The question of how calibration varies between various surfaces has been with us for decades. What we know is that it varies. We have yet got only fragmentary information. Part of the problem is that a surface may be described in words, but it will vary even within its category. Mike Sandford has shown that calibration can differ on an asphalt road even when the calibration paths are parallel and separated by only a meter or less.
I keep records of my calibrations. Each time I measure I pump the front tire to 100 psi, ride around the block, and calibrate. Below is how my calibrations came out. It includes postcalibrations as well as precalibrations.
The black points are those from my Kirkham Road 1000 feet, OH05028PR. The two red dots are from the Cleveland Ave 326.64 meters, OH06040PR. Kirkham Road is asphalt. Cleveland Avenue is laid out on the sidewalk parallel to the road.
We see a difference. Whether it is typical I cannot say, but it is all the personal data I have on this subject. Past issues of Measurement News have more information, but still nothing definitive.
The calibration file is available to anyone who asks. All past issues of Measurement News are available on CD. Contact me if interested.