One HN-oriented analogy is conditioning heat out of a data center. If you wait till the heat coming off your servers mixes uniformly throughout the entire room, you have to process a lot of air. If on the other hand you catch the air just as it comes off the servers, you have to process less air in order to remove the same energy.
Pushing the analogy further-- at normal temperatures (for data centers and for the ultimate place they reject heat to, the outdoors), processing a small amount of air that starts at a high temperature can be done more cheaply than a large amount of air that starts very close to the temperature you want to maintain. By analogy, it's probably true that scrubbing a fixed amount of NOx is easier when you start at a high concentration, than when you have a greater volume at lower concentration.
Further, think about the ratio of boundary layer (or even at a lower height) air volume to the surface area of roads. Plus, there is no suction happening here. It's literally just the air touching the surface of the road. I would love to see some numbers on the efficiency of this process, but my gut tells me it's pretty poor.
You are also absolutely correct that treating at the tailpipe is even better (for exactly the reasons my analogy makes clear).
That said, I will defend my analogy. My point was that dealing with NOx at the road surface will likely be more effective than dealing with it more generally in the atmosphere-- in part because the effective volumes aren't as huge as you might think (see the comment I was responding to), and in part because you're dealing with the NOx at higher concentrations.
I'm sure you're right that NOx disperses rapidly, but I'd bet that concentrations still are demonstrably higher at the roadway surface than, say, a block away. For example, this study found a factor of 5 difference between NOx at the side of a busy intersection, versus that some distance away: http://www.atmos-chem-phys.net/10/2745/2010/acp-10-2745-2010...
Second, I bet the boundary layer isn't all that thick at the road surface. After all, by definition you have cars passing along, stirring up a lot of turbulence. (Besides, even in quiescent conditions, you still get diffusive transport through the boundary layer, due to concentration differences-- and destroying NOx at the surface will set that up for you.)
As far as whether it is effective or not... well, it's probably more effective than not doing anything (additional costs and construction techniques need to be considered). Overall, I think absorbing pollution near the vehicles that produce it is a good idea. The raw amount of paved area in the world makes it an attractive target for any 'green' development (I've heard of projects trying to put solar panels into roads).
That doesn't make sense. Resources are finite. If something is not effective, then you shouldn't be wasting time and money on it, definitely not public money.
The raw amount of paved area in the world...
Have you thought of the raw amount of air in the atmosphere?
... makes it an attractive target for any 'green' development
You hit on a good problem here. "Solutions" like this one end up being nothing but certificates for companies and governments to show that they are doing "sustainable green development (TM)".