🧑🎓
What exactly is the AQI number I see on weather apps? It says "Unhealthy" but what does that really mean?
🎓
Basically, the AQI is a single number that translates complex air pollution data into a simple health alert. It runs from 0 to 500. In practice, 0–50 is "Good," and anything over 300 is "Hazardous." Try moving the PM2.5 slider on this simulator above to 35 µg/m³. You'll see the AQI jump into the "Unhealthy for Sensitive Groups" category, which is a real trigger for people with asthma.
🧑🎓
Wait, really? So it's not just one pollutant? How do you get one number from PM2.5, ozone, and all the others?
🎓
Great question! The reported AQI is actually the *highest* value calculated from all the major pollutants. For instance, on a hot, sunny day with heavy traffic, ground-level ozone might be the "critical pollutant" driving the index. A common case is winter, where PM2.5 from wood stoves can dominate. Try the simulator: set Ozone (O3) to a high value like 120 ppb, but keep PM2.5 low. The overall AQI will be based on the ozone.
🧑🎓
So the health categories are based on the worst pollutant. But are all pollutants equally harmful? What about Carbon Monoxide (CO)?
🎓
They have different health impacts, which is why each has its own concentration-to-AQI formula. CO primarily affects oxygen delivery in the blood and is dangerous at high concentrations in enclosed spaces. In outdoor air, it's more of a tracer for traffic pollution. Notice in the simulator, you have to push the CO slider quite high (above 9 ppm) before it becomes the dominant pollutant and changes the overall AQI category. This reflects its different health threshold.
The core of the AQI is a piecewise linear function that converts the measured concentration of a pollutant (C) into an index value (I). The formula finds where C lies between two breakpoints.
$$I = \frac{I_{high}- I_{low}}{C_{high}- C_{low}}(C - C_{low}) + I_{low}$$
Where:
I = AQI value for the pollutant.
C = Measured pollutant concentration (e.g., µg/m³ for PM2.5).
Clow = Concentration breakpoint ≤ C.
Chigh = Concentration breakpoint ≥ C.
Ilow = AQI value at Clow.
Ihigh = AQI value at Chigh.
Each pollutant (PM2.5, O3, etc.) has a unique table of these breakpoints defined by environmental agencies.
The Overall AQI is not an average. It is determined by the single pollutant with the highest calculated index value.
$$AQI_{Overall}= \max(I_{PM2.5}, I_{PM10}, I_{O_3}, I_{NO_2}, I_{CO})$$
This "worst offender" principle drives public health alerts. The pollutant that achieves this maximum is called the "Critical Pollutant," and the health category (Good, Moderate, Unhealthy, etc.) is based solely on this maximum AQI value.
Common Misunderstandings and Points to Note
There are a few common pitfalls to watch out for when you start using this model. The first is the point that "the calculation results represent an 'average'". The Gaussian model outputs the long-term average concentration, assuming steady-state conditions (constant wind speed and emission). For instance, it's not well-suited for precisely reproducing instantaneous puffs of smoke or short-term phenomena where wind direction changes frequently. In practice, you'll often use these calculation results as a "rough estimate for worst-case scenarios" or for "long-term impact assessment".
The second point concerns parameter dependencies. Did you know that wind speed 'u' and emission rate 'Q' are not in a simple inverse relationship? Looking at the formula, concentration C is proportional to Q/u, but the diffusion widths σy and σz themselves also change with wind speed and atmospheric stability. Especially when wind speed is extremely low (e.g., below 0.5 m/s), diffusion isn't calculated well, and results tend to be overestimated. If you lower the wind speed in the tool to near zero, the concentration should become abnormally high. This isn't realistic, so in actual assessments, it's standard practice to set a lower limit value.
Finally, note that estimating the 'effective stack height H' is even more challenging than the simulation itself. H is the sum of the physical height plus the rise due to the exhaust gas's momentum and buoyancy. For example, for exhaust gas at 200°C with an exit velocity of 20 m/s from a 100m stack, H isn't simply 100m; depending on the calculation, it could be 150m or more. This tool requires you to input H directly, but in practice, you need to calculate this rise beforehand using separate formulas (like the "Holland formula").
Related Engineering Fields
While this Gaussian plume model is fundamental for air pollution prediction, its concepts are applied across various engineering fields. The first that comes to mind is Indoor Air Quality (IAQ) Simulation. A confined-space version of the Gaussian model is sometimes used to predict how hazardous gases or dust generated locally in factory workspaces or large offices will disperse. It's an important application that ties into ventilation efficiency evaluation.
Next, its connection to Risk Assessment and Safety Engineering is also deep. The foundation of "gas dispersion simulation," which predicts the range and concentration of a toxic gas cloud following a leak at a chemical plant, is precisely this model. In that case, a variation using an instantaneous puff (smoke cloud) model, rather than a continuous release, is employed.
Slightly different in nature, but it's mathematically similar to beam propagation analysis in Optics and Acoustical Engineering. The spread of a laser beam or the attenuation of sound from a point source is sometimes approximated by a Gaussian distribution. The concept of viewing a diffusion phenomenon as "spreading out from a center following a normal distribution" can be considered a common language across many fields dealing with 'propagation phenomena'.
For Further Learning
Once you're comfortable with this tool and want to learn more, I recommend following these three steps. First, Step 1: Thoroughly understand the model's 'assumptions'. The Gaussian model is built upon many assumptions like "flat terrain," "constant wind direction/speed," and "reflection only at the ground." Explore how the calculation needs to be modified if any of these break down. For example, considering the effects of building winds would call for "wind tunnel experiments" or "CFD (Computational Fluid Dynamics)".
Step 2: Explore the origin of the diffusion width tables (like the Pasquill-Gifford table). In the tool, these values are set automatically just by selecting a class, but they are determined by empirical formulas based on data from past field diffusion experiments. Strive to understand, by connecting the physical image with the data, why σ is large for Class A (unstable) and small for Class F (stable).
Finally, Step 3: Progress to non-steady-state, non-Gaussian models. For more complex and higher-accuracy predictions, you would use Lagrangian particle models or the CFD mentioned earlier. These methods can handle complex terrain and non-steady meteorological conditions, but at a vastly higher computational cost. If the Gaussian model is a "simple, rapid tool," these are positioned as "high-precision analysis tools." First, fully grasping the advantages and limitations of this Gaussian model is the best foundation for correctly using more advanced tools later on.