So, when you lay tracks what do you trust more, LIDAR or google maps? Sometimes there is slight offset, especially on embankments. Do you stray from satellite imagery and follow embankments or follow the imagery and adjust embankments?
Lidar data should be more accurate. Objectively straight track beds are sometimes rendered curved on aerial images, probably due to incorrect parallax correction, which is to be expected if the correction has been made from standard, low-res DEM, which is likely.
What about gradients? There might be big discrepancy between documented gradient and what LIDAR landscape requires. I do understand that track ribbon can affect this too, in some cases, however no to the extent I see. I end up having tracks floating above surface at the bottom of a slope if I follow documented gradient or I have steeper gradient if I follow the landscape
I would say, use all the sources including theoretical knowledge to get the best estimate and if there's a discrepancy or something you wouldn't expect, ask why it should or shouldn't be a thing. It isn't just about picking a source and sticking to it. For example if you have a very long straight and the track suddenly don't align in a station, you should realise that there is a slight wobble and that the tracks are more spaced out within the station or that the track has a small hump when crossing an underpass. After all, high-quality track laying goes a long way because it's one of the things that you actually get to feel as a player instead of just seeing it. Last thing, isn't gradient data that you have just calculated elevation difference over distance? In real world, you have transitions between grades and that makes them steeper in the middle. You should really cross-check the data to see what's plausible.
I try to follow LIDAR as much as possible, as it means less altering of the terrain. Also as I am working on a disused route, LIDAR is the only option as traces of the trackbed are only visible in LIDAR.
