A Record Making Effort

Economist Figure

On Thursday March 31st Richard Muller of Lawrence Berkeley Laboratory gave evidence to the energy and commerce committee of America’s House of Representatives on the surface temperature record. Without having yet bothered to check, Babbage can say with some certainty that this event will be much discussed in the blogosphere—as, oddly enough, it should be.

Here’s the short version of the reason why: a new and methodologically interesting study, carried out by people some of whom might have been expected to take a somewhat sceptical view on the issue, seems essentially to have confirmed the results of earlier work on the rate at which the earth’s temperature is rising. This makes suggestions that this rise is an artefact of bad measurement, or indeed a conspiracy of climatologists, even less credible than they were before.

Now here’s the much longer version.

There are two topics which, more than any other, can be guaranteed to set off arguments between those convinced of the reality and importance of humanity’s impact on the climate and those not so convinced. One revolves around the question of how reliable, if at all, statements about average global temperatures before about 1500 AD are. This is the so-called “hockey stick” debate.

The amount of computer processing power and data storage capacity devoted to endless online discussions of the hockey stick— the subject featured in a great deal of the brouhaha over the “climategate” e-mails—must, by now, have the carbon footprint of a fair-sized Canadian city, which of course would worry one side of the argument not a whit.

The second touchy topic is the instrumental record of the world’s temperature over the past 100 years or so. This is a more genuinely interesting subject, for two reasons. First: Consider a person who looks at all the non-hockey-stick evidence and arguments for thinking people are changing the climate (we won’t rehearse them now, but here’s a relevant article from The Economist last year). Imagine this person then saying “you know, that radiation balance and basic physics and ocean heat content and all the rest of that stuff looks pretty conclusive—but because I can’t say for sure whether it was warmer in 1388 than it was in 1988 or the other way round I’m going to ignore it all.” This would probably not be a person you would take very seriously.

(It is because of this that the wiser sceptical voices in the hockey-stick debate do not claim that uncertainties over what, if anything, can reliably be said about mediaeval temperatures invalidate the scientific case for a strong and worrying human influence on climate. They say instead that there are a variety of statistical and other flaws in some of the reconstructions of mediaeval temperature, that some of the scientists responsible for some of these reconstructions have not behaved well, and that if that is typical of climate science then climate science in a whole is in a bad way. Thus the hockey stick becomes a sort of meta-, and indeed metastasising, argument.)

If, on the other hand, you imagine a person who has looked at all the other relevant material going on to say “You know, this is all very well—but there doesn’t seem to be any conclusive evidence that the world has actually been getting warmer in a significant or surprising way over the past decades,” you might well think hmm; if that’s the case then he has a point. Evidence that the world really is warming does seem pretty apposite to the whole issue. Being able to trust the records of what thermometers spread out over the world have actually measured and the procedures by which  those records are combined into a series of average global temperatures matter rather more than the hockey-stick.

Another reason is that mediaeval data (from tree rings and the like) at issue in the hockey stick debate are necessarily sparse and patchy, and coming up with really robust answers to all the relevant questions on the basis of them may well prove impossible. The data on which the contemporary surface temperature record is based, though, is rich. There are a great many temperature records in archives around the world. If you can choose records that are demonstrably reliable and combine them in an appropriate way, you should be able to get a pretty solid answer. This thus seems like an argument that could conceivably end in agreement on an important issue.

Indeed, most climate scientists would say that it already has. There are three different combinations of instrumental temperature records that seek to show average surface temperatures back to 1900 or earlier. Two are American, with one produced by the National Oceanic and Atmospheric Administration (NOAA) and one by NASA; the other one is British, with data from the Met Office and the University of East Anglia’s Climatic Research Unit (which was the epicentre of climategate). They used many of the same raw data, but the ways in which they adjust them to remove presumed artefacts and then combine them differ. Yet they come up with very similar answers, and when they publish their figures with error estimates they come within each others’ margins of error. The fact that three different groups agree in this way would normally seem to justify relying on the result.

But there are many ways in which climate science is not normal, one of which is that it matters a great deal with respect to some very expensive policy decisions. Various criticisms of the methodology and probity of the temperature records have been made, though much more often in the blogosphere than in the scientific literature. Erring on the side of extra caution is not a bad idea, and various efforts are underway to develop, corroborate and better to underpin the work on temperature records that has been done to date. One such effort is the Berkeley Earth Surface Temperature programme, which Dr Muller heads.

Fearless physicists

Dr Muller is an astrophysicist, not a climate scientist, and was indeed seen by some as being a bit of a sceptic, in the unfortunate negative usage of the word. He is strongly spoken in his criticism of some of the behaviour revealed in the climategate e-mails, and talks admiringly of some of the amateur or non-credentialed scientists who have mounted critiques of published climate science.

He also has a sort of intellectual fearlessness most often seen in physicists; when applied to other fields of endeavour this can look uncannily like a form of arrogance, perhaps because that is often what it is. The initials of the Berkeley Earth Surface Temperature project could be read in this light, and indeed they were so interpreted by some climate scientists, who got rather bit peeved at the idea that interlopers should presume to claim a priori they were the BEST. Any arrogance they may be prone to, though, doesn’t invalidate the fearless physicists’ insights. Dr Muller’s beloved mentor, Luis Alvarez, was quite right when he and his son argued that the death of the dinosaurs had less to do with the environmental or evolutionary challenges palaeontologists concentrated on and more to do with the damn great meteorite or comet impact for which Alvarez père et fils had just found dramatic and unexpected evidence. On the other hand Dr Muller’s subsequent variation on the Alvarez's now broadly accepted contribution, which led to an unsuccessful search for a distant planet that might be directing killer comets at the earth on a regular basis, has as yet come to naught.

The Berkeley approach seems based on the idea that coming out of physics, not climate science, was going to be a strength not a weakness. Rather than look at carefully (and similarly) selected subsets of the data it would look at everything available, just as astrophysicists frequently seek to survey the whole sky. Rather than using the judgement of climate scientists to make sense of the data records and what needed to be done to them, it would use well designed computer algorithms. Put together under the aegis of Novim, a non-profit group that runs environmental studies, the team gathered up a bit over half a million dollars—including $100,000 from a fund set up by Bill Gates and $150,000 from the Koch foundation, whose animosity towards action on climate change made the Berkeley project look yet more suspicious to some climate-change activists—and got to work. There was also support from the Department of Energy’s Lawrence Berkeley Lab, where Dr Muller and some of his team work. It is probably fair to assume that Steve Koonin, an undersecretary of state at the energy department with whom Dr Muller has served as one of the “Jasons”, a group of particularly intellectually fearless scientists which provides blue-sky and sometimes far-out advice to the defence department, and who has also produced a report for Novim, had an unofficial eye on what was going on.

Dr Muller’s testimony was not exactly the unveiling of his team’s first results—you can find him saying much the same in a seminar on the web— but it was a particularly high-level early outing. It was also a strikingly robust defence of the record as others have interpreted it. Calling the three extant groups “excellent”, Dr Muller described preliminary work by the Berkeley team on the overall magnitude and course of recent warming that backed them up. Instead of picking a relatively small number of weather stations to look at, this work simply took 2% of all the records the Berkeley group has access to at random. The results look very like what the other three teams have seen. The Berkeley team says that it has run such 2% experiments a number of times now, and the results are robust. The earth has warmed by about 0.7°C since 1957, just as the other teams claimed. Adjustments made to the data on a site-by-site basis which have had some suspicious sceptics hopping mad seem to have made no appreciable difference.

The Watts and wherefores

Dr Muller also, more controversially, reported on results that pertain to a specific point made by climate sceptics; that the temperature record is contaminated because many of the stations used to compile it are inappropriately located. This idea is particularly associated with Anthony Watts, a former television weatherman who runs an extremely popular website catering largely to a climate-sceptic crowd. Mr Watts has led an impressive crowdsourcing movement devoted to checking out the meteorological stations that generate climate data in America. This has found that a really surprising number of the instruments concerned are not sited in the way that they should be, being inappropriately close to buildings, tarmac and other things that could cause problems.

A compendium of Mr Watts’s concerns was published early last year by the Science and Public Policy Insitute, which specialises in airing doubts about climate science and policy, under the title “Surface Temperature Records: Policy Driven Deception?” Dr Muller’s answer to that question in front of Congress was pretty clearly no. The Berkeley team compared the data from the American sites Mr Watts thought were worst situated and the sites he thought best. It found no statistically significant difference in the trends measured in the two different categories, though the warming trend in the better sites is slightly stronger.

