Click images for more details



Recent comments
Recent posts

A few sites I've stumbled across recently....

Powered by Squarespace
« Keifer Sutherland | Main | Private prosecutions »

It's all in the adjustments

This is odd. Or do I mean appalling?

When you measure the surface temperature, the data that comes out of the station network is poor, and has to be "fixed". This is done by means of a series of adjustments which are added stepwise to the raw data to give the final answer. 

I've show below a graph of the difference between the raw temperatures measured in the USHCN surface station network, and the final temperature delivered as an output.

What this appears to show is that most of the observed warming is coming from the adjustments, not the weather stations. (I'm assuming here that the trend in the final temperature is not more than 0.6oC)

The page from which the graph is ripped explains what the adjustments are:

  • Time of observation. Different stations measure temperature at different times of day, but you want every station's midnight temperature. You therefore adjust anyone who is not reading at midnight, creating an estimate of what temperature it would have been if they had have done it at the correct time.
  • Station moves
  • Changes in equipment
  • Missing data
  • Urban heat island - as urbanisation takes place, an man-made warming trend is introduced, which needs to be eliminated to give the true temperature.

It also gives the impact of each. In the graph below, each line represents one of the adjustments.


From this, we can see that the warming trend is being produced by the time of observation adjustment (black) and  by the station move adjustment (yellow). 

I can think of no earthly reason why time of observation adjustments would produce this shape. The upward slope of the adjustment implies that there are many stations recording temperature at a time when it's colder than midnight. This means the wee small hours I guess. Why would this be? And why would the effect be increasing? I mean, over the last century more and more stations will be automatic, which presumably means that you could get temperature exactly when you want. Why then, does the raw data appear to be getting worse - ie the adjustment required to correct it is getting larger?

It all looks a bit fishy if you ask me.

PrintView Printer Friendly Version

Reader Comments (21)

A bit fishy... Your Grace is unusually restrained.

I'm sure the scientists at NOAA must be doing what they believe is truly "scientific", but out here none of us has had any idea what kind of data devilling they've been doing.

Jul 23, 2007 at 8:59 AM | Unregistered CommenterWat Tyler
I note the page claims "The cumulative effect of all adjustments is approximately a one-half degree Fahrenheit warming in the annual time series over a 50-year period from the 1940's until the last decade of the century."
As far as I can make out this compares to the reported rise of about 0.6 degrees Farenheit over the same period (The dates are imprecise) - and a lot of that warming is claimed to be due to reduced particulate pollution....
Jul 23, 2007 at 9:43 AM | Unregistered CommenterThe Englishman
But then you should try making sense of the *sea* temperature readings which also play such a large part in this travesty of science. Some months back Steve McIntyre's site, Climate Audit, explained some of the intricacies, and the confusions, to be found there; arising in part from the change (after WWII, as I recall) from hawling a bucket of sea-water out and taking its temperature, to the use of built-in sensors many of which are placed to close to warm components of the ship itself thus producing a 'maritime heat island' effect.

You couldn't make it up - but climate 'scientists' do!
Jul 23, 2007 at 9:59 AM | Unregistered CommenterDavid Duff
"Hawling"? Hauling! Hell, you say 'potaytos, I say 'potartos'.
Jul 23, 2007 at 10:02 AM | Unregistered CommenterDavid Duff

I think the 0.6 degree rise you are referring to is the global figure. The graph in my post is for the US network alone.


I've seen McIntyre's bucket adjustment piece. It is another truly awe-inspiring piece of "science".

If anyone reading this thread hasn't seen it, read these:
Jul 23, 2007 at 4:08 PM | Registered CommenterBishop Hill
I've been a regular visitor to ClimateAudit and it's qute breathtaking sometimes to read some of the stuff that Steve M post son his blog. Fro the whole hockey stick debacle to the bucket adjustments to Antony Watts' recent auditing of the weather stations its become increasing clear that the AGW Emperor most definitely doesn't not have any clothes.
Jul 23, 2007 at 6:25 PM | Unregistered CommenterKevinUK
I agree. There is a real problem in getting the public level of knowledge on the subject up though.
Jul 23, 2007 at 7:39 PM | Registered CommenterBishop Hill
I have long been supicious of results where the adjusted, averaged, massaged and generally buggered about with numbers come up with answers with places of decimals when the original measurements were to the nearest degree, with a plus/minus half a degree error!
As a sort of related comment that I have been dropping around the place to highlight mankind's insignificance.
Imagine a mile cube, or if it's easier, a building ten miles long, one mile wide and 528 ft high filled with compartments approximately 6ft high, 2ft wide and fifteen inches front to back.This would hold the entire population alive today! Not comfortably, indeed, probably not alive, but that's all there is of humanity, less than one cubic mile. (quite a bit less, actually)
And we think we can affect the climate! We can't even keep our houses dry.
Jul 23, 2007 at 11:52 PM | Unregistered CommenterTony Edwards

"2ft wide and fifteen inches front to back."

You'd struggle to get your average midwestern American in that footprint.

Jul 26, 2007 at 2:53 PM | Unregistered CommenterCleanthes
Yeas, but I'm talking averages, the children alone would make up the difference. Any way, make it two foot three wide and eighteen inches fat, you'ld still fit them in.
By the way, it has been estimated that ants would fill a ten cubic mile volume. So who matters most?
Check out Junkscience, he has this
Major interest, albeit long and complicated read.
Jul 27, 2007 at 12:19 AM | Unregistered CommenterTony Edwards
Englishman, According to the IPCC, the increase is 0.6 degrees Celcius not Farenheit. 0.6 degrees C is 1 degree F, that would make a 50% difference, non-trivial. The USHCN is probably the best temperature information in the world and its so polluted that it is almost unusable. Corrections have been made that just don't make sense, some of these are detailed at Climate Audit.
Jul 27, 2007 at 5:30 AM | Unregistered CommenterJohn Nicklin
Hey, bish, the settings on my laptop that make the fonts on most blogs big enough for me to read don't work on your blog.
Jul 30, 2007 at 10:21 AM | Unregistered Commenterdearieme
Hi Dearieme

I thought I'd increased the font size by one to get it working for you.

I'm going to take it up with Squarespace to see if they can get things working for you.
Jul 30, 2007 at 3:30 PM | Registered CommenterBishop Hill
The increase font size works for me. Squarespace are saying it works for them too.

I'm at a bit of a loss.
Jul 30, 2007 at 9:51 PM | Registered CommenterBishop Hill
I have looked into the Time of Observation adjustment in the USHCN data and all I can say with certainty is the adjustment is part Time of Observation Bias and part estimating error.

TOB occurs when observations are made close to when the high of the day occurs (afternoon) or the low of the day (early morning), either double counting warmer than average days (to get a warming bias) or cooler than average days (to get a cooling bias).

This is where it gets interesting. TOB can be significant on a monthly basis, because the day you double count could be from the previous month, which in temperate latitudes is statistically warmer or cooler for most months.

Observers do record their time of observation, but this is not used and a statistical technique is used to estimate the time of observation and then standard TOB adjustments derived from a sample of stations are applied.

BTW, the justification for estimating time of observation is 'to save data entry'. I kid you not.

I have been unable to find any documentation on how monthly TOB is turned into yearly TOB and I assume they just average the monthly values.

The problem here is that yearly TOB is smaller than monthly TOB and I am sure they are magnifying whatever error there is in the monthly data (the originator of the TOB adjustment method, Karl, has said it is around 25% in the monthly data).

IMO the annual TOB adjustment is mostly error.

Now to the trend since 1950. I can explain the magnitude of it (a little over -0.2C) but not why it's applied as a +ve adjustment of around +0.2C in recent years.

Automated measurement should occur at midnight and hence have no TOB. Over time automated measurement has replaced manual measurement, which was recommended to be done late in the day, and which would have produce a +ve TOB, requiring a -ve adjustment. So as automated observations increased we should have seen a decreasing -ve TOB adjustment, and not an increasing +ve TOB adjustment.

I'm surprised other people haven't picked up on this. That increasing +ve TOB adjustment since 1950 has to be wrong and bang goes nearly half of global warming over that period (although I should point out as a reduction in average temperature and not as a reduction in the trend).
Aug 29, 2007 at 5:54 AM | Unregistered CommenterPhilip_B
I've always been a global warming sceptic. The more I look into it, the more sceptical I become.

It never ceases to amaze me that the last climate crisis has been conveniently ignored - the fact that 30 years ago there was a genuine worry about global *cooling* and the fact that since WW2, global temperatures had been decreasing. It is all BS designed to give governments an excuse to generate revenue.
Taxing motor vehicles in the UK, according to supposed CO2 output is a prime example of this, but something smells fishy about it: If each vehicle was putting out nearly a quarter kilo of CO2 per kilometre, as claimed, the concentration of CO2 in cities and around busy roads would be so high that people would be collapsing through lack of oxygen and our dogs would be dropping dead on the ends of their leads, as we walk them. I'm no expert, but these figures must be out by 3 or 4 orders of magnitude.
My point? More fake BS to back the pseudo science being spouted as an excuse to tax citizens harder and prevent developing countries getting anywhere near western levels of development.

I'll shut up now, because I'm getting angry at how easily the sheep around the world believe the pap governments feed to them about anything.


Aug 29, 2007 at 2:54 PM | Unregistered CommenterFr J

Run that by me again. They don't use the actual time of observation for each station, but use a statistical technique to estimate it? Is there an explanation for why they do this?
Aug 29, 2007 at 8:44 PM | Registered CommenterBishop Hill
The fraction of observers recording at various hours of the day was calculated and interpolated for intervening years (extrapolated for subsequent years). For these seven states, the ending time of observation was grouped into three categories: AM, PM, and MD. The AM category included observers who ended their climatological day between 3 AM and 11 AM; the PM category between noon and 9 PM; and the MD category between 10 PM and 2 AM; all local standard time. The fraction of observers in these categories was calculated, and it was assumed the 7 AM observation time best represented the AM category; the 5 PM observation time, the PM category; and midnight for the MD category. The reason for the simplification was to test if a faster method, requiring significantly less bookkeeping and keypunching, could not provide nearly as good results as calculating the fraction of observers at each of the 24 hours of the day.

Reference: Karl, et al. (1986): "A model to estimate the time of observation bias associated with monthly mean maximum, minimum, and mean temperatures for the United States" (Thomas R. Karl, Claude N. Williams, Jr., and Pamela J. Young, National Climatic Data Center, and Wayne M. Wendland, Illinois State Water Survey, Journal of Climate and Applied Meteorology, January 1986, vol. 25, pp. 145-160, American Meteorological Society, Boston, MA).

BTW, the Karl TOB estimating method referred to is different to the method described here. So the method of estimating Time of Observation probably varies between datasets.
Aug 30, 2007 at 12:05 AM | Unregistered CommenterPhilip_B
And note the considerable irony of spending untold billions based on data that contains estimating errors made to save a few thousands.
Aug 30, 2007 at 12:16 AM | Unregistered CommenterPhilip_B
The ability of the climate science community to surprise is quite stupefying. I just don't know what to say.
Aug 30, 2007 at 8:53 PM | Registered CommenterBishop Hill

see this

Oct 3, 2009 at 3:30 PM | Unregistered CommenterDeeptPexNup

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>