NCDC Temperature Adjustments In Alabama
By Paul Homewood
Climate At A Glance
In a recent article, I reported how the introduction of the new “Gridded Divisional Dataset”, (GrDD), by NCDC had, in Alabama at least, the effect creating a warming trend by both cooling older temperatures and also warming more recent ones.
For instance, according to the NCDC chart above, which is based on the new system, mean temperatures for Alabama were 57.0F in November 1934 and compare to 55.5F in November 2011, making the latter 1.5F cooler. But the original state reports published by NCDC show figures of 57.6F and 55.3F respectively, leaving 2011 as 2.3F cooler. So somewhere the new GrDD dataset has added a warming trend since 1934 of 0.8F. (The full report is here).
One possible reason for this discrepancy is that the temperature measurement in 1934 was based on a different mix of stations than the current one. It is easy to see how, for instance, older datasets could have concentrated on warmer valley sites, while later sets may include a higher proportion of cooler higher altitude sites. However a properly weighted analysis of Alabama stations shows that this does not seem to be the case.
There is, however, another explanation. NCDC explain the differences between the new GrDD and old TCDD datasets. As well as addressing the mix and weighting of stations, they say
Finally, none of the TCDD’s station-based temperature records contain adjustments for historical changes in observation time, station location, or temperature instrumentation, inhomogeneities which further bias temporal trends.
In other words, temperature adjustments, and in particular time of observation adjustments or TOBS. (It of course makes sense to adjust for station location changes, but these would tend to be self cancelling). So what difference have TOBS adjustments made to Alabama temperature trends?
There are fifteen USHCN stations listed for Alabama, of which two, Muscle Shoals and Troy have no data for last year. I have listed the rest below, showing the originally recorded raw mean temperatures and temperatures after TOBS adjustments have been made, for Novembers, 1934 and 2011.
|Nov 1934||Nov 1934||Nov 2011||Nov 2011||TOBS Adj|
|1934 v 2011||2.2||1.5||0.7|
USHCN temperatures are available here.
The change in raw temperatures between the two years, 2.2F, is virtually identical to the figure of 2.3F we have already identified above on the original state reports. Equally the change in the adjusted temperatures, 1.5F is exactly the same as the new NCDC dataset.
So we now have three datasets all pointing to a difference of about 2.3F between 1934 and 2011 – the original state reports, my weighted reanalysis, and now the USHCN stations.
Coincidence? I think not. It is clear that pretty much all of the difference between the old and new datasets is due to temperature adjustments.
And it is not just Alabama. For instance the Southeast Region, covering Virginia down to Florida and across to Alabama, shows a similar adjustment, cooling the past and warming the present by the best part of a whole degree. So does the Northeast Region.
I should point out US temperature records nationally have been based on adjusted USHCN temperatures for a few years now. What has now changed is that State and Regional datasets have now also been put onto the same basis. But this is the first time we have really been able to see clearly the effect these adjustments.
How significant is 0.7F? Well, nationally, US temperatures according to NCDC have been increasing at a rate of 0.10F/decade sine 1930, or about 0.80F during the last 80 years or so. Without these temperature adjustments, this increase effectively disappears.