Tim Lambert (deltoidblog AT gmail.com) is a computer scientist at the University of New South Wales.
John Quiggin has another post on the right wing attack on science, this time describing the Australian front. Chris Mooney has great article in the The American Prospect about James Inhofe's part in the attack on science.
And Iain Murray is at it again. He has a post where he refers to graph on the left, saying that it is one of the most important elements in the debate, and writing:
"The fact that the ten hottest years happened since 1991 may well be an artifact of the collapse in the number of weather monitoring stations contributing to the global temperature calculations following the fall of communism (see graph)."
As I've said before, I'm reluctant to comment on global warming because many others are better informed on the matter. But even though I'm not an expert, it took me all of ten seconds to think of way to test to see if the increase was an artifact of the change in the weather stations reporting. All you have to is produce another graph of average temperature just using the weather stations that have data for the whole period. If this graph shows a similar increase, then Murray's suggestion is proven false. If it doesn't show an increase, then Murray's suggestion is proven true. And if you have the data to produce this graph, then you have the data to produce the graph that tests his suggestion.
There are three possibilities:
- Murray didn't think of this really obvious test. In this case he isn't competent to write about global warming.
- The test was done and Murray knows that it showed that his suggestion was false. In this case it would not be honest for him to present his suggestion the way he did.
- The test was done and Murray knows that it showed that his suggestion was true. If this was this case, why wouldn't he say so?
Update: In comments, Christopher Enckell provides the source of the graph Murray showed: a paper by Ross McKitrick. McKitrick writes:
Figure 3 shows the total number of stations in the GHCN and the raw (arithmetic) average of temperatures for those stations. Notice that at the same time as the number of stations takes a dive (around 1990) the average temperature (red bars) jumps. This is due, at least in part, to the disproportionate loss of stations in remote and rural locations, as opposed to places like airports and urban areas where it gets warmer over time because of the build-up of the urban environment. This poses a problem for users of the data. Someone has to come up with an algorithm for deciding how much of the change in average temperature post-1990 is due to an actual change in the climate and how much is due to the change in the sample. When we hear over and over about records being set after 1990 in observed global temperatures this might mean the climate has changed, or it means an inadequate adjustment is being used, and there is no formal way to decide between these.
I'm stunned. As I wrote above, it took me ten seconds to think of way to test if the increase was due to a change in the sample and McKitrick writes that "there is no formal way to decide". It would appear that my possibility 1 applies to both Murray and McKitrick.
Comments
Doesn't your argument depend on the (probably right but not explicit) assumption that USSR temperatures are positively correlated with those in the rest of the world? One should also consider the error bars, I think.
Posted by: rilkefan | April 22, 2004 2:31 PM
Think you're being a bit harsh here Tim (one of those "false"s above ought to be "true" as well, shouldn't it?) [Yes. I fixed it. TDL]
Take a look at the crappy aliasing on the chart in the .pdf file Murray links to. It's completely different from the print in the rest of the newsletter.
This suggests to me (from long experience of this sort of thing) that the chart has been cut 'n' pasted from another .pdf file and then resized to fit the space. In other words, Iain Murray doesn't have access to the underlying spreadsheet or the data, so he's guilty of passing on bullshit rather than making it up.
Posted by: dsquared | April 22, 2004 5:01 PM
I'm commenting from memory, but I don't think the decline in weather station numbers was due to the collapse of communism.
The general point Murray is making is similar to claims about "urban heat islands", which have been tested in exactly the way you suggest and found to be unimportant.
Posted by: John Quiggin | April 22, 2004 5:44 PM
please visit Bizarre science where it is explained for you
Posted by: Louis Hissink | April 22, 2004 6:11 PM
Using five long range stations in Europe (carefully selected for heat island trends) it shows that (at least for europe) there is indeed a jump in temperature. If it is anthropogenic is a completely different topic.
graph:
Posted by: Hans Erren | April 22, 2004 6:48 PM
Hans,
So? then these data also must be considered - http://www.john-daly.com/stations/mawson.gif
Which is correct?
Posted by: Louis Hissink | April 22, 2004 6:58 PM
dsquared, Murray said that the source for his graph was Essex and McKitrick "Taken by Storm". According to Louis what Murray actually did was combine a graph of the number of weather stations from E&M with temperatures from some other source. They don't seem to have reported the result of the obvious test, so possibility 1 applies to Murray, while one of possibilities 1,2 or 3 applies to E&M.
I was able to locate and download the data set in 15 minutes. I guess that doing the test would take another hour or so.
Posted by: Tim Lambert | April 22, 2004 8:38 PM
Louis, milder winters in the northern hemisphere will not show up in antarctica.
Posted by: Hans Erren | April 23, 2004 3:04 AM
The graph with weather stations shown with surface temperature does indeed exist. It was used in a presentation by McKitrick to the Department of Economics Annual Fall Workshop, The University of Manitoba, November 7, 2003.
