[email protected] wrote:
1) Can you measure temperature with error less than 0.1C?
Why do you *think* they take a whole bunch of measurements and average them?
Because they believe averaging will reduce that error. I think they are
mistaken.
2) Do you know about how big is measurement uncertainty for your data?
You *do* realize that instrument calibration is a pretty well understood field?
Yes, I have background in measurement theory..
3) Do you know that satellite temperature measurements are models,
calibrated by ground temperature?
Did you know that the mercury thermometer is just a model based on expansion
and contraction of a liquid? Oh noes, we can't trust them. Or the ones
based on two strips of metal with different coefficients of expansion. Oh
wait, we can't trust *any* measurement, because its a model calibrated against
something else...
Lol, but that satellite measurement is done through poorly understood
area (the atmosphere), calibrated with biased temperature measurements
and designed with belief that layered structure of the atmosphere is
constant over whole earth (which is not true). And then presented as
reality. What should I think of such a people?
4) Do you know that most of temperature measurement is done in urban
areas, biasing the measurement?
Unless youre referring to the fact that there's a higher density of
measurements in urban areas (for example, could be 1 per square mile in
the city but only 1 per 40 square miles in the country). Rest assured
that is *not* a major challenge for anybody who actually understands how
to do modelling, because you *never* get a nice perfect rectangular mesh
of perfect sensor readings. You get messy data, broken sensors, that
one area you couldn't put sensors into the experiment because there was
a structural support there, the subject sneezed and moved slightly,
etc etc etc. So dealing with missing/incomplete data has been understood
for as long as scientists have been analyzing datasets.
It is more difficult than you think. I tried to find info on some of
weather stations in the US. The ones which seemed suspicious the moment
I looked at the data. Then I googled photographs of some of them. One
was on a parking lot, second was just after pub's ventilation etc.. At
the moment I suspect that there are more of highly biased weather
stations. If I'm right, when anyone calculate average temperature from
such a data, it will show bigger temperature growth. If there is only a
small number of highly biased stations, everything is ok of course.
I should cluster the stations by data, but it wouldn't prove anything
without someone who would look at the actual station in natura. I have
no time for trip over the US photographing weather stations, so I can't
verify it.
5) Do you know something about 'butterfly effect'? (great differences in
state trajectories by some time, which reduces prediction horizont of
your model, depending on measurement error)
Did you know that the scientific discipline that first understood the
butterfly effect was, in fact, meteorology? They've been working at
understanding it longer than anybody else.
Yes, I know about Lorentz, who discovered that effect. I do not take
'they have been working on it longer than..' as an argument.
6) Do you know that reality is more real than any model?
Oh, give me a *fucking* *break*. It's called the Scientific Method. Make
a hypothesis, see how close it fits to reality, and fix it when it doesn't.
I know this is trivial. But I have seen some graphs, related to
climatology, which speaks for itselves. :-( Look for example at IPCC
reports. Look at the graphs. You understand this, but they don't. (Or
they do but want to produce more scary graphs to get more money)
Yes, I am slightly paranoid on IPCC. I consider IPCC as
environmentalistic institution.
I am quite sure most of IPCC climate scientists won't pass even 3 of
these 6 points. :-((
I'll take that wager. :)
:-)
--
Martin Tomasek
_______________________________________________
Fun and Misc security discussion for OT posts.
https://linuxbox.org/cgi-bin/mailman/listinfo/funsec
Note: funsec is a public and open mailing list.
