(06/09/2003)
3DVAR produces a
number of diagnostics file that contain useful information on how the
assimilation has performed. This section will introduce you to some of these
files, and what to look for.
By this stage you
have successfully run the 3DVAR observation preprocessor, as well as 3DVAR.
Hopefully, you’ve also had a chance to plot some of the MM5 format analysis
increment fields using GRAPH. If not, you may like to return to the 3DVAR
Tutorial Page to try these out.
1. Which
are the important diagnostic to look for?
Having run 3DVAR,
it is important to check a number of output files to see if the assimilation
appears sensible. Change directory to where you ran the previous case-study:
cd
${RUN_DIR}/${RUN_ID}
ls –l
You will see
something like the following:
total 16507
-rwxr-xr-x 1 mmm01 system 8765 Jun 4 16:45 LANDUSE.TBL*
-rw-r--r-- 1 mmm01 system 4144 Jun 4
16:45 namelist.input
-rw-r--r-- 1 mmm01 system 2285 Jun 4 16:45 namelist.3dvar
-rw-r--r-- 1 mmm01 system 4638396 Jun
4 16:50 DAProg_3DVAR.analincs
-rw-r--r-- 1 mmm01 system 4638396 Jun
4 16:50 DAProg_3DVAR.analysis
-rw-r--r-- 1 mmm01 system 2156 Jun 4 16:50 DAProg_3DVAR.cost_fn
-rw-r--r-- 1 mmm01 system 2156 Jun 4 16:50 DAProg_3DVAR.cost_fn_grad
-rw-r--r-- 1 mmm01 system 19821 Jun 4 16:50 DAProg_3DVAR.statistics
-rw-r--r-- 1 mmm01 system 6097 Jun 4 16:50 da_3dvar.out
-rw-r--r-- 1 mmm01 system 806707 Jun
4 16:50 fort.47
-rw-r--r-- 1 mmm01 system 1182 Jun 4 16:50 fort.48
-rw-r--r-- 1 mmm01 system 3465430 Jun
4 16:50 fort.50
-rw-r--r-- 1 mmm01 system 3274840 Jun
4 16:50 inc
The most
important output files here are
da_3dvar.out:
Text file containing information
output as 3DVAR is running. Again, there is a host of information on number of
observations, minimization, timings etc.
DAProg_3DVAR.statistics:
Text file containing O-B, O-A
statistics (minimum, maximum, mean and standard deviation) for each observation
type and variable. This information is very useful in diagnosing how 3DVAR has
used different components of the observing system. Also contained are A-B
statistics i.e. statistics of the analysis increments for each model variable
at each model level. This information is very useful in checking the range of
analysis increment values found in the analysis, and where they are in the
grid.
DAPROG_3DVAR.cost_fn:
Text file containing values of the
components of the cost function at each minimization iteration.
DAPROG_3DVAR.cost_fn_grad:
Text file containing values of the
components of the gradient of the cost function at each minimization iteration.
DAProg_3DVAR.analincs:
MM5 format file containing analysis
increments (this is for diagnostics only and was covered earlier).
DAProg_3DVAR.analysis:
MM5 format file containing the
analysis. Rename this to MMINPUT_DOMAIN1 and run MM5 from it (later!).
Take time to look
through the textual output files to ensure you understand how 3DVAR has
performed. For example,
·
How closely has 3DVAR
fitted individual observation types? - Look in the DAProg_3DVAR.statistics
file to compare the O-B and O-A statistics.
·
How big are the
analysis increments? – Again, look in the DAProg_3DVAR.statistics
file to see minimum/maximum values of A-B for each variable.
·
How long did 3DVAR take
to converge? Is it converged? – Look in the da_3dvar.out
file, which will indicate the number of iterations taken by 3DVAR to converge.
If this is the same as the maximum number of iterations specified in the
namelist (DA_NTMAX) then you may need to increase this value to ensure
convergence is achieved.
5. What
next?
OK, you’ve run
3DVAR, checked out the diagnostics and are confident things are OK. Before
running a forecast, you must first modify the tendencies within the lateral
boundary condition files to be consistent with the new 3DVAR initial
conditions.
g) Updating lateral boundary conditions
Miscellanies:
Trouble Shooting: If you have questions, ask mesouser.