[ExI] Living temperature dataset

[Rafal Smigrodzki]

2009/12/12 Alfio Puglisi <alfio.puglisi at gmail.com>:

>
> If one suspect that adjustments are systematically biased in one direction,
> the correct thing to do is to take all of them and look at their
> distribution. This is the result for the entire GHCN dataset, on which GISS
> is based:
>
> http://www.gilestro.tk/2009/lots-of-smoke-hardly-any-gun-do-climatologists-falsify-data/
>
> You get a nice gaussian distribution with an average of 0 degrees. That is,
> there are as many negative adjustments as there are positive ones

### See here:

http://wattsupwiththat.com/2009/12/09/picking-out-the-uhi-in-global-temperature-records-so-easy-a-6th-grader-can-do-it/

- please watch it and read the article before commenting.

Basically, since GISS homogenizes data containing a mixture of urban
and rural stations, the procedure introduces a spurious warming trend
due to the UHI effect. The overall distribution of adjustments proves
nothing about the actual net effect of the adjustments.

Let me explain on a hypothetical scenario:

Imagine that you take all rural site temperatures and adjust them
upward by 1 degree. Then you take an equal number of urban sites and
adjust them downward by 1 degree. Obviously, the net adjustment per
site will be zero, just as described in Alfio's link. However, note
that adjusting rural sites up doesn't make physical sense, since there
is no "rural cooling island effect" you would need to adjust for -
these data should be consumed raw. Adjusting urban sites down makes
sense, since their temperatures are the result of the UHI - but in
this hypothetical this physically correct adjustment is negated by a
physically improper adjustment of the rural data. The overall effect
is that the UHI is fully transferred into the homogenized temperature
record, and a spurious warming trend is seen.

Lest you think this is just theorizing - homogenization (averaging) of
urban and rural measurements does produce exactly the same effect on
the raw data. This is not surprising, since homogenization based on
proximity (as used in the GISS/CRU procedures) is a naive approach
which fails to take into account the underlying physics. The
physically proper procedure, aimed at assessing true global
temperature variability, would simply discard urban data, since they
reflect *local* influences. Indeed, for rural sites in the US that can
be paired with urban sites, there is no warming trend, as demonstrated
by GISS data.

So, Alfio, it looks to me like the consensus GISS/CRU climate record
is systematically biased upward, over multiple stations, and the
analysis you quoted does not disprove this assessment.

Rafal