Guest post by David R.B. Stockwell PhD
I read with interest GHCN’s Dodgy Adjustments In Iceland by Paul Homewood on distortions of the mean temperature plots for Stykkisholmur, a small town in the west of Iceland by GHCN homogenization adjustments.
The validity of the homogenization process is also being challenged in a talk I am giving shortly in Sydney, at the annual conference of the Australian Environment Foundation on the 30th of October 2012, based on a manuscript uploaded to the viXra archive, called “Is Temperature or the Temperature Record Rising?”
The proposition is that commonly used homogenization techniques are circular — a logical fallacy in which “the reasoner begins with what he or she is trying to end up with.” Results derived from a circularity are essentially just restatements of the assumptions. Because the assumption is not tested, the conclusion (in this case the global temperature record) is not supported.
I present a number of arguments to support this view.
First, a little proof. If S is the target temperature series, and R is the regional climatology, then most algorithms that detect abrupt shifts in the mean level of temperature readings, also known as inhomogeneities, come down to testing for changes in the difference between R and S, i.e. D=S-R. The homogenization of S, or H(S), is the adjustment of S by the magnitude of the change in the difference series D.
When this homogenization process is written out as an equation, it is clear that homogenization of S is simply the replacement of S with the regional climatology R.
H(S) = S-D = S-(S-R) = R
While homogenization algorithms do not apply D to S exactly, they do apply the shifts in baseline to S, and so coerce the trend in S to the trend in the regional climatology.
The coercion to the regional trend is strongest in series that differ most from the regional trend, and happens irrespective of any contrary evidence. That is why “the reasoner ends up with what they began with”.
Second, I show bad adjustments like Stykkisholmur, from the Riverina region of Australia. This area has good, long temperature records, and has also been heavily irrigated, and so might be expected to show less warming than other areas. With a nonhomogenized method called AWAP, a surface fit of temperature trend last century shows cooling in the Riverina (circle on map 1. below). A surface fit with the recently-developed, homogenized, ACORN temperature network (2.) shows warming in the same region!
Below are the raw minimum temperature records for four towns in the Riverina (in blue). The temperatures are largely constant or falling over the last century, as are their neighbors (in gray). The red line tracks the adjustments in the homogenized dataset, some over a degree, that have coerced the cooling trend in these towns to warming.
It is not doubted that raw data contains errors. But independent estimates of the false alarm rate (FARs) using simulated data show regional homogenization methods can exceed 50%, an unacceptable high rate that far exceeds the generally accepted 5% or 1% errors rates typically accepted in scientific methods. Homogenization techniques are adding more errors than they remove.
The problem of latent circularity is a theme I developed on the hockey-stick, in Reconstruction of past climate using series with red noise. The flaw common to the hockey-stick and homogenization is “data peeking” which produces high rates of false positives, thus generating the desired result with implausibly high levels of significance.
Data peeking allows one to delete the data you need to achieve significance, use random noise proxies to produce a hockey-stick shape, or in the case of homogenization, adjust a deviant target series into the overall trend.
To avoid the pitfall of circularity, I would think the determination of adjustments would need to be completely independent of the larger trends, which would rule out most commonly used homogenization methods. The adjustments would also need to be far fewer, and individually significant, as errors no larger than noise cannot be detected reliably.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
As an example, the Geneva station (but all stations follow the same pattern):
http://data.imagup.com/12/1165303601.png
The adjustment of 1962 corresponds to the move of the station from city center to airport. Where are UHI continuous adjustments ? Subsequent cooling to the remoteness from the city is however well done!
@phi. Local gradual trends are more difficult to adjust accurately than discontinuities. I only wanted to argue that you can and should do it. Hansen did not argue that homogenization is expected to have a neutral effect on trends. We know that the temperatures measured in the past were too high due to problems with protecting the thermometers for solar and heat radiation. And in a specific network you can also expect biases due to typical changes, such as the transition from Liquid in Glass thermometers to automatic weather stations in the US.
Do you have reasons to expect problems with urbanization in Geneva?
Victor Venema:
I am now able to give a proper and considered reply to your post to me at October 18, 2012 at 3:59 pm. As I said in the early hours of this morning, I am genuinely grateful for your reply.
Firstly, I apologise for mistakenly thinking you were involved in or with the compilers of the GHCN global temperature time series. Much of my questioning of you was based on that misunderstanding and so was my frustration at what seemed to be your evasions.
Clearly, I need to explain how I gained the misunderstanding which has required this apology. I give you that explanation briefly here.