You will find it at: http://www.uoguelph.ca/~rmckitri/research/econ-persp.pdf
That said, it does show correlation, but not necessarily causation. But it invites to study the quality of the surface temperature network more carefully than has been done so far.
Posted by: Christopher Enckell | April 23, 2004 12:28 PM
Tim,
I suspect Murray used the graph which Christopher Enckell points to, but probably, as we all do, from time to time, misattributed it. No big deal - so I am quite happy that the graph exists, since I also decided to extend my search to other papers Essex and McKitrick might have written subsequent to their book.
As for Murray actually did, I never said that he actually did, I suggested he might have collated temperatures from Taken by Storm and the station data, and combined both in one graph. Clearly Murray did not, since the source of the graph has been identified.
Posted by: Louis Hissink | April 23, 2004 1:26 PM
Hans, Of course not, how could it - and of course eliminating heat island effects is rather subjective - so perhaps you should point to the data and we proceed from there.
Posted by: Louis Hissink | April 23, 2004 1:26 PM
Tim,
This is what I wrote
"However Murray seems to have collated two disparate data sets to good effect, and one concludes that the Mann Hockey Stick graph, which seems to mimick this graph, may well be another artefact of the reduction of GHCN stations, rather than a result from a set of statistically representative samples."
So what Murray "actually did" is quite inaccurate - he didn't. I suspected he might have - guessing on my part of course.
Posted by: Louis Hissink | April 23, 2004 2:29 PM
There's an interesting post waiting to be written about the McKitrick-McIntyre attack on Mann. According to their account, the original paper was commenced in summer 2003 and published in (I think) October of that year. Mann et al replied almost immediately pointing out all manner of problems with the MM critique.
MM made a big fuss about problems they'd had getting to the data set and said (November) that a substantive rejoinder had been circulated for comment and would be released soon. Similar promises have come out at steadily increasing intervals, but, as of late April, no actual response.
Posted by: John Quiggin | April 24, 2004 5:30 PM
Chris Enckell wrote:
But it invites to study the quality of the surface temperature network more carefully than has been done so far.
It has been quite carefully studied, many times, in fact.
D
Posted by: Dano | April 25, 2004 9:18 AM
John Q, do you mean the bagpipe lessons, rather than the trend analysis? Very interesting indeed.
D
Posted by: Dano | April 25, 2004 9:20 AM
The bagpipe lessons were a nice touch. Perhaps the rejoinder will be made with claymores.
Posted by: John Quiggin | April 25, 2004 2:40 PM
So, Tim, have you had a chance to analyze the data? If not, I volunteer a few hours of my time and my 23 years of experience as a practicing statistician to help.
Posted by: Paige | April 26, 2004 2:44 AM
Yes, I've analyzed it. You still see an increase in the 90s. I'll have a post on it soon.
Posted by: Tim Lambert | April 26, 2004 5:52 AM
The answer to what happened is, I believe, simple. To understand though, you have to know a bit about how global surface temperature records are constructed.
There are two acknowledged surface temperature records, the one at GISS http://www.giss.nasa.gov/research/observe/surftemp/ and the one at the Climate Research Unit http://www.cru.uea.ac.uk/cru/info/warming/
Both show rises of ~0.4 degrees since ~ 1950 and in both the rise starts at ~1970
Both start by selecting stations at which temperatures are measured by an agreed method with calibrated instruments. They then decide on a base period and find an average temperature for EACH STATION over that base period. Following this, one subtracts the "average annual" temperature (those quotes hide a lot of detail)from the base period average to find the ANNUAL ANOMOLY for EACH STATION.
Now you divide the surface up into equal area blocks and average the anomolies in each block. Finally you average over the blocks to get a global average temperature.
You are measuring differences between the temperatures at an individual station against some base period average for that station. As long as the area around the station stays the same the anomoly for that station will not be affected by urban heat island effects. While the temperature read on a thermometer in an urban area will be higher than in some nearby rural area, the anomolies will be close to each other.
The effect seen in the chart above is almost certainly a selection for higher temperature stations (I assume the data in the chart is correct), probably more urban, less rural. If you compare the graph above to the GISS and CRU graphs, you see that in the latter two there is no jump at 1990, which is what one would expect if the anomoly method were robust. It is.
The three graphs taken together are rather strong proof that the observed rise in the surface temperature record is not controlled by urban (or suburban) heat island effects. If they were, the jump at 1990 would be in all three.
The Murray piece is a target rich environment. I especially liked the part about Sir John Houghton not contributing to climate science.
Posted by: Eli Rabett | April 26, 2004 11:08 AM
Well said Eli.
Note also the AGU has calculated the cooling of the globe by volcanism in the 1900s as ~ 0.75 deg F, and the recent papers attempting to quantify the cooling effect of land cover change in the US (which has risen less than many other parts of the planet).
Posted by: Dano | April 26, 2004 1:03 PM