If the shooter(s) really do have the cops totally confounded there's one approach that might bear fruit. If they have reason to believe that he/they are using the main highways to either enter or egress from the target areas they could install traffic cams at the on/off ramps of the exits. I'm pretty sure there are some wireless solutions out there so running cable (time consuming and expensive) could be avoided. The cameras connect to an always monitored central facility. Tape record all camera activity and archive. If a shooting takes place review all camera feeds for suspect vehicles. You may not get success the first time but eventually unless the perp(s) are switching vehicles all the time you're going to see the same one or two vehicles consistantly either entering or leaving the area via the freeway exits in close relation to the time of the attacks. Proceed to apprehend as possible. Might seem like grabbing at straws but unless they have more to go on than they're telling us (and I suspect they do) this could yield benefits. Pick some key traffic flow points and hope for the best.
Only problem with using cameras is they don't currently exist in this area if I'm interpreting people's posts correctly. That means they have to be installed. Hard to do with no one noticing, plus the cameras may be visible after installation. In this case the perp simply changes their pattern or moves outside of the coverage area.
I wish I had a viable answer to this problem. I suspect we all do.
There are tons of traffic and red light cams in the area already, though I reckon the red light cameras wouldn't have much application in this situation. The cameras are, by the way, visible - but they don't really look out of place. I would suspect that resources are an issue here - whither the money to install all these cameras? LE resources, financial and otherwise, are already stretched quite thin, to the point that locals are asking for federal financial assistance. Also, as stated previously, putting cameras everywhere is a tall order for the simple reason than this is Roadland. It's impractical to consider covering every stretch of likely road from Maryland down to Richmond - too many roads, too many miles to cover.