------
Expense
Relative timings:
WRF meteorology 11
+ gas phase chemistry 11+61
+ aerosols 11+61+88
+ SOA (Zaveri's version) 11+61+88+74
+ cloud/aerosol interactions 11+61+88+74+89
For a 24 hour simulation with chem_opt=2 and 2 domains (146 x 103 x 29) (175 x 175 x 29) with time steps of 200 and 66.67 seconds a run on discover with 32 processors takes 1 hour and 7 minutes.

(Elena) I ran an Aug 1 -17, 2002 simulation with a restart on Aug 9. So a one-week simulation should run within the allotted 12 hours of wall time with 32 processors requested. But this is a run with radm2 and sorgam only, no aqueous chemistry. Sorry I don’t have the specific numbers.


------
1-way vs 2-way nesting?
Currently 2 way nesting with feedback is not working for WRF-chem. Hopefully this will be fixed in the next release of WRF-chem.
------
Restarts with and w/o aerosols
Currently restarts are not working with the MADE/SORGAM aerosols routine. This may be because there are some arrays that need to be output in the restart file that are not included in the registry file.
------
nudging
Currently nudging is available in WRF, but does not work in WRF/Chem. Nudging is supposed to work in the next version of WRF/Chem.


------
global boundary conditions
Beta code that will add chemical initial and boundary conditions from MOZART and RAQMS output can be obtained from . This code can be modified to extract other global model output for WRF-chem initial and boundary conditions.
------
SMOKE-RT
John McHenry at Baron Advanced Meteorological Systems is preparing the SMOKE emissions processing system to be compatible with WRF-chem. This is scheduled to be released with the next version of WRF-chem.
------
aqueous chemistry
(Elena) Documentation about turning on aqueous chemistry is very terse. The options I use:

chem_opt=11 (radm2, sorgam, and aqueous chemistry) (package radm2sorg_aq = found in registry.chem)

progn=1 (based on error messages this option is needed in &phys section)

The model core dumps after several time steps. I am investigating whether this is an NCCS issue or problem with WRF/Chem setup.
------
space issues on DISCOVER
(Elena) I do all work in the nobackup directory. The allotted space of 100Gb is not enough to run the simulations. Especially when the date specific emissions files for a one/two week run are huge. So I move the output files to $ARCHIVE as they are complete by WRF.
------
throughput issues on DISCOVER

(Elena) For small test runs, the jobs are executed very quicly with little or no wait time. But for long, intensive runs wait time can last from a few hours to days, especially on the weekend.
------
emissions: How realistic is the default emission inventory
(Elena) The default emission inventory is the EPA NEI99 inventory. Comparison of WRF and SMOKE emissions shows that SMOKE emissions are more sophisticated and up to date. SMOKE includes a plume rise mechanism. WRF/Chem emissions are dispersed based on an hourly climotological wind fields for a specific region (New England?). Biogenic emissions are calculated online using the Gunther scheme. Biogenic emissions fields are hard to check, they are not output by the model.