Hello,
I am a new CMAQ user interested in running CMAQv5.4 using EQUATES data. I’ve noticed some pre-processing that will be needed for updates in 5.4, specifically using SMOKE to create an ocean file and also to create BEIS_NORM_EMIS if using BEIS for biogenic emissions. However, I’m also interested in whether recent improvements in other schemes could also be used with EQUATES input. For example, using MEGAN (with BDSNP) for biogenic emissions, cb6r5_ae7 or CRACMM for an updated mechanism, and M3dry over STAGE for an updated deposition option.
Is there any data available for EQUATES inputs formatted for v5.4 or information on what changes may need to be made? I understand that some of these require additional pre-processing, I just want to be sure that what I would need to do that already exists from EQUATES or other sources.
Thank you kindly for any advice.
There is a dataverse entry that describes the metadata for EQUATES.
Please let us know if that provides the information that you need.
In general, the EQUATES emissions, meteorology, EPIC, and boundary condition input datasets can be used with CMAQv5.4.
If you wish to use MEGAN instead of BEIS to calculate biogenic emissions online, you generally would have to generate the necessary input files (see CMAQ Release Notes: Emissions Updates: Model of Emissions of Gases and Aerosols from Nature (MEGAN) Biogenic Emissions · USEPA/CMAQ Wiki · GitHub). However, as long as you are working on the same 12US1 modeling domain used in EQUATES, you can also utilize the MEGAN input files for that domain distributed as part of the 2018 modeling platform AWS bucket. (note that this directory also includes updated OCEAN files that include the DMS concentrations needed by CMAQv5.4).
Selecting the M3Dry vs. STAGE dry deposition option does not require any changes to model input files, only a change to that model configuration option in the bldit_cctm.csh script.
The model-ready EQUATES emission files were speciated for cb6r3_ae7_aq and work without modification for the cb6r5_ae7_aq mechanism available in CMAQv5.4. Switching to CRACMM would entail creating new model-ready emission files. This would require working off of the SMOKE-ready input files and reprocess everything through SMOKE, substituting CB6/AE7 speciation profiles and cross-reference files with their CRACMM counterparts. Doing so for the EQUATES period would likely be a major undertaking and require good familiarity with SMOKE emissions processing, particularly speciation.
4 Likes
Thank you both for the kind and helpful replies. We are indeed operating over the 12US1 domain. We plan to at least simulate the month of July. However, we may want to also simulate other years, such as 2017 or 2011 (we are focusing on agricultural emissions, and 2011 seems to be the baseline year for management data).
Would these 2018 files for MEGAN or the OCEAN files be valid for other years? For example, EQUATES appears to use static emission files for BEIS and its ocean emissions across all years. Of course, it seems like we could also calculate the necessary input files for the specific simulation year with the tools you’ve outlined.
The OCEAN files in the 2018 modeling platform AWS bucket are not year specific. The DMS and CHLO fields in these files represent climatological values, and you can obtain more information on the source of the climatology in the documentation sections of the dmschlo tool used to generate them.
To obtain more information on the underlying MEGAN data used to generate the CMAQ-ready input files for the 12US1 domain, you would probably have to look at (non-CMAQ) MEGAN documentation or contact the MEGAN developers, but I am quite certain that they are not specific to 2018.
Sorry to revive this thread from quite a while ago, but I’ve done more experimentation and it seems like the data is all quite usable for 2017 model runs. One issue is that I do not have an MSOILOUT for my 2-week spinup start date (June 15 or 16 off the top of my head) to use for a soil restart file (EQUATES provides BSOILOUT, of course). What are the caveats of beginning a 2-week spinup with or without a restart file, and what could be a way to generate one if needed?
I don’t have first-hand experience to which extent results might differ after the 2-week spinup period depending on whether you start the first day of the spin-up period with or without the information information contained in the previous day’s MSOILOUT file (or, for that matter, the previous day’s BSOILOUT file if using BEIS instead of MEGAN). My gut is that 2 weeks spin-up should be sufficient to wash out these specific initialization effects, but I don’t know.
To test whether two weeks spin-up are indeed sufficient, you could perform one simulation that starts on June 1 using the EQUATES CGRID and MEDIA_CONC restart files (but of course no MSOILOUT file), and a second simulation that starts on June 15 using the CGRID and MEDIA_CONC files from the first run but no MSOILOUT file. You could run both simulations through June 30 (using the MSOILOUT, CGRID, and MEDIA_CONC files from your runs for each day after the first day), and compare differences between both runs during the June 15 - 30 time period. These differences and their temporal evolution should give you an estimate of the persistence of the MSOILOUT initialization effect.
In general, when selecting an appropriate spin-up period, you would want to consider not only the persistence of MSOILOUT initialization effects but also the persistence of the CGRID and/or MEDIA_CONC initialization effects. For an earlier version of the model, we looked at those initialization effects in this technical note and found that 10-14 day spin up periods might not be sufficient in all situations.
2 Likes
I wanted to revisit this post once I had results, both for feedback and so that others could inform their decisions with spin-up periods without an MSOILOUT file. I’ve attached two gifs showing the difference between simulated PM2.5 concentrations and simulated NO emissions from MEGAN since this is one of our primary interests. I included 06/25-06/30 since I thought the last few days may show whether the model has converged. The 30-day spinup model diverges in simulated PM2.5 in the midwest on June 30, with a difference of up to 0.5 ug/m^3. Based on this, I’d want to use at least a 30-day spinup. One caveat to this result is that the two-week spinup had to be restarted from the 28th since the cluster job ran out of time, but the run script was configured to replicate how a simulation would normally continue. I’d appreciate any feedback and if concluding a 30-day spinup requirement is reasonable. Thanks for the help!
EDIT: I reviewed these results with my advisor, and we thought it was odd that conditions converged but then diverged after the 28th. Is this normal during a spinup rather than monotonic convergence?
