Date: Wed, 12 Oct 1994 16:23:14 -0500
From: fan@meteor.atms.purdue.edu (Zengquan Fan)
Message-Id: <9410122123.AA17652@meteor.atms.purdue.edu>
To: ccm-users@ncar.ucar.edu
Subject: HELP
Problem w/CCM2 Perpetual Run Simulations!!
- We recently ran a perpetual run using January initial conditions with
the intent of running out about 400 days; thus, the only difference from
the original code is that we flipped the switch from seasonal to perpetual
mode!! After approximately 50 days the model blows up, indicating that we
have incorrect values for specific humidity which we definitely do as the
model outputs negative specific humidity values!? Upon closer inspection,
the error comes about due to a sudden, dramatic increase in the global
RMS divergence, which we feel is probably due to an anomalous regional
increase in surface convergence in the tropics (this may not be confined
to tropics but just happens to be where the model bombs on us; 35th long.
and 37th lat. parallel). The anomalous convergence field leads to a huge
perturbation in the vertical velocity field, resulting in a violation
of the vertical CFL condition in the model!! In turn, in the subsequent
time step the horizontal CFL condition is violated(dramatically) and the
global rms. temp rises by 25K, presumably in response to ridiculously
large condensational heating. The negative 'Q' values result from
excessive condensation as more water vapor is condensed than is actually
available in the local atmospheric column. So it is undoubtedly a
positive feedback loop, probably kicked off by an anomaly in the near sfc.
horizontal wind field, which screws up the div. field, etc., etc...
Has anyone encountered this before? Is it something specific to
a perpetual simulation??