One wonders why the story of a new paper covered on WUWT: NCAR: Number of record highs beat record lows – if you believe the quality of data from the weather stations did not include the 1930′s and 1940′s and earlier, conspicuously missing from the NCAR graphic below:
From: “The relative increase of record high maximum temperatures compared to record low minimum temperatures in the U.S.”Authors: Gerald A. Meehl, Claudia Tebaldi, Guy Walton, David Easterling, and Larry McDaniel Publication: Geophysical Research Letters (in press)
The answer: those decades are inconvenient to the conclusion Meehl makes from a cherrypicked portion of the US data. There were many many temperature records during this period. For example, Richard Alan Keen writes in email:
My book, Skywatch West, covers the weather and climate of the 11 western states, plus Alaska, plus 6 western Canadian provincs and territories.
The chapter on temperature extremes includes a chart of the occurrences (by decade) of the all-time extreme temperatures for each of the 18 states, provinces, and territories (a total of 36 records in all).
Some fun statistics from this are:
- Of the all-time record maximum temperatures, 10 occurred before 1940 (the first six decades), and 8 after (the second six decades).
- For record minimum temperatures, the reverse is true: 8 records before 1940, 10 afterwards.
- Half of the records – 8 maximum and 10 minimum, a total of 18 – occurred during the middle three decades of the 1930′s, 40′s, and 50′s, and of these nearly a third of the total (10) were during the 1930′s alone.
- No records occurred in the 2000′s up to the publication date of the book (2004). Since then Arizona’s record maximum was tied, but not broken, in 2007.
Here is his graphic:
Granted this is not the entire USA dataset, only western states, and one could say that I’m engaging in the same sort of cherrypicking that Meehl et al engages in by illustrating it here. But there’s more.
The graph in #127 may be overly influenced by single extreme heat waves or cold snaps. So, here’s a look at broader populations.
The first is a look at the decades in which summer high temperature records were set. This covers the contiguous US for the three typically hottest months (June/July/August). A single nationwide event would affect the records for one month but not for all three, so this plot should be less-influenced by single extreme events.
The appearance is similar to the record high plot of #127.
Here is a similar plot except that it is for record lows in December, January and February:
There appears to be a modest downward trend in extreme cold events.
(Note: The final bar in each chart covers 2000-2003 (records posted as of May 2004) and is prorated so as to make an apples-to-apples visual display.)
Here’s the combination of the two:
Conclusion – the 1930s in the US were rough.
Note: The trendlines for all three graphs are essentially flat (no trend) if the prorated early 2000s are excluded from the trend calculations.