Guest Post By Bruce Hall of Hall of Record
Dr. Pielke has kindly offered me the opportunity to address a basic issue regarding the difficulty in holding meaningful discussions with regard to climate change based on an email exchange I had with a person who agreed with my conclusions, but was uncomfortable with my methodology.
The following statement was part of a recent email I received:
I have just ‘stumbled’ upon your fascinating page and have been reading about the U.S. extreme temperatures. I do not challenge the numbers but I do think their description can be a bit misleading. You list the temperatures as monthly when, in fact, they are single day extremes for a given month. The true monthly temperature extremes (record highs or lows) are the average of the readings taken during a given month. I have downloaded the NCDC-NOAA state-by-state database and the monthly highs and lows are almost always different. Neither figures support a warming in the contiguous US.
Three years ago, I published an analysis of U.S. all-time monthly temperature extremes by state from NOAA data and then updated it again in 2009. The worksheets are available as Excel files here:
Extreme Temperatures By State – Database and Analysis [Excel 2007 File]
This material provides a somewhat different perspective of the U.S. climate over the past 14 decades than popularly shown.
Most of us are comfortable with the idea that average temperatures accurately reflect weather. We like to know that the average normal for July is 74° and for January it is 27°. It gives us a sense of understand. The idea of using extremes as a measure makes us uncomfortable. What do we really learn from a statement that the July high temperature record was 113° and the record low temperature for January was minus 37°? In terms of simple expectations, averages are more comfortable than extremes. We can’t plan for extremes. We don’t purchase a wardrobe for extremes.
Nevertheless, when it comes to climate, I propose that averages… as we derive them… are fraught with problems, not the least of which is the practice of changing historical data to “fix” changing conditions. An average fails to give insight into variation; what is the temperature range upon which the average is derived? For example, average temperatures could increase in these scenarios:
In the case of high and low average temperatures increasing, one would expect to see more frequent high temperature records… a basic assumption of global warming. In the case of only low average temperatures increasing, one might question why. Indeed, Anthony Watts has done an extensive study of weather stations across the United States and concluded that a large part of the problem stems from poor station siting and encroachment of urban areas around weather stations… “lows tonight in the upper 50s… cooler in the outlying areas.”
Certainly some cyclical warming occurred after the 1970s, but the record of new high temperature extremes shows that … whatever the derived averages … there was no significant climate change versus past cyclical warm periods.
My response to the email was:
While I understand how one can find fault with calling the records “monthly” as opposed to daily, the data are for the highest and lowest all-time recorded temperatures for each month for each state since 1880. They are records of extremes as opposed to calculated or derived averages.
While this may not satisfy some, I believe it has specific advantages over averages.
1. The data are not adjusted. They are recognized as valid by NOAA and have not been “corrected” by interpolation with data from other weather station data.
2. They represent the climate “boundaries” for statewide geographies. A record either stands on it’s own or is replaced by a subsequent reading that ties or exceeds it.
3. It allows testing of the tenet that global warming necessarily results it an increased frequency of temperature extremes.
While one can argue that US records are a small fraction of earth’s geography coverage, they represent the most consistent sampling for the past 130 years.
My conclusions from this exercise were that:
1. The last two decades were warm, but not abnormally so.
2. The 1930s were significantly warmer over a wider geography.
3. The last decade was unusually berift of temperature extremes, with low temperature extremes occurring in the NE US.
From the January, 2009 update:
For further reading concerning the problem of framing discussions about climate change, see this:
Climate Science: Roger Pielke Sr. 