WunderBlog Archive » Dr. Ricky Rood's Climate Change Blog

Category 6 has moved! See the latest from Dr. Jeff Masters and Bob Henson here.

Temperature Data and Urban Heat Islands

By: Dr. Ricky Rood, 7:15 PM GMT on August 19, 2007

Temperature Data and Urban Heat Islands

I am going to start this entry with the end of the last entry. Two things I want to talk about are the temperature data set and the urban heat island.

I introduced the graph of the Central England observations and said that it was for the purists. These are data made with thermometers, and the thermometers are "rural." For sure there have been changes over the years, but in some sense the spirit of this data set is one of a consistent simplicity for a very long time. Here is that plot again, and if you follow the links you will find references and discussions. Temperature measurements from Central England.


Figure 1, Central England Temperatures, 1772 --> present.

I asked what people saw in this figure. Here is how I would analyze this figure. First, this is a figure that is a difference from some average; therefore, what is the averaging time, which is according to the reference, 1961-1990. Given this, you can look at that time in the graph, and over that time the positives and minuses should balance each other. If not, well there is a problem. (If there is a strong trend during the averaging time, that would show up in the differences, and suggest a choice of a different averaging time.)

Next, I would notice that there have been times of sustained difference from the average. In the past, before 1900, there were a number of times where it was cooler than the average, and very few times where warmer than average was sustained. There were a few very cold years, and a number of years that were nearly as warm as the most recent years. The recent warm time is the longest sustained warm time, but there are a number of cool times of comparable length. For this time record, the recent years are definitely the warmest. This does not establish cause and effect.

If you were to look at the time series closest to the averaging span, then this is when there is the most oscillation above and below the average. Going into the past, again before 1900, the temperature lies mostly below zero. If you calculated a trend between most any average before 1900 and today, it would show warming.

The questions that arise about this graph: What is the impact of changing the averaging length? What is sensitivity to the choice of time span for the average? Of course there are questions of data quality and how representative are they of the temperature in the region that they cover. As a reminder, this time series is made up of only a small number of thermometers in a part of central England. This is not meant to represent the globe.

In terms of understanding cause and effect, this is most accurately thought of in terms of what meteorologists call the thermodynamic equation, which is an equation that describes the behavior of temperature in time. Most simply, the temperature can change based on heating, cooling, and variability associated with changing weather patterns, like the North Atlantic Oscillation. This being a plot of regional temperatures it is likely sensitive to changes in weather patterns. In terms of heating and cooling, one would look at, for instance, solar variability, volcanic activities, changes in composition, and changes in local environment of the measurements, and of course the accuracy and precision of the measurement instrumentation.

It has been long recognized that changes in the local environment impact the measurement of temperature. This makes the job of defining trends and determining cause and effect difficult. This is a challenge for all observing systems. All. Clearly, any measurements of the last 100 years would have to consider the role of urbanization in the temperature record. There is a correlation, but again, correlation does not determine cause and effect.

Some have written in response to this blog as if urban effects have not been considered. Or if it has been considered, then it has been considered in either an inaccurate or a disingenuous way. The impact of urban heat islands is a problem that has definitely been considered. Typing "urban heat island" into the Web of Science, the old Science Citation Index, finds 63 pages of peer-reviewed references. (Most big libraries have a subscription to Web of Science.) Here is a link to a paper from the American Meteorological Society's Journal of Climate. (Parker, 2006, A Demonstration that Large-scale warming is not urban.) Here so you can read what is done in the GISS analysis of temperature is Hansen et al., 1999, GISS Analysis of Surface Temperature Change. You will see that in these papers the authors take some trouble to account for difficulties in the data sets. Further, they provide a list of references that show how others have tried to address the problem.

It is fair and legitimate to read these papers and to challenge the methods and the conclusions. (In fact, this is taught formally in graduate school. Students are given such papers and asked to analyze and evaluate methods and the conclusions.) I don't think that there is any justification in stating the scientists have been negligent in considering, for instance, the impact of urban heat islands. Nor is there justification in saying that there is some attempt to hide methodology. It is not in the best interest of science or scientists. (Seriously, if you think it is in the best interest, tell me why.)

Now back to the temperature data. I expect that there will be more errors found in all of the data records. I can think of very few things that if one continues to revisit over and over whose perfection stands up to such scrutiny. Each of the corrections will be incorporated and the quality of the data record will be improved. The question arises - are there errors in the observations or the analysis that will challenge the basic conclusion that the Earth is currently in a sustained period of warm climate? Or are we only refining the accuracy of that conclusion? I personally do not think that we will find any systematic error in the observations or the analysis that will change the conclusion. And if there is something from a single observation that changes it all --- well, I think that would be magic, and there is no magic.

The views of the author are his/her own and do not necessarily represent the position of The Weather Company or its parent, IBM.