June 2, 2014, 9:29 am
This chart illustrates a data analysis mistake that is absolutely endemic to many of the most famous climate charts. Marc Morano screencapped this from a new EPA web site (update: Actually originally from Pat Michaels at Cato)
The figure below is a portion of a screen capture from the “Heat-Related Deaths” section of the EPA’s new “Climate Change Indicators” website. It is labeled “Deaths Classified as ‘Heat-Related’ in the United States, 1979–2010.”
The key is in the footnote, which says
Between 1998 and 1999, the World Health Organization revised the international codes used to classify causes of death. As a result, data from earlier than 1999 cannot easily be compared with data from 1999 and later.
So, in other words, this chart is totally bogus. There is an essentially flat trend up to the 1998 switch in data definition and an essentially flat trend after 1998. There is a step-change upwards in 1998 due to the data redefinition. This makes this chart useless unless your purpose is to fool generally ignorant people that there is an upwards trend, and then it is very useful. It is not, however, good science.
Other examples of this step change in a metric occurring at a data redefinition or change in measurement technique can be found in
- The hockey stick (and here)
- Ocean heat content (sorry, can't find the link but the shift from using thermometers in pails dipped from ships to the ARGO floats caused a one time step change in ocean heat content measurements)
- Tornadoes
- Hurricanes
March 7, 2014, 9:24 am
I try to make it a habit to criticize bad analyses from "my side" of certain debates. I find this to be a good habit that keeps one from falling for poorly constructed but ideologically tempting arguments.
Here is my example this week, from climate skeptic Steven Goddard. I generally enjoy his work, and have quoted him before, but this is a bad chart (this is global temperatures as measured by satellite and aggregated by RSS).
He is trying to show that the last 17+ years has no temperature trend. Fine. But by trying to put a trend line on the earlier period, it results in a mess that understates warming in earlier years. He ends up with 17 years with a zero trend and 20 years with a 0.05 per decade trend. Add these up and one would expect 0.1C total warming. But in fact over this entire period there was, by this data set, 0.3C-0.4C of warming. He left most of the warming out in the the step between the two lines.
Now there are times this might be appropriate. For example, in the measurement of ocean heat content, there is a step change that occurs right at the point where the measurement approach changed from ship records to the ARGO floats. One might argue that it is wrong to make a trend through the transition point because the step change was an artifact of the measurement change. But in this case there was no such measurement change. And while there was a crazy El Nino year in 1998, I have heard no argument from any quarter as to why there might have been some fundamental change in the climate system around 1997.
So I call foul. Take the trend line off the blue portion and the graph is much better.
