Skip to content

GHCN Temperature Adjustments In Iceland–A Closer Look At Stykkisholmur–Part I

October 28, 2012
tags: ,

By Paul Homewood




In a recent post, GHCN Adjustments In Iceland, I reported how GHCN had adjusted past temperatures at every single station Iceland. The adjustments all took the same pattern, reducing temperatures up to about 1965, usually by about 0.7C, thereby creating an artificial warming trend.

This is quite a long post, so  I am splitting it into two parts. The first section recaps what some of you may already know about. You may wish to skim over the first bit, but please stick with it as there is a lot of new and significant material in the second half.

The reason GHCN give for such adjustments is inhomogeneities in data, i.e. changes to temperature records caused by non climatic factors such as station moves, observation changes etc. GHCN identify these by looking for abrupt shifts in temperatures, and check against nearby stations through a Pairwise Algorithm to test whether the changes are non climatic. Full details can be seen in their technical reports.

According to the temperature records supplied to GHCN by the IMO, (Iceland Met Office), there was a sharp drop in temperature at all Iceland sites in 1965, and again the following year. It is clearly this event that has triggered the algorithm into making adjustments. Furthermore, as the fall in temperatures and subsequent GHCN adjustments appear at all six Icelandic stations, that are used by GHCN, the algorithm must have been comparing with stations hundreds of miles away.

So the question is – was the sudden drop in temperatures a genuine event, or was it caused by simultaneous observational changes? For instance, was it caused by the operators all being eaten by polar bears and being replaced by new guys who did not know what they were doing, as suggested by reader John G?

To get to the bottom of this, we are going to look at one particular station in more detail. Stykkisholmur is a small town of 1100 people in the west of Iceland. According to Hanna et al, “Stykkisholmur has the longest running and most uniform temperature record in Iceland.” It is also one of the three main stations regularly used by the IMO in climatic reports, so it is clear the records should be reliable.

[A note at this stage. There are three peer reviewed papers I will referring to throughout, “An Analysis of Icelandic Climate since the 19thC” by Hanna et al,”One Hundred Years In The Norwegian Sea” by Dickson and Osterhus, and “Sea Changes Ashore: The Ocean And Iceland’s Herring Capital” by Hamilton et al. The links are shown in the references in Part II.]


What GHCN say

The following two graphs show the raw and adjusted temperature records using GHCN data as presented by GISS.


Figure 1



Figure 2


Two things immediately stand out :-

1) The last two years shown, 2010 and 2011 appear to be much warmer than any previous period, according to the adjusted version, whereas in the raw version both 1941 and 1964 were almost as warm. The difference, as we already know, is because the earlier temperatures have been adjusted down, in this case by 0.6C. For instance, 1941 was originally recorded as 5.1C, but has been changed to 4.5C by GHCN. For some reason, the temperature for 2011 has also been adjusted up by 0.14C, so there is a net warming added of 0.74C.

Temperatures were adjusted downwards for each year up to and including 1963 as shown below. There is no adjusted data offered by GISS for 1964, but both raw and adjusted were shown as the same for 1965, and remain the same up to 2010, except for 1972-80, when they were adjusted up slightly.




2) There is a gap in the GISS records between 1981 and 2008 in the raw data, and 2001 and 2009 in the adjusted. This is a total mystery, since the IMO have full records throughout, as can be seen in Figure 4. It also shows the same trend as Figure 1.



Figure 4

The abrupt shift

Now let’s take a closer look at the annual figures.



Figure 5

In 1964, the annual mean temperature reached 5.05C, but dropped to 3.88C the following year, and even further to 3.1C in 1966. (N.B. GISS annual number are always MetAnn, i.e Dec-Nov). There is no doubt this was an abrupt shift. However, if this attracted the attention of our friend, the algorithm, I would question why it was not triggered by the equally abrupt rise in temperature between 1963 and 1964, from 3.58C to 5.05C. (N.B. the gaps in the GHCN V2 data from 1981-2000, referred to above have been replaced with the IMO data, for ease of presentation).

It is also notable that, by the early 1970’s, and again by the late 1980’s, temperatures had recovered to the levels seen in the 1940’s and 50’s. There is certainly no sign of any permanent shift downwards.

We can, though, look more closely at the monthly data, in order to see if there really was an abrupt shift.



Figure 5


In comparison to 1964, temperatures in the following two years were certainly much lower for January to May. However, check the months that follow. From June to October, the temperatures in the “cold” years were as warm as, or warmer than, 1964, before going downhill again.

And not only in comparison with 1964. These summer months were consistent with average temperatures from 1954-63. It is hard to see any evidence here at all for a permanent shift downwards in temperature, that could have been caused by non climatic factors.

Already any justification for these adjustments seems to be falling apart at the seams . But we don’t have  to simply rely on statistics. We can look at what the experts say about this period, the scientists who have spent years studying the climatology and ecology of this particular part of the Arctic.

We will see what they have to say in Part II.

One Comment
  1. October 30, 2012 3:05 am

    Data crime. >:-[

Comments are closed.

%d bloggers like this: