date: Fri, 29 Aug 2008 17:35:23 +0100
from: "Douglas Maraun" <d.maraun@uea.ac.uk>
subject: Error found :-) :-(
to: "Osborn Timothy Dr (ENV)" <t.osborn@uea.ac.uk>

Hi Tim,

I found the error. Could be important, so here it is:

I told you that I replaced the reading in from a file just by the
standard set of the parameters (esp for MAGRUN.CFG).
Later in the program, where the loops for the ensemble start, these
are overwritten, first by the values set in the loops, but also by a
set of standard parameters, which are set just before the climate
sensitivity loop.

First of all, some of these standard parameters are just altered some
lines down (TE, XKLO,XKNS), so they are actually never used. Anyway,
this is not important.

But:
Apparently, before these loops, the data read in from the MAGRUN.CFG
are already used to do some calculations. The error occured, because
at that point I commented these values out where they used to be read
from MAGRUN.CFG, because they were set later anyway. Without
commenting them out, there is no difference.
That means, a part of the program (some initialisation) runs with the
parameters from MAGRUN.CFG, the main integration with the parameters
set later in the loops.

Could this be important for the results? I mean, it causes a quite big
difference in the results, so it should be important...

Quite scary, the whole program...

Anyway, the program runs now, and I can read in data from a new
parameter file which R will produce for every iteration. Each one
should take less than a second, so we might end up with reasonable
times (less than a week on a PC) even for the specificity study for
different observational noise.

Cheers and have a good weekend!

Douglas
-----------------------------------------------------------------------
Dr. Douglas Maraun
Climatic Research Unit, University of East Anglia
+44 1603 59 3857
http://www.cru.uea.ac.uk/~douglas
