This is highly off-topic. I'm hoping that some of the map-geeks out there will have some ideas for me.
A friend of mine at U.C. Berkeley is working on a research project auditing road construction quality in India and wants to chart road deterioration using aerial/satellite imagery and/or LIDAR data.
The idea is to use visual signals that might indicate potholes, washboarding, or other deterioration.
Does anybody know whether this has been done or have ideas about how it could be done?
A few lidar ideas:
return intensity to determine surface material, asphalt vs dirt
linear ruts are probably detectable and measurable in an unsharp image
washboarding would require a measure of surface smoothness, which would need a very tight post spacing, same with potholes.
A few thoughts:
Pavement engineers use laser profilometery to measure the roughness and some major pavement defects. I think that is lidar-like, but I think it requires collecting the data by driving the pavement with the laser collecting the data. It isn't making use of existing lidar or air photos. If you actually have access to the roads, you can collect ride data to measure smoothness. There's a relationship between smoothness and pavement quality. Some contracts may require a specified level of smoothness.
Depending on the quality of the air photos and/or lidar intensity images, you can spot some common pavement problems. In particular utility cuts and pothole patches show up in photos. You can also see some large cracks in some photos.
You might be able to get rough estimates of the age of a surface based on the color. Asphalt and concrete surface both change color relatively quickly as they age.
You should be able to use air photos and lidar intensity to measure surface areas. If you're auditing construction, the more things you can do to verify quantities that you pay for the better off you are. Construction contracts often pay by the amount of material used (usually tons of asphalt or concrete). Contractors have been known to overcharge or overuse materials. You have to have a good idea of the system in place for payments to really figure out the various ways to test for overcharges. But, I'm guessing that UC Berkeley isn't looking for fraud on these roads.
You could use topography from lidar data to figure out where water flows. That might help identify areas vulnerable to pavement problems. If you find a big watershed that feeds runoff to a little pipe under a road, you're going to see problems with that road.
I don't know what sort of administrative records they have, but in a project I worked on here, I was able to project pavement problems based on the specification in place at the time of construction. But that took access to the administrative records and I had data from pavement inspections (where inspectors look for various defects, most of which are easily visible at the pavement level but not in a photo or lidar).
Depending on what they mean by "auditing" you could probably project pavement performance by looking at the contracting management practices and internal controls. If you don't have good practices and internal controls, you're likely to get bad pavement.
Thanks folks. That's useful insight.