrees during the month. I wouldn't call that insignificant.
Moving beyond one particular story in Slate, there is the question of whether journalists like Holthaus and myself should be rushing to publish these month-by-month accounts of global temperatures in the sometimes breathless way that we do.
I think the official analyses published by major agencies like NASA and NOAA are indeed legitimate news stories that should be reported. These are your tax dollars at work. And arguably, you should have an opportunity to learn what knowledge those investments are producing.
But in our coverage, we need to be very careful not to conflate short-term variations in weather with long-term climate changes, which by definition occur over the course of decades.
As NASA's Gavin Schmidt puts it:
. . . individual months or very short-term excursions are not predictive of longer term change, and when they turn down again (relative to where we are now) there’ll be some confusion and (perhaps justified) accusations of cherry-picking and bias.
Teasing out underlying long-term changes in the climate system from the noise of shorter-term natural variations can be challenging for scientists. So I don't think this is something a journalist should be trying to do by himself — especially not with an unreliable set of data and a reliance on Tweets and links to other journalists' articles instead of interviews and skeptical reporting.
I think it is fair to say that most journalists who cover this topic are ultimately motivated by a desire to help people understand the changes our planet is undergoing. But unless we slow down a bit, go back to journalistic basics, and keep the old adage that "if your momma' tells you she loves you, check it out," the result can be the opposite of what we intend.