From: Steven Pawson [mailto:pawson@dao.gsfc.nasa.gov]
Sent: Thursday, September 20, 2001 8:57 AM
To: Scott Denning
Cc: Steven Pawson
Subject: Re: Reanalysis, etc.


Dear Scott,

   all is returning to 'normal' again here.

   A couple of days ago I had a small meeting with some
DAO staff to discuss a reanalysis. The basic answer we have come up
with is this: while we could technically manage this, there is a
concensus that it would be too premature to complete 25 years before
the middle of 2004.  The reason for this is that there are 'issues'
that we could work on to improve the science quality of the product
(for carbon cycle and other users). Having more time to work on these
and provide comprehensive assessments for shorter periods would lead to
us having a much more viable product, but this would need at least one
more year (maybe two).

   The mid-2004 timeline would mean that within a year from
now we would have to start production of the 1979-present period.
(Throughput on the computers is an issue, and even this starting date
would assume no major catastrophes - long real-time periods were we
could not run the system.) That means that we would be running the
high-resolution system in 'reanalysis' mode beginning only just after
we start it in 'real-time' operational mode, so we would not have a
large validation dataset.

   We at the DAO would also not be able to do the nested
(10-km) analysis (we kind of knew this bit anyway).

   There are options. We could aim to to the reanalysis, but
with a more realistic completion date (say mid 2006). This would be
good for retrospective analysis of the NACP intensive field data.  We
could provide model runs at high resolution (but here we are, of
course, subject to model bias): we have vast experience with AMIP-like
runs.

   What do you think right now? It means my text will have
to be modified.  This does not affect our ability to produce real-time
support during the campaigns.

   Please share this message with others, if you think this
will be productive.

   Steven