What air dispersion modeling is for

Every page in this guide serves one question: will the air around this facility stay within the limits the law sets for it?

Air quality standards are written as concentrations in outdoor air — so many micrograms of fine particulate in a cubic meter, so many parts per billion of nitrogen dioxide. A facility, though, is permitted on its emissions: pounds per hour out of a stack, tons per year from a process. Nothing in an emission rate tells you what a person standing at the property line will breathe. The distance between those two units — grams per second leaving a stack, and micrograms per cubic meter arriving at a receptor — is the entire subject of this guide.

A dispersion model closes that distance. It takes a description of the emissions, a year-by-year record of the weather, the shape of the ground, and the buildings in the way, and it computes a concentration at every location you ask about, for every hour of the record. AERMOD, the model this guide covers, is the one U.S. regulators expect to see: the Guideline on Air Quality Models — the regulation at 40 CFR Part 51, Appendix W — lists it among the preferred models, meaning that if you use it as prescribed, you need not argue for the model itself:

No further evaluation of a preferred model is required for a particular application if the EPA requirements for regulatory use specified for the model in the Guideline are followed.

Source · 40 CFR 51 App. W §3.1.1(e).

That sentence is worth pausing on, because it defines the job. The modeling is not a research exercise where any defensible method wins. It is a prescribed procedure, and most of the skill lies in knowing what the procedure requires, what it leaves to judgment, and how to document the difference.

The question a demonstration answers

Regulatory modeling almost always asks one of three questions:

Does this facility, by itself, matter enough to look at further? — compared against Significant Impact Levels.
Is the air, with this facility plus everything else already there, within the standard? — the cumulative demonstration against the National Ambient Air Quality Standards (NAAQS).
Does this facility consume more than its share of the air quality still available in a clean area? — the Prevention of Significant Deterioration (PSD) increment analysis.

All three are answered by comparing a number the model produced with a number the regulation fixed. Getting the comparison right requires the model to produce the number in the form the standard defines — not the peak hour, but (depending on the pollutant) the eighth-highest daily maximum averaged over several years, or the annual mean of the worst individual year. Part I of this guide is devoted to that form, and the last part of the guide covers where the model reports it.

Why one model needs five programs

AERMOD does not read weather files, terrain files, or building blueprints. It reads processed inputs — and each kind of processing is a separate EPA program with its own inputs, options, and output files. This is the single biggest source of confusion for people new to the field: "running AERMOD" is really running a small pipeline.

The AERMOD system: raw data feeds five preprocessors, whose products feed AERMOD along with the source, receptor and option inputs you supply

Read the diagram left to right. Raw data on the left, preprocessors in the middle, AERMOD on the right, and outputs beyond it:

AERSURFACE turns land cover around the weather station into the three surface characteristics AERMET needs — how reflective the ground is, how it splits the sun's energy between heating air and evaporating water, and how rough it is.
AERMINUTE reads the archived 1-minute and 2-minute wind records from automated airport stations and produces better hourly winds, recovering hours that the standard observation would have discarded as calm.
AERMET is the meteorological processor proper: surface observations, upper-air soundings (a balloon profile of the atmosphere, launched twice a day), and the products above go in; two files come out — a surface file of hourly boundary-layer parameters and a profile file of winds and temperatures aloft.
AERMAP reads digital elevation data and assigns each receptor and each source its terrain elevation, plus the "hill height scale" AERMOD uses to decide how a plume behaves in terrain.
BPIPPRM reads building corners and heights and computes, for every wind direction, the effective building dimensions that control downwash — the aerodynamic pull of a plume into a building's wake.

The arrows are not decorative: each preprocessor's choices propagate into every concentration AERMOD reports. A surface-roughness decision made in AERSURFACE changes turbulence in AERMET, which changes dilution in AERMOD, which changes the design value you compare against the standard. That chain is why this guide gives each preprocessor its own pages rather than treating them as file-format details.

How this guide is arranged

The parts follow the order the work actually happens in:

Air quality standards — the targets: what the standards are, what form each design value takes, and how background and increments enter.
Meteorology — the biggest input by far: representativeness, AERMET, AERSURFACE, AERMINUTE, onsite towers and prognostic data, and the options each tool offers.
The model — how AERMOD itself works: plume rise, the boundary layer, terrain, the thirteen source types, nitrogen dioxide chemistry, deposition.
Downwash — buildings, wakes, and Good Engineering Practice stack height.
Outputs — reading what came back: the ranked tables, every output file AERMOD can write, and the source-contribution analysis (MAXDCONT) that shows who caused a high value.
Regulatory review — the protocol, emission inputs, receptor placement, and the refinement ladder you climb when a run does not pass on the first try.

That last item deserves a word, because it is what the job usually is. A first run that fails does not mean the facility cannot be permitted; it usually means the analysis is still carrying screening-level conservatism that the data can now replace. Knowing which assumptions may be refined, in what order, and with what documentation is the craft this guide is trying to teach.

In PlumeSmart

The platform runs the entire pipeline in the diagram for you — surface characteristics, hourly winds, meteorology, terrain, and building dimensions are prepared and passed to AERMOD in the right order, with the choices at each stage visible and recorded. What you keep is the judgment; what you skip is the file plumbing.