A Cautionary Note on Interpreting the BEST Global Warming Data November 4, 2011
Posted by geoff in News.trackback
A UC Berkeley professor assembled a team and reviewed the land temperature data since 1800. He told the world that after reviewing the data he was no longer skeptical of global warming claims, and that global warming was real and significant. But the Global Warming Policy Institute (GWPI) got hold of the data and said that it proved that there has been no warming since 2001. The Daily Mail then published this graph based on the GWPI’s results:
"The Inconvenient Truth" should be "The Lame Data Analysis by GWPI Clowns"
You may have seen this running around the internet – Hot Air had it up today. It’s been pretty popular on conservative blogs.
The thing is, it’s wrong. Just wrong.
Do you see that weird data point at the end of the lower graph – the one that spikes way down? Turns out that it was based on a tiny number of stations, all of them in the Antarctic. The uncertainty associated with the point is more than 10 times the normal uncertainty for this data. And that one point is enough to mess up trend analysis.
If you exclude it, as you certainly should, there is a positive trend of 0.14 K/decade over that time period. The trend is pretty sensitive to the starting and ending points – but for any time period extending farther back than about 5 years, it’s definitely positive and significant (and most intervals in the 10-year range give a 0.2 – 0.25 K/decade increase).
Over the past few years the trend has flattened and even shows a decline for some time intervals, but it’s too early to say whether that’s a real change in climate trends. So we’ve got to wait.
In the meantime, don’t stick your neck out over the GWPI/Daily Mail data interpretation.
Has there been any urbanization around or upwind of a significant number of the data collection stations since 1800? If so what affect would that have on the assumptions?
Yes, Joe, there has been an “uban heat island” effect that affects climate data from individual weather stations. Climate researchers attempt to statistically adjust (“normalize”) the data to account for this, but that opens up the possibility that their “normalization” was biased by their assumptions about what the outcome should be. AGW skeptics have uncovered numerous examples of apparent bias. Nobody knows the extent of such bias. Somehow, most of the original data went missing at the University of East Anglia Climate Research Unit.
It’s my understanding that NOAA now provides climate data from satellites that, so far as I know, is very accurate. The problem is, the satellite data is all relatively recent, and not directly comparable to historical data from terrestrial weather stations, so it does not help much to establish the extent of long-term warming.
Unless, of course, NOAA embarks on some “normalizing.” Whoomp, there it is!
The BEST team looked at the heat island effect, and found that it was tiny. I’m not sure I’m thrilled with the way they did their analysis, but I will probably never have enough time to do it myself.
In any case, the point of this particular post is that you can’t simply take the BEST data and claim that it shows that the warming trend has flattened. The data may or may not be any good, and the trend may have flattened, but contra GWPI & the Daily Mail, this data does not show that trend.
I don’t get your point geoff. Probably because I’m a moron but the mean may still be above the 30 year average but the trend is still essentially flat especially given the increase in C02 over that time which I believe was the Daily Mail’s main point.
but the trend is still essentially flat
It looks flat, but it really isn’t. If you do the least squares fit, you end up with 0.14 K/decade temperature increase. It’s a case where you can’t trust your eyes – you have to crunch the numbers.
you have to crunch the numbers.
http://tinyurl.com/3gfwts4