U.S. temperatures and the BEST study

For those who have been following my review of the temperatures around the United States, there was an interesting set of papers released by the Berkeley Earth Surface Temperature (BEST) study in the last week. Four sets of findings have been released, based on:
1) statistical methods
2) Urban Heat Island
3) Station Quality
4) Decadal Variations.

For the moment I am most interested in the second of these, since it relates to the effect of population around the measuring station on the resulting temperature reported. This is not quite the same as the UHI effect, although they are both generated by the same factors.

In the paper, to greatly over simplify, the authors appear to have looked at the trends in temperatures in urban and rural sites, and then looked at the difference between the trends in temperature since 1950 for each. They report that the overall trend (i.e. temperatures in urban areas – temperatures in rural areas) is negative rather than the expected positive. They find an average negative trend of 0.342 deg F per hundred years. (Or 0.00342 deg F per year).

I have a couple of concerns about the results they have reported, which requires that I carry out some comparative plotting of my own to generate similar trends. Since I will be looking at the 48 contiguous states (though some will drop out as being too small) this will take a little while. So I ask you to bear with me if posting is a bit sparse while I run these numbers.

The Met Office, NASA GISS and is there a Thermometergate?

Because there are very obvious current differences between the weather in the United Kingdom and the forecasts of the Met Office, the BBC are reporting that the some at the Office are considering stopping the practice of making long-term forecasts. One of the problems has been identified:

The Met Office has now admitted to BBC News that its annual global mean forecast predicted temperatures higher than actual temperatures for nine years out of the last 10.

The bias is very small overall, about 0.05 degrees C, but the fact that it is a persistent warming bias should, perhaps, be of concern, relative to the accuracy of the models using it. Of course there are those, such as Professor Mobbs at Leeds, who would argue differently:

"All models have biases and these are very small. It may be, as the Met Office suggests, that the observations are wrong, not the model."

Must be all that illusion of snow that must be confusing folk in the UK after the model promised a greater chance of a warm winter.

Thermometergate, the concern over the quiet, and at the time un-noticed switch whereby the NASA Goddard Institute for Space Studies (GISS) changed number of stations used in assessing global temperature, is slowly getting more visibility. There was an hour long special (visible in five parts) on KUSI in which (during part 4) John Coleman introduced and talked with E. Michael Smith, who runs the site that has reported on this, and which has reviewed the GISS code and discovered that they now rely on a much smaller number of data acquisition sites than they have previously.

One final note: There has been A Great Dying lately for thermometers. Since about 1990, there has been a reduction in thermometer counts globally. In the USA, the number has dropped from 1850 at peak (in the year 1968) to 136 now (in the year 2009). As you might guess, this has presented some “issues” for our thermal quilt. But do not fear, GIStemp will fill in what it needs, guessing as needed, stretching and fabricating until it has a result.

Going through the list of stations that survive and looking for those in Missouri, there are apparently three:
(the first and second sets of numbers after the station name give the latitude and longitude of the stations)

42572434000 ST.LOUIS/LAMB 38.75 -90.37

42572445000 COLUMBIA/REGI 38.82 -92.22

42572440000 SPRINGFIELD/M 37.23 -93.38 .

Going to the actual GISS site to see how the values check out, GISS does recognize the drop in the number of station records that are used:

Number of stations used to generate the GISS temperature record

The full list of Missouri stations (there are 35 of them) are also given. The relevant plots given are:

Lambert St Louis temperatures

Columbia MO temperatures

Springfield Mo temperatures

The consistent part of which plots seem to be the lack of current evidence of warming and the higher temperatures during the dust bowl era. I also looked at the University site in about the middle of the state (at Rolla) and got this:

Temperatures at Missouri University of Science & Technology (until recently UMR)

One thing that did strike me about that record was that it did not look the same as the plot that I got from the USHCN data base, and which I had used once before. So I went and got that plot for Rolla:

Plot of the Mean Rolla Temperatures from the United States Historical Climate Network (USHCN) data base.

The impression of a rising trend evident in the GISS plot is not as evident in this one, and the highest temperature is barely, rather than clearly, 2006.

What is, however, odd, is that the USHCN data base for Missouri does not carry the station data for the three sites chosen by GISS.

I am going to continue to monitor the Rolla site, since (apropos Anthony Watts project on the validity of station data there has been an new technical addition near the measuring station and it will be interesting to see what, if any, effect the wind turbine has on the data generated from the weather station.

Current status of the station at Missouri University of Science and Technology (note the wind turbine on the right, and the weather station on the left).

The debate over the impact of the reduction in the number of sites that are being used to generate global temperature averages will no doubt continue, though it is interesting to note the specifics of what Dr. Hansen has said in his response to the KUSI report.

“NASA has not been involved in any manipulation of climate data used in the annual GISS global temperature analysis. The analysis utilizes three independent data sources provided by other agencies. Quality control checks are regularly performed on that data. The analysis methodology as well as updates to the analysis are publicly available on our website. The agency is confident of the quality of this data and stands by previous scientifically based conclusions regarding global temperatures.”

Now NASA (GISS) recognized that there has been a reduction in the number of monitoring stations (it is after all their graph that I used above) what it does not do is comment on the impact that changing the data base has on the overall result.

We will see how long it takes to getting around to addressing that issue.

Puzzling Graphical Comparisons raise questions of veracity

“Dire Predictions” by Michael Mann and Lee Kump, sets out to provide an illustrated guide to the findings of the IPCC. It is a relatively simple volume, full of the promised pictures, small graphs to illustrate points and condensed comments on the different aspects of the climate change debate. In short it is the sort of text that might be provided to a class in the United States to help them understand the prevailing arguments about climate change.

It has some nice initial illustrations of the way in which climate is generated that are easy to follow and which are therefore initially persuasive. But my eye was caught, from the beginning by the different graphs that are scattered throughout the chapters. I was a bit surprised by the first graph, since it didn’t quite look like the graph of the temperature plots given by NOAA, so after looking at the two separately:
From page 20 of the text

Global temperatures from 1880 to 2008 from NCDC

You can see quite a difference in the curves around 1940 – the actual peak back then has disappeared from Dr. Mann’s graph and if I superimpose them you can see how a fluctuating temperature record has been smoothed. (The heavy black line comes from the Mann and Kump curve)



The difference is more than subtle – the peak and stable or declining temperatures between 1940 and 1970 have been magically eliminated. But wait, those of you who have read the book respond – he puts a more detailed temperature plot on page 36.
Here it is:

Mann and Kump – Figure on page 36 – Trends in Global Average Surface Temperature.

But if you look at this figure – relative to the official plot you can see that while the official temperature is “debatably” flat in the official record, here the plot is steadily increasing from 1950.

Skipping forward through the book, let me pick out one more graph that caught my attention – the regional trends shown on Figure 71. Here is the Mann version:
Regional Continental Temperature Trends (after Mann and Kump) Blue are the temperatures taking into account natural trends only, pink includes both human and natural factors). Red is what they say happened.

But this is the official temperature record for North America from NASA.



Again I won’t bother superimposing the pictures, but you can see that the trends that are actually occurring don’t quite follow the curves in the book.
Having discovered which there really isn’t much point in continuing reading it, since it takes such liberties with easily verifiable figures, one is wondering what else has been “quietly adjusted” to make the facts more supportive of the argument.

Now it isn’t as though I completely agree with the official figures, given the corrections that have been imposed on the initial raw data, and that, as a result, trends appear that weren’t there before the “tweaking.’ But I do think that this book is taking the trend of “adjusting” the data just a bit too far in the process of making a point. There comes a point where this stops being Science and becomes Propaganda.