PROCESS_ANALYSIS_NOTES - 30 September 2008
The following describes procedures for conducting process analyses with
the CMAQ modeling system. Process analysis is a diagnostic tool that can
be used by the modeling analyst to quantify contributions of individual
science processes to model predicted concentrations. Such information is
useful for understanding a model's predictions and for determining why
predictions change when the model configuration or when model inputs are
changed.
Technical details of the CMAQ implementation of process analysis are
contained in Chapter 16 of the "Science Algorithms of the EPA Models-3
Community Multiscale Air Quality (CMAQ) Modeling System (EPA-600/R-
99/030), and will not be repeated here. The focus here is on the
operational steps necessary to invoke process analysis in the CMAQ system.
However, this science document has not been upgraded to reflect
modifications to process analysis, required by the new mass-conserving
advection scheme now available in the CMAQ chemical transport model
(CCTM).
The CMAQ build scripts included with this release are set up such that
process analysis is de-activated. This is accomplished by including three
"no-op" include files that effectively turn off process analysis -
PA_CTL.EXT, PA_DAT.EXT and PA_CMN.EXT. Whenever process analysis is to be
omitted, these three no-op files should be used when the CCTM is compiled.
To activate Process Analysis, the three no-op files must be replaced with
three include files that define how the information is to be collected
during model simulations. These files are generated by running the
Process Analysis Control Program (PACP), for which bldit and run scripts
are provided with this release. The steps to invoke process analysis
follow.
1. Create an ASCII text file containing process analysis commands. See
Chapter 16 of the aforementioned science document for command rules
and syntax. New syntax for the mass-conserving advection scheme in
the CCTM is discussed in special note d), below.
2. Build the PACP using the bldit script. Note that the CVSROOT
directory should point to the PROCAN subdirectory of the source
archive (e.g., in the accompanying bldit script, the variable
"Project" is set to $M3MODEL/PROCAN). The only change to the release
script that may be required involves the chemical mechanism. The
script variable "Mechanism" must be set to the chemical mechanism
that will be used in the CCTM.
3. Run the PACP using the run script. The input file to be used can be
defined using the environment variable PACP_INFILE. (If this
environment variable is not set, the PACP will look for an input file
named "pa.inp" in the run directory.) The output of the PACP
consists of four ASCII files: the aforementioned three include files
(PA_CTL.EXT, PA_CMN.EXT, and PA_DAT.EXT), and a PA_REPORT file that
is included for informational purposes. All four output files will
be placed in the run directory.
4. Use the three include files generated by the PACP program in lieu of
the no-op files when the CCTM is built (i.e., change the "ICL_PA"
variable in the CCTM bldit script to point to the path of the include
files generated by the PACP). The process analysis output files
generated by the CCTM will be included with the other CCTM output
files in the OUTDIR specied in the run script for the CCTM. See
special note b) below regarding setting the output domain before
running the CCTM.
Special notes:
a) When invoking process analysis, be sure to use the same chemical
mechanism when building the PACP and the CCTM. At present no
consistency check is performed, and the use of inconsistent
mechanisms will produce erroneous results.
b) As described in the science document, the domain for process analysis
can be set smaller than the CCTM computational domain to reduce
computer resources requirements. The science document indicates that
the process analysis output domain is set using the OUTPUT_DOMAIN
command in the PACP input file. That is no longer the case. The
process analysis domain is now set in the CMAQ run script via the
environment variables PA_BCOL_ECOL, PA_BROW_EROW, and PA_BLEV_ELEV.
When process analysis is invoked, these variables must be un-
commented and set appropriately. The PACP is now programmed to
terminate with an error message if the OUTPUT_DOMAIN command is
included in the PACP input file.
c) An example PACP input file and corresponding output files are
included in the "data" directory of this release, but these files
have not been used in the example CCTM run.
d) There's a new "base" operator, HADV for the mass-conserving advection
scheme that combines the X- and Y-advection. These cannot be simply
separated as is the case for PPM advection. All the operators that
combine the base operators have been renamed to contain only four
characters, like the base operators. This was done to allow a larger,
maximum 11 character length of the chemical species names, which get
prepended by the operator name (with the additional "_" character) to
become the I/O-API variable names (16 character limit) written to the
output file(s).
New combined operator names (created from base operator names):
1) ADV2 = XADV + YADV
2) ADV3 = XADV + YADV + ZADV
3) MADV = HADV + ZADV
4) TADV = XADV + YADV + ZADV + ADJC
5) TDIF = HDIF + VDIF
6) TRAN = XADV + YADV + ZADV + ADJC + HDIF + VDIF
7) TRNM = HADV + ZADV + HDIF + VDIF
(1) is horizontal advection for the PPM scheme
(2) is total advection without mixing adjustment for the PPM scheme
(3) is total advection for the mass-conserving scheme
(4) is total advection for the PPM scheme
(5) is total diffusion
(6) is total transport for the PPM scheme
(7) is total transport for the mass-conserving scheme
(1), (2), (4), and (6) can apply only to the PPM advection, whereas
(3) and (7) can apply only to mass-conserving advection.
e) Integrated Reaction Rate (IRR) Analysis. CMAQ includes integrated
reaction rate analyses that can be performed to assist the analyst in
understanding the underlying reasons for model predictions. In IRR,
numerical solutions of the gas-phase chemistry solvers are used to
calculate integrated rates of reactions during the model simulations.
Special accuracy requirements for these calculations preclude the use
of some solvers and may necessitate changes to the convergence
tolerances in other cases. Tests have shown that the EBI solver as
configured in CMAQ is insufficiently accurate for use in IRR analyses
(the inaccuracy here applies to the IRR part, not to the gas-
chemistry solution part). Although accuracy could be improved
somewhat by adjusting convergence tolerances and reducing the
chemistry integration time step, the accompanying loss in
computational efficiency would negate any advantage of this solver.
Hence, IRR analysis is not implemented in the EBI solver. With
respect to ROS3 and SMVGEAR, tests have shown both to be sufficiently
accurate for IRR analyses in most cases. However, if IRR results are
being computed for fast reacting radicals such as OH and HO2 with the
ROS3 solver, then it is recommended that the absolute tolerance be
decreased from the default value of 1.0E-07 ppm to 1.0E-09 ppm by
setting the environment variable RB_ATOL as described above. The
default SMVGEAR convergence tolerances should be sufficient for most
applications.