Following your repeated assertions that “no more than a few per cent of the data are affected by urbanization”, at October 15, 2012 at 10:13 am you said
I interpreted this – I now know wrongly interpreted this – to mean you and your “colleagues” were working on compiling the GHCN data set but you have not specifically addressed the UHI issue.
Then, at October 15, 2012 at 2:44 pm you wrote to ‘laterite’ (aka David Stockwell) saying
That lifted the hairs on my neck because science is not about “sides”; it is about assessing and challenging data and hypotheses. Importantly, it seemed to confirm that you were ‘at one’ with compilers of the GHCN data set.
Subsequently, you repeatedly referred me to your article – which you linked – that you said may answer my questions. But that link told me nothing I did not know and did not mention the fundamental issues of data reliability, accuracy and precision which I had repeatedly queried. However, it did seem to be a presentation of ‘insider’ knowledge of the GHCN data compilation.
Hence, I progressively obtained impressions which I put together so (2+2)-=5.
I wrongly thought you were a compiler of the GHCN data set who was defending the GHCN method while avoiding evaluation of that method. That thought was an error. I completely apologise for my misunderstanding and any difficulties which my misunderstanding may have created.
Having got that out of the way, I can respond to your substantive point which is
I also “couldn’t care less whether the temperature goes up, down or stays the same” but I would like the omniscience to know which it is going to do. 😉
It pleases me that you were asked “to lead the validation study because [you] had no stake in the topic”. That is how it should be. However, you and I would have differed in our approaches to that. As I understand your article you were interested in data “quality” whereas I would have investigated effects of homogenisation on data reliability, accuracy and precision and to what degree those effects could be determined.
As an addendum I point out that interacting with people outside the immediate research domain provided me with many benefits when I was directly involved in scientific research. I was at the UK’s Coal Research Establishment (CRE) and then would find excuses to discuss the work with non-scientists such as mechanics in the workshops, gardeners and lavatory cleaners. This was rewarding for several reasons.
Firstly, if I could not explain the work to one of them then I knew I did not have sufficient clarity of understanding of the work myself.
Secondly, it gave them an ‘involvement’ in the research which was our common purpose so they wanted to perform well with several resulting benefits including improved conduct of the work (e.g. best quality constructed research equipment).
Thirdly, they were not constrained by their training and background so would ask ‘naïve’ questions which those of us ‘in the box’ would never ask. This could have unforeseeable benefits. For example, I had a theoretical explanation of – so possible solution to – the longstanding problem of heat-exchanger tube wear in FBC fluidised combustor beds. A discussion with a mechanic in CRE’s workshops informed me of a new ability to make long, narrow, longitudinal holes in tube walls. I recognised that this could enable thermocouples to be positioned inside a tube wall along the length of a tube. And a conical insert inside the tube would provide a range of outside wall temperatures along the tube. With that knowledge the problem was solved. And the idea of linking this new ability (to erode long, thin holes) with my problem may never have occurred to me without the discussion with the mechanic.
So, I think you may find the very wide range of backgrounds, knowledge and experience on WUWT may prove useful to you if you ‘tap in’ to it.
Richard
Victor Venema:
Camuffo wrote a very detailed exposition on 7 historic european data sets via the ‘Improv’ project. He believed there to be a consistently warm bias.
I think the problem is much more complicated than that and wrote about the problems with historic temperatures here
http://wattsupwiththat.com/2011/05/23/little-ice-age-thermometers-%e2%80%93-history-and-reliability-2/
.In short the methodology is highly suspect until the advent of digital stations in the 1980’s which then had their own problem with siting as Anthony WAtts has chronicled.
In general-when you also take into account information such as crop prices and observational evidence of the time, there are many times when camuffos instrumental detective work must be questioned and the warm bias doesn’t always exist. This is complicated by uhi and a physical move of the station to a different micro climate which still bears the stations original name .
Personally I think UHI is very real, but once urbanisation reaches a certain level the effect spreads over a wider geographical area, rather than intensifies.
There are some 30000 stations worldwide of which around one third are cooling accrdig to BEST. A great percentage of the remainder are in an urbanisation where the warming could be caused by concrete rather than co2 and this is not properly accounted for.
tonyb
Dear climatereason, thank you for pointing us to your posts on historical temperature measurements. Your explanation of the many reasons why observations from before 1900 likely have a warm bias are more likely to believed here at WUWT. It explains why the trend in the raw data is too shallow and becomes steeper and closer to the true trend after reducing these problems by homogenization.
