RFM v4 Bugs
Bug-Finder Hall of Fame (excluding myself since I have an unfair advantage)
There are a number of other issues associated with LUTs and isotopic
profiles which will be sorted in the next release.
Note that the original patch rfmgrd.for introduced a new bug, so make sure you have the version dated 26FEB18
F-REAGRD: Unrecognised interpolation method=-71
F-REASPC: I/O error reading file, IOSTAT= 5001
when using irregular spectral grid output
F-INPNAM: Logical Error - unrecognised TYP
when creating Cooling Rate spectra using the
F-ATMCHK: PRE profile doesn't decrease monotonically
when LAY Flag enabled.
F-KEYCHK: *NTE section missing from Driver File
even though the *NTE section is present.
F-CHKLIM: Elev.Ang=60.0 [deg] has geom.tan.pt. < atmos., alt=-3.17E+03 [km]
when used in limb-viewing mode with line-of-sight specified as an upward (+ve)
I-XSCFIL: Using x/s data:
SB = exp ( c2 ELS
[ 1/Tref - 1/T ] )
c2 is the second radiation constant (1.44 K/cm-1).
ELS is the lower state energy [cm-1]
(taken from the HITRAN data), Tref is the reference temperature
(296 K) and T is the local atmospheric temperature.
Usually the ELS is a few hundred cm-1 but in HITRAN 2012
there are a number of very weak NO lines throughout the mid-IR with
ELS > 20 000 cm-1. A local temperature T=220 K
results in [ 1/Tref - 1/T ] ~ -0.001, which keeps the exponent
argument a sensible size (~ -30) but in the upper thermosphere,
temperatures of 1000 K are possible, giving a large positive argument
(>+80) which means that SB cannot be expressed as
single precision. The solution is to perform the calculation at double
precision and include some other scaling terms before attempting to
convert to single precision line strength.
F-RFMXSC: Logical error#1
when used with 2 or more
W-QTFCT: Extrap.beyond 70-1500K limits, T= -133.5273
when run in NAD/ZEN mode, probably with a finely spaced tropospheric grid
and ozone profile. Also shows up as a -ve Curtis-Godson temperature
in the PTH diagnostics.
STOP F-LUTTAB: Logical error#1 statement executed
when run with LUT flag and
isotopic Jacobians specified in the
section and the *JAC appears
before the *LUT
section in the driver table.
RVAL = 1.0
This seems to cause a problem with some compilers, but writing
it first into a separate character variable seems to get around the problem
WRITE (*,*) C9FLT(RVAL)
RVAL = 1.0
Similar problems affect the construction of error messages in chkprf.for
CSTR = C9FLT(RVAL)
WRITE (*,*) CSTR
F-VALATM: Logical error#1
when used with the FOV flag,
or with the OBS flag and the
Observer above the top of the atmosphere.
There are a number of other issues associated with LUTs and isotopic profiles which will be sorted in the next release.