Ozone source apportionment

I want to run the ozone source apportionment tool to evaluate the impact of emissions from Mexico on ozone in Arizona. My only issue is I can’t decide whether the 12 km grid domain or the 36 km domain is appropriate for this analysis. The 12 km domain only covers the northern part of Mexico while the 36 km domain covers the entire country of Mexico. However, if I am not wrong, CMAQ performs better on the 12 km domain compared to the 36 km domain. Is there any other criteria I missed that would help me decide between these two domains?

Ideally, you should run a nested 36/12 system, with the 36 providing the BC’s for the 12, and using a 900-sec to 1200-sec time step for the met-files, in order to model temporal variability of the transport and the mixing behavior (that last changes quite rapidly during the after-sunrise transition).

Or better, just run the whole thing in a 36/12 nested WRF+CMAQ or WRF-Chem…

@cjcoats can you explain more how the nested simulation addresses the issue of the 12km domain not covering the entire country of Mexico?

The 36km domain in the nest system should cover the entirety of Mexico. It then provides boundary condition data for the 12km domain, that gives you (somewhat coarser) inflowing information about the air quality coming into the 12km domain from the portion of the 36km domain outside the 12km.

You also need to keep in mind that the CMAQ ISAM tool currently does not support nesting, i.e. it does not support passing source tags from the 36 km ISAM simulation to the 12 km ISAM simulation. The 12 km ISAM simulation can attribute ozone to emission sources within that domain as well as to initial and boundary conditions. The boundary conditions would include contributions from all sources outside the 12 km domain which includes Mexican sources not covered by the 12 km domain as well as all other international sources and large scale background ozone.

You could run ISAM for both the 36 km and 12 km domains separately and use the 36 km ISAM results in which you could tag a complete set of Mexican sources to provide context for the 12 km ISAM results in which the effect of Mexican sources would be split between those that can be explicitly tagged because they are located within the 12 km domain and those that are implicitly included in the boundary conditions derived from the 36 km simulation.

As you noted, it’s a trade-off between being able to explicitly attribute concentrations to the full set of Mexican sources which requires a 36 km ISAM run and likely better model performance resulting from better-resolved meteorological and emission fields in the 12 km setup.

@hogrefe.christian that’s a good point. I will look into your suggested method as well as CAMx ozone source apportionment since it supports nesting. Thank you for your feedback.