cc: c.harpham@uea.ac.uk, c.g.kilsby@ncl.ac.uk
date: Sun, 2 Nov 2008 11:19:11 -0000 (GMT)
from: C.Harpham@uea.ac.uk
subject: Re: Adjusting the WG to MOHC/UKCIP08 Tmean and Range
to: P.Jones@uea.ac.uk

Phil,

I have gathered together all the plots done so far for London (attached).
There is an added bonus in doing this that needs to be emphasised - the
heat island effect, the UKMO data takes land use into consideration and
from the adjusted WG runs (which are virtually the same as UKMO) the heat
island effect has been captured (to some extent anyway). For areas such as
the Sheffield grid the changes are fairly small for all but the urban
areas and then <1.0 deg. London's are only > 1.0 for the city.

The WG Vas sent me takes just over a minute a run, lets say 1.5mins which
for 8000 cells equates to roughly 8 days continuous cpu time. I could do
1000 cells at a time and then calculate the stats (which takes longer than
a WG run). So, approx a month perhaps.
Problems.
The WG does not run reliably all the time, it occasionally loses server
connection and stops, a pain for overnight runs.
When running the WG I am unable to do anything else, it seems to hog all
resources even with a dual processor. Could temporarily get another PC
though.

Chris, need to resolve above before launching into this - it was not a
problem for the smaller runs.
If I were to do the calcs in the same order as the batch runs I could
start feeding cfs (by the 1000 say) as we go along.
Phil, will check mail tonight if you have any queries.

cheers
Colin

>  Colin,
>   Here is a doc that came from Chris whilst I was away.
>  I thought you were on it, but you're not. Anyway here
>  it is. Also attaching my ppt that I've just made up
>  for Monday's meeting with DEFRA in London.
>
>    What I'm going to tell them s that we will have
>  to do the adjustements for all boxes - a la what you've
>  done for SCORCHIO.
>
>    Need to do for Tmean and also for Range. Can you make
>  an estimate on Monday how long this will take? We will
>  need to get these adjustments to Chris before he runs
>  any of his batch runs. There is still some time, as
>  he doesn't yet have the final CFs due to the temperature
>  variance issue.
>
>    I've convinced myself looking at these that doing the
>  adjustments improves all these ten locations. I know
>  why the spatial interpolation does this, so not difficult to
>  explain.
>
>  Cheers
>  Phil
>


Attachment Converted: "c:\eudora\attach\London_various_plots.doc"

Attachment Converted: "c:\eudora\attach\Manc_Shef_various_plots.doc"

Attachment Converted: "c:\eudora\attach\lond_cfs_Tmean.csv"

Attachment Converted: "c:\eudora\attach\manc_cfs_Tmean.csv"

Attachment Converted: "c:\eudora\attach\shef_cfs_Tmean.csv"