LITTLE_R Tutorial : Trouble Shooting
- The
little_r program produce extensive printout if the user sets
the print switches in the namelist to TRUE. Most of the
time, this is enough to capture the problem. The first
place to look is towards the end of the log file.
- To
find how many observation reports are available to the program,
search for "Number" in the log file:
Number
of observations successfully ingested:
1191.
Number of empty observations discarded:
4.
Number of observations discarded outside of domain:
534.
-
The little_r program must have the namelist.input file in the
working directory, the gridded input file must exist, the 3d
obs files must exist (though they may be empty).
The FDDA obs need to exist if the f4d flag is TRUE (again,
though they may be empty). The requested
dates in the namelist file must exist in the first-guess data
set.
- Frequent
troubles arise from trying to write observations into the format
required by the little_r program. If you are writing data
in this format, refer to the utility program upa.f
residing in the LITTLER/util directory.
- A
large disparity in the data distribution for the observations
(such as would happen if half of the domain has lots more observations
than the other half of the domain) increases the change for
a poor analysis if the multiquadratic scheme is chosen.
The multiquadratic scheme is two-dimensional, but the effective
radius of influence is the entire analysis domain.
|
|