Because that's what happens when contaminants are released from a point source into the atmosphere (or any fluid for that matter). They disperse due to eddies, pressure/temperature gradients, or other causes of turbulent diffusion.
It's just a larger scale version of what happens if you watch smoke from a cigarette, etc. In the immediate area there will be large differences in concentration. So if you're say 10 feet from a smoker moving just a few feet in the right direction can make a huge difference. But if you're a mile a way then you would never even notice and even if you had instruments to measure it, moving them a few feet left or right would likely make no difference as whatever detectable contaminants remain would be pretty evenly dispersed relative to the local area.
Even if there are differences in measured values you'd have to understand the properties of your sensor well to validate those very small differences are true vs variations sensor to sensor, sampling methods, or other repeatability issues.
Which is not to say this is necessarily a bad idea or it shouldn't be done (seems cool to me), but simply to state my opinion that I wouldn't expect to see much difference and even if you did validate a true difference it would probably be so small as to not impact a rational decision.
But who knows, maybe you find something interesting. Just seems like a smarter use of resources to try to look for a problem on a small scale before trying to quantify it on a large one. E.g. Go to 10 places on Oahu and see if there is any difference.
OTOH, there are probably lots of other contaminants that one would reasonably expect to vary around the island at meaningful levels. E.g. Pollution from HECO, traffic, other industry point sources.
- bri