[RegCNET] RegCM3 crash - kind of like CFL violation
Michal Belda
belda at cs.cas.cz
Mon Feb 6 18:37:29 CET 2012
Dear RegCNETters,
I'm facing a problem with the model crashing after a couple of years of
RegCM3 simulation forced by ERA-Interim on the 50km European domain as
defined by the CORDEX guidelines. In the last 6 hours of 20080531 the
model crashes with ps=NaN message no matter how short a time step I set.
I've gone down to dt=30s, which is pretty short for a 50km simulation,
but still no improvement. The strange thing is, this always happens
after reading the 2008053118 BCs even with different time step settings.
Such situation usually translates to a corruption in the ICBCs, only I
couldn't find any, the fields look consistent enough. I've been using
the ERA-Interim data from the RegCM4 data web page, I even downloaded
the 2008 data again and went through the ICBC preparation once more, but
nothing helped.
By the way RegCM4 run with ICBC files prepared from the same ERA-Interim
files finishes the whole 1989-2008 simulation without any crash.
Can anyone help with this please?
Thanks for any pointers.
I'm also pasting a fragment of the log file below.
BATS variables written at 2008053115 180.000000000000
at day = 7090.6552, ktau = 116550 : 1st, 2nd time deriv of ps
= 0.79588E-05 0.27204E-07, no. of points w/convection = 141
at day = 7090.7073, ktau = 116600 : 1st, 2nd time deriv of ps
= 0.69961E-05 0.20833E-07, no. of points w/convection = 71
*** solar declination angle = 22.10 degrees.
dectim = 178560.000000000
OUT-history written date = 2008053118.00000
BATS variables written at 2008053118 0.000000000000000E+000
Writing rad fields at ktau = 116640 2008053118
BCs are ready from 2008053118 to 2008060100
at day = 7090.7594, ktau = 116650 : 1st, 2nd time deriv of ps
= NaN NaN, no. of points w/convection = 0
at day = 7090.8115, ktau = 116700 : 1st, 2nd time deriv of ps
= NaN NaN, no. of points w/convection = 0
at day = 7090.8635, ktau = 116750 : 1st, 2nd time deriv of ps
= NaN NaN, no. of points w/convection = 0
--
Michal Belda
Department of Meteorology and Environment Protection
Faculty of Mathematics and Physics
Charles University in Prague
More information about the RegCNET
mailing list