This analysis echoes one carried out last year by scientists at NOAA, which when looking at a subset of Mr Watts’s data found much the same thing. The Berkeley team’s result, though, is perhaps more striking, in that Mr Watts had made all his data available to Mr Muller and his colleagues, a step he seems now rather to regret.

Impressed by the Berkeley set up, Mr Watts wrote in a post published March 6th:

I’m prepared to accept whatever result they produce, even if it proves my premise wrong. I’m taking this bold step because the method has promise. So let’s not pay attention to the little yippers who want to tear it down before they even see the results. I haven’t seen the global result, nobody has, not even the home team, but the method isn’t the madness that we’ve seen from NOAA, NCDC, GISS, and CRU, and, there aren’t any monetary strings attached to the result that I can tell. If the project was terminated tomorrow, nobody loses jobs, no large government programs get shut down, and no dependent programs crash either. That lack of strings attached to funding, plus the broad mix of people involved especially those who have previous experience in handling large data sets gives me greater confidence in the result being closer to a bona fide ground truth than anything we’ve seen yet.

Responding to Dr Muller’s testimony, Mr Watts e-mailed that when he shared his data he “was expecting a study done by peer review, months out, not a job rushed in three weeks for political theater in the House of Representatives.” Though some of his results and conclusions were published in “Policy Driven Deception?” he has since worked on a peer-reviewed publication, sensible to the accusation often levelled at his blog that such publication is the way to do proper science. That this is not what the Berkeley team did, choosing instead to present preliminary results to Congress, has upset him, and he has asked to have a statement of his own read into the congressional record. The Berkeley team, for its part, argues that while it would rather publish in a peer-reviewed journal before discussing its results, and has “begun the submission process to do this”, an invitation to address the committee deserved what the team saw as the best available answers.

Mr Watts’s paper, on which he has a number of co-authors including climate scientist Roger Pielke Sr, is now said to be close to the end of its road towards publication. It claims to find that there is indeed a difference between the good and bad sites: though they may provide indistinguishable results for trends in average temperature, they differ when you look at trends in minimum and maximum temperatures, which has implications for the diurnal temperature range.

It is not clear what the climatological significance of this might be, and it is always worth bearing in mind that these data only apply to stations in America, which is a fairly small part of the planet.

Overall, the takeaway from Dr Muller’s presentation of his team’s data is that, in the words of one climate scientist, a “Koch-brothers-funded study confirms the previous temperature reconstructions.” Dr Muller says the team will now be looking into a number of other effects, including the bias that the “urban heat island” effect—cities are warmer than surrounding countryside—might have. The question of good versus bad location is linked to this (a good site can become bad as a city sprawls over it) and so is the issue of which records you choose to use (long records, preferred by earlier reconstructions, may be more prone to changing urbanisation around them). But there is more to the problem, and Dr Muller hopes to look into it further, as well as into issues that might arise from the times of day at which observations are made, stations moving from one place to another, and changes in the instrumentation used.

The Berkeley work, especially after it is published and disseminated in full, may increase the acceptance of the reality of global warming among people who have so far managed to maintain a comforting and sometimes self-serving feeling that maybe the people who deny that anything is going on are actually right. It doesn’t in itself show how much of the warming is due to human activity. Dr Muller, in a somewhat cavalier way, chose to suggest that about half of what had been seen since 1900 was. Other scientists would put the proportion higher.

Nor does it say how much warming is yet to come. Carbon dioxide and other widespread gases can warm the earth, but dust, smog, sulphate particles and other things can cool it. There is no very reliable record of how these cooling factors changed over the 20th century, but they must have played a role. So you can’t simply look at temperature changes over the 20th century, and scale them up according to the amount of carbon dioxide you expect, to find out what will happen next. Though Dr Muller, in his testimony, seems to differ on this—as might perhaps be expected for someone proud of a new contribution to the field, and hoping to make more—most climate scientists do not believe that further improvements in the accuracy of the average temperature record over the twentieth century will add greatly to their ability to predict the magnitude or timing of the changes to come.

But broader agreement that the temperature has indeed risen quite steeply over the past century is nevertheless a thing worth having. If the Berkeley team can help provide it, that is all to the good.

Comment policy: anyone who thinks this post was telling the truth about the fair-sized Canadian city mentioned near the beginning should please refrain from commenting.