Many of the problems you mention “only” cause a bias (for some stations) or make the measurement more noisy. This is a problem if you want to use the data to validate a weather prediction or if you would like to draw maps with isotherms and for many more detailed studies, but as long as a bias stays constant it does not preclude an analysis of the temporal variability and trends in the climate. If the bias changes and you have multiple stations, you can use such data to study temporal changes after applying homogenization to correct for the change in the bias.
Victor Venema,
I reply to your message as follows: I read on your blog (http://variable-variability.blogspot.ch/2012/01/homogenization-of-monthly-and-annual.html
) :
“For example, for the Greater Alpine Region a bias in the temperature trend between 1870s and 1980s of half a degree was found, which was due to decreasing urbanization of the network and systematic changes in the time of observation (Böhm et al., 2001).”
So, according to Böhm, on average in this area (including Geneva) UHI effect on thermometers in the nineteenth century was 0.5 ° C higher than in 1980.It is perfectly paradoxical.When one knows the evolution of urbanization in the twentieth century (UHI sources more than tenfold), it is just obvious that it has had a significant warming effect on temperatures measured. A continuous effect which must imperatively be taken into account.
You will reply perhaps that I forget TOBS. Actually, no. In the particular case of Böhm et al. 2001, I could demonstrate that the consideration of TOBS hardly changes anythingt to the value of 0.5 ° C for a century. But that is another topic.
No it is not because you forgot the TOBS. The statement claims that both TOBS and the decreasing urbanization of the network biases the trend.
With the “decreasing urbanization of the network” of the HistAlp dataset Böhm did not mean that the stations in the cities experience less urbanization, but that at the end of the series a larger fraction of stations was not situated in urban areas.
Zürich is probably a good example for this. The station is now at the airport and thus experiences a smaller Urban Heat Island effects as when it was in the city. Also cities are often founded in valleys, thus airports are often situated at higher altitudes and thus colder. This is typical for many countries.
I never studied it, but I would expect that the first stations were at universities, monasteries, capitals and courts and often operated by scientists, school teachers, apothecaries, lawyers, etc. Consequently, in the beginning most stations were probably in cities. Later on, when the network became denser and people tried to spread the stations evenly, more stations needed to be in smaller towns and villages. If this tendency exists, it would probably also be happing in most countries.
In Austria as a mountainous country and with lots of winter tourism, I can also imagine that having mountain stations and locating them in touristic villages became more and more important and improvements in communication and transport made them easier to maintain.
Victor Venema,
I talk about the steady increase of UHI which must imperatively be corrected. It was, according to Böhm already at least 0.5 ° C in 1870.
At this time the station of Geneva was still surrounded by meadows near a city of 50,000 inhabitants. In 1961, the typology is that of a city center of 300,000 people with a consumption of energy which has increased tenfold.
For Böhm for GHCN for the CRU, BEST and national offices, there is no further increase of UHI since the nineteenth century. In Geneva and around the world !
It is important, (and not difficult to achieve), that the original raw data be preserved and be readily available. Data sets should be linked to this raw data. Information on the homogenization methods used should be also accessible.
I agree that data should be open and that is the current climate of mistrust it is preferred that the algorithms used to produce data are also published.
There are a number of open datasets (GHCN, USHCN, ISTI, HadISD) and also most state-of-the-art methods for homogenization can downloaded from http://www.homogenization.org.
Thus you can do your climate research, if you do not trust climatologists to do it right or simply have an interesting new question.
Unfortunately much of the data from Europe is still closed. The European small-government advocates, wanting to pay less taxes, want the weather services to make money by selling the data. Due to abuse of raw data, such as in the book “State of Fear” by Michael Crichton, climatologists used to have reservations against giving out the original raw data. Nowadays, all colleagues I speak with are in favour of releasing the data, but the weather services are not allowed to. Putting pressure on the government to release climate data is a cause climatologists and climate “sceptics” can work on jointly.
The original data is nowadays always preserved. In the past, some of the quality control (removal of measurement mistakes, outliers) was performed on the paper records or in the beginning of the computer era when computer memory was expensive directly on the digitized records. For the monthly or annual means this is completely insignificant (the percentage of the data involved is very small), but for research on changes in extreme weather on daily data this could be more important and the daily data may need to be re-digitized so that the quality control is identical for the entire period. Sometimes, the original data is lost, for example in case of Austria the original data was lost during the WWII and all we have left are the monthly averages, which were reproduced in annual reports.