A major catastrophe occurred at NCEP yesterday, and it
will create major changes for the RUC (and all operational
models). There was a fire in the Cray3 system, the one where the
models are run, which had to be extinguished with a chemical
powder. Needless to say, much damage was done to the system,
and the damage is still being evaluated but appears to be
quite serious.

 What does this mean for the RUC? Well, it is fortunate
the we just recently finished converting the current RUC code
for use on the new IBM supercomputer. The original plan was
to make the IBM machine operational in January 2000, but for the
next few days, the RUC will be run operationally on the IBM.
While we have examined a lot of test output from this new
system along the way and the code is intended to match that
which has been running on the Cray3, we can't guarantee that
everything is perfect and encourage you to post to the forum
if you see something unreasonable in the output.

 Things get tricky later this week. The IBM machine is
being moved to a new location, and the plan before this
disaster was to begin moving it on Friday. If this plan
stays in place, we will not be able to run the RUC on that
machine for about a month. Instead, we would resort to
using an older version of the model run at a 60-km resolution
on Cray5. Due to constraints on that machine, the RUC could
only be run at the 00,03,06,09.... times. It would not be
hourly. Again, this scenario is assuming the worst for the
Cray3 - we don't know things for certain yet.

 As you can imagine, things are very hectic at NCEP today,
and no final decisions have been made. I will post a follow-up
later this week when I know more. In the meantime, however,
be aware that the model is running (with a version intended
to be very close to the Cray version) on the IBM machine
this week.