Measuring Crr.
 

I thought this was kinda nifty. If you want to approximately measure your rolling resistance coefficient over a portion of road, you need to know the car's loaded weight, air density, CdA, and the length and change in elevation of your route. You need a timer in seconds and speedo. Obviously this works better for unmodified cars, and assumes the CdA provided is accurate, but anyway...

Pick a route using google maps/gmaps pedometer, measure the length of the route and the change in elevation. Because gmaps pedometer measures the distance as the crow flies, you may need to adjust the length of the route if the change in elevation is significant. When driving, hit that initial point of that route at some speed (preferably low) in neutral, measure the time it takes to get to the second point and the speed once you hit it, do a few runs. We now have an initial velocity, final velocity, and time. Double check the time with the distance to check if your speedo is accurate (mine was off) and figure your average speed, Va. We'll use this.

Va(W(h/l)-.5(ro)Va^2(CdA)-W(Crr)-Fs)=(.5m(Vf^2-Vi^2)+Wh)/t

Where Va, Vi, and Vf and the average, initial, and final speeds, t is time, m is mass, W is weight, Fs is some stray force for energy dissapation in the suspension or whatnot, ro is the air density at the altitude/temp, h is the change in elevation, l is the length of your route, and CdA is the reference air multiplied by the drag coefficient, which may be found here for stock vehicles, and that's about it. I found my Crr to be ~.014 assuming a stray force of 50N, which seemed a bit high, until I realized the road I tested on was far from what organizations such as green seal would use. Ideally, if you can pull the suspension beam off of a car and have an inclined smooth road with no traffic, you can time what it takes to go from zero to the end of your route to get a much more accurate figure for Crr, maybe vary the tire pressure to get a curve for different pressures/temperatures...

While the Crr may not be especially accurate, it allows you to quantify a proportion of Crr and test other changes wrt a baseline. I.e. even if I can't state what the specific Crr is, I can state that dropping my tire pressure from 50lbs to 35lbs results in some percentage change in Crr, which is very useful. The same method could probably be used at high speed to quantify changes in drag coefficient.