Massive Tampering With 1934 US Temperatures
By Paul Homewood
The heatwaves and droughts that affected large swathes of America in the 1930’s are a historical fact. The hottest year of the decade was 1934, but just how hot was it? NCDC insist that 2012 was 1.2F hotter, with 1934 beaten into 4th place by 2006 and 1998, but are they right?
NCDC say that the annual mean temperature, for the CONUS, was 54.14F in 1934, compared to 55.34F last year. But what numbers were being declared at the time?
Fortunately, we can reconstruct the original figures by using the individual State Climatological Reports, that are still archived by NOAA here, (see example below).
Appendix A lists the annual mean temperatures for each state for 1934. These temperatures are then weighted according to geographical area, which are available from NOAA here.
The sum of these weighted temperatures gives the national figure of 55.37F. In other words, 1.23F higher than is now declared by NOAA for 1934, and actually slightly higher than 2012.
So why are the figures so different? It is often claimed that the mix of stations has changed over the years. So, for instance, there might be more mountain stations than there were in the past, and consequently current temperatures would be biased down. In fact, this is certainly true in Arizona, the usual example given, and possibly one or two other states. But, as Appendix B shows, nearly every State has been adjusted in the same way.
It would be ridiculous to suggest that, in every State, there are proportionally more stations in colder places. Even excluding Arizona and other outliers from the overall numbers would leave a national adjustment of well over 1.0F.
Furthermore, if proper consideration is made for station mix, then there should also be allowance made for the UHI effects of towns and airfields.
It is also worth pointing out that, back in 1934, the State Climatological Reports were already factoring in any potential effect from changing station mix. They did this by the use of climate divisions, areas that tended to be climatologically distinct, but of a similar size. Typically each State would have between six and eight divisions.
Just as now, the stations within a division were averaged together, giving a divisional average. The divisional averages were then averaged to give the State figure. The addition/deduction of stations within a particular area, therefore, would in most cases not have a significant effect on the State temperature.
Previous analysis in Alabama and Virginia also strongly suggests that changing mix could not explain NCDC adjustments. I have also carried out some more detailed analysis of New Jersey, Delaware and Iowa, which, in those states at least, rules out a changing mix as an explanation.
A close look at the Appendices shows that the adjustment from the original figures to today’s version is split into two stages.
1) The original figure is reduced from 55.37F to 54.51F, a drop of 0.86F, when each individual original State record is compared with NCDC’s latest version, and then a weighted average produced.
2) However, this figure of 54.51F then morphs into a national number of 54.14F, a further fall of 0.37F. In other words, the individual State temperatures don’t support the national figure.
Perhaps it really was as hot back in the 1930’s as they believed at the time.
APPENDIX A – ORIGINAL TEMPERATURE RECORDS
|MEAN TEMP||AREA||WEIGHTED TEMP|
APPENDIX B – CURRENT NCDC STATE RECORDS