After a few months of anticipation, this past Wednesday (November 20) the final rule revising the Guideline on Air Quality Models was signed by the EPA Administrator. As part of that, EPA updated AERMOD with a few changes (the only one likely to affect most of you reading this is a new NOx to NO2 conversion technique available as a regulatory option, the Generic Reaction Set Method (GRSM)) and published a final version of its Background Concentration Guidance[1].
Given the stringency of the new Annual PM2.5 NAAQS there’s more attention being paid to background concentrations than ever before. To that end, it’s the new Background Concentration Guidance that I was most interested in with this update, and that’s what I’ll speak to in this blog—starting with a summary of the updates to the final guidance and concluding with my commentary on what this does and doesn’t mean for the modeling community.
Summary of changes
A little over a year ago EPA released a draft version of this guidance document. After receiving comments from the modeling community, EPA made very few changes for the final version, with the one meaningful one being that they added two appendices that provide “real-world” examples of going through the mechanics of selecting both an appropriate monitor for a background concentration and for determining which sources to explicitly model.
A quick sidebar—you may be wondering why, if this is background concentration guidance, are we talking about which sources to explicitly model. In this context EPA regards the “background concentration” as a combination of modeled impacts from nearby/offsite sources and a value from a monitor. So, this guidance speaks to how to deal with both.
EPA provides two hypothetical scenarios for addressing background concentrations, one for when the source being permitted is isolated and one for when the source being permitted is amongst other sources. Going through these examples you can get some insight into what EPA thinks is important when it comes to identifying monitors for a background concentration and offsite sources to model. For instance:
Consider the scale/purpose of the monitor—is it regional, neighborhood, etc.?
Look at predominant winds in the area—is a monitor downwind of a source?
Consider the averaging period you’re modeling—if you’re looking at a 1-hr averaging period, assuming a source’s emissions impact a monitor 50 km away probably isn’t reasonable (because it’s unlikely emissions would be transported that far in an hour)
Consider the total emissions in the vicinity of a monitor—how do they compare to total emissions in the modeling domain?
Consider the expected footprint of air pollution impacts from an offsite source being considered for inclusion in the modeled inventory and how it might overlap with other predicted impacts—if it has tall stacks it will have a larger footprint than if it has short stacks
Commentary
While these approaches make sense from an academic perspective, I’m not sure how useful they are in practice. If I’m putting together a modeling analysis, I’m probably not going to take the time to go through the exercise of determining whether a nearby source should or shouldn’t be included in my inventory—it’s easier and faster just to include it and worry about it later if you run into problems. If it does end up causing problems then sure, I’ll go through that exercise. But needing to do that is fairly rare in the modeling world.
When it comes to Annual PM2.5 NAAQS modeling, usually what is now the driver in terms of compliance is the background concentration—the value you get from a monitor that gets added to every predicted concentration at every receptor (yes, there can be temporal/spatial variations, but my fundamental point remains). Removing some sources from the modeled inventory I think, in most situations, won’t really move the proverbial needle when it comes to NAAQS compliance. What I think is needed is flexibility to allow for technically-justified and common sense approaches to using a background concentration lower than what the monitor data show. As this guidance touches on, there are occasions when a monitor is impacted by sources far from the modeling domain, and I think that should be accounted for in a modeling analysis. In my opinion, it’s unfortunate this guidance doesn’t explore that possibility.
I’ll close with what I think is a very important quote taken directly from this guidance:
“Appropriately accounting for all source contributions is an inherently discretionary exercise with use of best professional judgment in determining a representative background concentration and identifying nearby sources that need to be explicitly modeled.”
I hope that in the future we can lean on the fact that determining background concentrations is discretionary and, now more than ever, best professional judgment is key.
[1] https://www.epa.gov/system/files/documents/2024-11/guidance-on-developing-background-concentrations-for-use-in-modeling-demonstrations.pdf