Skip to content

Changing Station Mix?

February 19, 2013

By Paul Homewood


In my previous post, I discussed the large adjustments applied to the original US temperature record of 1934. One of the justifications often thrown up, is that there are a greater proportion of stations in “cold locations” than before. For example, more stations at higher altitudes.

This may be true in one or two states such as Arizona. Equally it could be argued that there may be more stations today in extremely hot places!

But, is there any evidence of this in the vast majority of states, or are there other explanations for the adjustments? I have looked at three states, and the results suggest station mix is not a significant factor.



New Jersey

The original state records show the annual mean temperature in 1934 was 51.6F. Current NCDC figures give 51.2F, a drop of 0.4F

In 1934, the State was split into four divisions:-



Sandstone Plain

Southern Interior

Sea Coast


The Highlands is located in the NW part of the state, the rest of which is essentially flat and low lying.

In 1934, there were 3 stations over 500’. (Altitudes are given in the State Climatological Reports):-

Charlotteburg – 719’

Dover – 600’

Long Valley – 520’


In 2012, there are just two stations:-

Charlotteburg – 719’

Sussex – 649’


There is therefore no evidence of any shift to higher altitude sites.



As the second smallest state, there is very little room for variability from one part of state to another, yet NCDC have adjusted the 1934 temperature down from 54.7F to 54.1F.

In 1934, there were 6 stations declaring, only one of which was over 100’. This was Wilmington at 260’.

Fast forward to today, and there are 5 stations. Again, there is only one station over 100’, the same station at Wilmington.



Iowa makes an interesting example, as it still retains exactly the same nine climate divisions, as it had in 1934. Therefore, changes in station distribution from one part of the State to another cannot create any bias, as the divisional system acts as a firewall. Yet the original temperature for 1934, of 51.5F, has now been adjusted down to 50.3F, a reduction of 1.2F.

Any change in station mix can, therefore, only have had an effect within divisions. The nine divisions range from 4993 to 7368 square miles, or, on average, about 6200 square miles. Say about 80 x 80 miles. These are much too small to have any significant variation within, unless there are variations in altitude.

To test this, I have looked at Division 1, which is the Northwest. The State Climatological Report shows a mean temperature of 23.2F, as below.




Yet the NOAA’s ftp site now shows the temperature as 21.9F, a drop of 1.3F, as per the screenshot below. The coding works, (see here),  as follows:-

13 = Iowa

01 = Division

02 = Denotes temperature

1934 = Year

The next 12 columns are monthly values.


ScreenHunter_54 Feb. 18 17.28


In 1934, there were 15 stations in the Northwest Division, with an average altitude of 1365’. There are currently 17 stations in this division (see below), with an average altitude of 1354’. (Altitudes are listed on Page 26 of the Report below).




So the large adjustments, made by NCDC in Iowa, simply cannot be justified by changes in the distribution of stations, either from warmer to cooler parts of the State, or from lower altitudes to higher ones.



In none of the three states are there any changes in station distribution, which could justify the adjustments NCDC have made. I have no doubt it will be claimed that these adjustments are justified by “peer reviewed literature”, but where do NCDC actually explain they have reduced 1934 temperatures by 1.3F?

NOAA/NCDC have become too big, too bureaucratic, too remote, too opaque, and too reliant on “algorithms” rather than practical, local knowledge.

Full responsibility for climate recording, including compilation of historical data, should be handed back to the individual states. At state level,they would possess the local knowledge necessary, to ensure that measuring equipment is properly sited and operated. They should also be able to ensure that any adjustments needed for local factors, such as station moves and UHI, were made, transparently, on the basis of that local knowledge.





All State Climatological Reports, used in the above analyses, are available here.

  1. Coldish permalink
    February 19, 2013 3:55 pm

    This is good work, Paul. Any chance of extending your treatment to some states with larger altitude range?

  2. Ian permalink
    February 19, 2013 6:02 pm


    Does NCDC justify or explain each adjustment that they have made? If not, have you inquired of them on specific examples (such as the ones you give here). I’d be curious to see if they have a response (or would deign to respond).

    Good work.


    • February 19, 2013 7:21 pm

      I have asked them to explain where they got a particular month’s numbers from (for Virginia). Deke Arndt effectively admitted he did not know!!

      On a global basis, I have specifically requested GHCN to tell me how they arrive at some of the adjustments made in Iceland. Despite being told this would be simple, and despite me chasing several times, I still have had no reply 4 months later.

      Good idea, though. I’ll try Deke again.

      • Ian permalink
        February 20, 2013 6:32 pm

        It will be interesting to see if there is a coherent response. Keep up the good work!


  3. Ian permalink
    February 20, 2013 8:10 pm


    A quick check on Kansas shows that the difference between the historical state report and the current figures used by NOAA is a whopping 2.85 degrees F over the year (so, they’ve lowered the average temperature for 1934 by 2.85 degrees F; on a monthly basis, the changes range from 0.3 degrees to 3.9 degrees F – in every case, the 1934 temperature has been lowered from the original report).

    1934 report:

    1934 State report on temperatures:

    Jan Feb Mar Apr May June July Aug Sept
    36.3 36 44.1 57.1 69.6 80.5 87.2 83.6 65.2

    Oct Nov Dec
    62.3 47.3 33.1

    Current NOAA temps for 1934:
    1401021934 34.40 33.60 41.70 53.40 69.30 77.10 84.80 79.80 62.40 59.00 43.40 29.20

    The challenge in Kansas is that there have been significant changes in the number of recording stations, and the divisions. Back in 1934, there were 3 divisions (Eastern, Middle and Western) with an aggregate of about 91 reporting stations (E – 31; M – 32; W – 28). There are now 9 divisions and around 150 recording stations.

    2011 Kansas report:

    That still doesn’t really explain why EVERY temperature should be lowered in 1934. Does NOAA have a coherent explanation for how and why the temperatures record from the past has been changed?

Comments are closed.

%d bloggers like this: