Skip to content

Temperature Adjustments At Valentia Observatory

October 27, 2014

By Paul Homewood

 

A little bit more information on Valentia Observatory, the long running, high quality, rural site in Ireland. As we saw yesterday, GHCN algorithms had added an artificial warming trend, based on trends from nearby urban sites.

 

First, the GHCN plots showing the unadjusted and adjusted data.

 

62103953000

ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v3/products/stnplots/6/62103953000.gif

 

Let’s home in on the graphs, to see them more clearly. The top, red, plot is the unadjusted trend.

Essentially, all temperatures prior to 1967 have been arbitrarily reduced by 0.4C. There is nothing in the temperature record to suggest any break points.

 

62103953000

 

 

 

We can also see the same effect on the GISS temperature record. The next graph is based on the raw data, available through the now defunct GHCN V2 database, which gives data up to 2011.

 

station

http://data.giss.nasa.gov/cgi-bin/gistemp/show_station.cgi?id=621039530005&dt=1&ds=1

 

The data is available here.

 http://data.giss.nasa.gov/tmp/gistemp/STATIONS/tmp_621039530005_1_0/station.txt

 

Now compare with the current picture, based on GHCN V3 adjusted data.

 

station

http://data.giss.nasa.gov/cgi-bin/gistemp/show_station.cgi?id=621039530000&dt=1&ds=14 

 

Data here.

http://data.giss.nasa.gov/tmp/gistemp/STATIONS/tmp_621039530000_14_0/station.txt

 

 

Finally, Ronan Connolly includes this map in his paper, showing the “neighbouring” stations, which have been used to homogenise Valentia’s record. Incredibly, they take stations as far away as France, including, it would appear, Paris.

 

image

 

The whole thing is a nonsense.

5 Comments
  1. October 27, 2014 1:28 pm

    Why even put up thermometers and keep records if one just makes up the data anyway?

  2. David permalink
    October 27, 2014 1:48 pm

    Thanks Paul,

    For some reason the raw GISS data disagrees with the raw BEST data: http: //berkeleyearth.lbl.gov/auto/Stations/TAVG/Text/155842-TAVG-Data.txt

    In any case, as the unadjusted plot shows, the NCDC algorithm doesn’t “introduce” a warming trend as there’s already one in the raw data. In BEST for 1880-2011 the raw trend is +0.42C/century; in GISS it’s +0.34. If the NCDC algorithm produces “roughly +0.40” then it’s a reasonable fit for the raw data from either source.

    The BEST algorithm also spotted the mid-1960s breakpoint identified by NCDC: http://berkeleyearth.lbl.gov/auto/Stations/TAVG/Figures/155842-TAVG-Alignment.pdf

    The adjusted BEST data has an even faster trend than the NCDC adjusted data (+0.55C/century from 1869-2013): http://berkeleyearth.lbl.gov/auto/Stations/TAVG/Figures/155842-TAVG-Comparison.pdf

    • October 27, 2014 3:42 pm

      The NCDC algorithm adds a warming trend to the raw data they have, because they are comparing against urban sites. This is fundamentally wrong.

      Whether BEST have a different set of raw data is neither here nor there.

      • David permalink
        October 27, 2014 5:32 pm

        There’s already a warming trend in the raw data used by NCDC. The top red plot of the second chart clearly shows a warming trend in the unadjusted data.

        If this is the same raw data as published on the GISS link, then it’s +0.34C/century, which is just fractionally slower than +0.40C/century.

        The adjustments may have made the tend slightly faster, but they didn’t ‘add a warming trend’, because one already existed.

  3. October 27, 2014 5:50 pm

    Actually you can see the Gleissberg cycle in these data,
    with again a shift in temperature from the time around the seventies when we started recording automatically (with recorders and computers) every second, with thermo couples, instead of before:
    human observations 4-6 x per day, [if you were lucky and nobody went on leave] with non-re-calibrated thermometers before the 1950s …..
    (show me a re-calibration certificate of thermometer before 1948?)
    Which results do you trust more?

Comments are closed.

%d bloggers like this: