We saw the same type of thing in East et al. (ACP - Inferring and evaluating satellite-based constraints on NOx emissions estimates in air quality simulations). The description of rural vs urban is a good generalization, but there is certainly variation. The supplemental Figure 5 shows that our result was seasonally dependent.
In general, NOx lifetime is determined by the OH + NO2 reaction. This converts NOx to NOz – effectively a terminal product on short time scales. The availability of OH dependends on the propagation factor P_{OH}. I am going to simplify here and use RO2 to mean both HO2 and RO2:
P_{OH} = f_{OH+VOC} \times Y_{RO2/VOC} \times f_{RO2+NO}
- Y_{RO2/VOC} = depends mostly on the VOC mixture
- f_{OH+VOC} = fraction of OH reacting with VOC (i.e, not reacting with NO2)
- f_{RO2+NO} = fraction of RO2 oxidizing NO to NO2 (i.e., not reacting with another RO2)
So, you can see that the abundance of NOx effects the first and last parameter. All else being equal, more NO2 decreases the first parameter and more NO increases the second. The impact will depend on the environment.
Sillman and Kleinman showed that in NOx-limited environments (i.e, low-NOx) there are more RO2+RO2 than OH+NO2. Essentially, there isn’t enough NOx so the RO2 are reacting with themselves. If you add, NO then radical propagation will increase.
Put another way, in low-NOx conditions, there is an abundance of VOC and not much NOx. Adding more NOx decreases f_{OH+VOC} and increases f_{RO2+NO}. When the limiting factor is f_{RO2+NO}, emitting more NO leads to more RO2 + NO reactions, which produces more OH. More OH leads to more NOx loss via OH + NO2. In that case, you might add a unit of NOx via emission and lose some of what would have been in the air already.