and the next sentence
"Nonetheless, if you have ready access and
you're primarily interested in changes in CdA at zero/low yaw angles,
testing on the track can be a very powerful tool (and it certainly
beats having to get up at 5 a.m. to drive to some isolated stretch of
road out in the country)."
so basically there's very little difference in data accuracy between testing outdoors vs. an indoor velodrome. but a velodrome does give you the benefit of repeatability, absence of traffic, etc. that was my point (albeit unstated) in suggesting a velodrome.