Hottest July In Virginia? Or Maybe Not.
By Paul Homewood
NOAA recently announced that July 2012 was the hottest month on record for the CONUS. As the above map shows, though, only Virginia actually broke the state record for July, logging up 79.0F, beating the previous record set in July 1934 of 78.6F. Or, at least, that’s what NOAA tell us, but what do the thermometers say?
“The USHCN is actually a designated subset of the NOAA Cooperative Observer Program (COOP) Network the USHCN sites having been selected according to their spatial coverage, record length, data completeness, and historical stability.”
In other words they are picked as being robust and reliable. Out of the nineteen, only thirteen have records back to 1934 and these are listed below with comparisons of this year’s temperatures with those of July 1934. The temperatures shown are the actual original readings, and not adjusted, homogenised or otherwise tampered with! (The original handwritten/typed monthly records for these can be accessed on the USHCN site here).
|Station||Division||July 1934||July 2012||Difference||Rural Y?|
|Degree F||Degree F|
(Although operational at the time, the original records for 1934 are not available for Bremo Bluff and Piedmont, so figures for both years have instead been sourced from GISS).
As can be seen, only three stations out of the thirteen show this year as being warmer. You will also note that I have indicated the climatological division for each station. There are six such divisions for Virginia, as listed below, so all divisions are represented. NOAA currently work out state temperatures by averaging temperatures across all stations within a division, and then weighting each division’s temperature according to its area. We can therefore repeat this logic with the “differences” calculated above.
|Division||Area Square Miles||Difference||Weighted
|2- Eastern Piedmont||7751||0.4||0.1|
|3- Western Piedmont||7392||0||0|
|5- Central Mountain||5200||-0.7||-0.1|
|6- Southern Mountain||7681||-2.7||-0.5|
So only two out of the six divisions have recorded an increase, and only one of these is more than the 0.4F announced by NOAA for the whole of Virginia.
There are three other points worth making here :-
1) NOAA do not take TOBS adjustments into account when calculating state temperatures (although they do when working out national figures). This is due to change next year when the new GrDD system (Gridded) is introduced. Therefore TOBS cannot be the reason for the difference between the USHCN records and NOAA’s. (And of course, there are no allowances for UHI either).
2) I have indicated which sites are classified as “rural” by GISS (i.e. <10000 population). Out of the three sites that report higher temperatures in 2012, only one is rural.
3) The only significant temperature increase is in the Tidewater Division, and here of course, we only have one sample, the town of Hopewell. So just how robust is the Hopewell temperature record?
There are two factors that suggest Hopewell’s record may not be totally reliable.
a) Hopewell, although a city in its own rights, is part of the Greater Richmond Metro Area. There must, therefore, be a question mark over any possible UHI effects.
b) The USHCN station is situated within a water treatment plant, which may not be ideal.
So how can we explain the difference between NOAA’s figures and the evidence from the USHCN stations?
As Anthony Watts has often complained, NOAA seem to be very reluctant to divulge how they operate. We do, however, know how the current temperatures are worked out.There are 124 stations in Virginia, split into the six divisions, with a simple average taken of the set of stations in each division. The divisions, as previously noted, are then weighted to give a statewide figure.
The real mystery, though, is how they arrive at the 1934 number. There are State Climatological Reports available from then, but no direct comparison can be made with current ones because there were less stations then and they were allocated to only three divisions instead of six. Also the majority of stations reporting in 1934 no longer report now. (For interest, the State temperature given at the time for July 1934 was 78.8F, i.e. 0.2F higher than now shown).
When I asked Deke Arndt of NOAA how they calculated the temperatures for 1934, he seemed as confused as me!
“Climate divisions weren’t in use then, so the circa 1934 is likely a straight up average of available stations at "press time"
The more modern CAG value uses the CDs to build the state value (this is a factor if stations are distributed unevenly within the state), it incorporates the late-arriving (in 1934) stations, several QA tests, etc.
Those are probably the factors in play here.”
So it would appear the temperatures that NOAA now declare for 1934 are no more than a “theoretical reconstruction”. But without the transparency to show how this has been done, how can we have any confidence that it has been done properly. There is a potential discrepancy of 0.6F in Virginia; if this is extended across the whole country, it would destroy the claim that this July was the hottest month on record as July 1936 was only 0.13F cooler on NOAA’s record. Indeed this would only the 5th warmest July in the CONUS.
So here’s my challenge for NOAA. Prove me wrong. Prove that the USHCN station analysis is not giving the true picture. Show how you come to the conclusions you do and prove that they are correct.