From my various reading, it appears that in routing a railroad over long distances was first a reconnaissance trip by horse/and or wagon. This will provide a general idea of practical routes and unfeasible passages. Nowadays, like the Santa Fe did in the 1960s, this can be accomplished by studying USGS maps.
The next step is then to send out one or more survey crews to find a more precise route. This may just start at "point A" and move toward the destination. If part of the route is through a mountain pass a crew may start at the top and work their way down toward the flatland. This survey many be enough to construct the railroad or in places a more detail final survey would be needed.
The C&C over Montgomery Pass is an interesting example. There had been wagon road over the top and down the two sides. The initial surveyor for the railroad could probably be called a flatland surveyor -- he looked over the pass and said yes we can do it and it will take some bridges to do it. But when they got to actually locating the route he was out of his element. So the company brought in another surveyor to assist who had mountain experience. To achieve the final alignment it sounds like several lines were run to find the best locations and the final alignment was a combination of several of these survey lines. The approach from the northeast was obtained by back and forth across the ascending terrain. The descent toward the Nevada-California state line involved some serious surveying as the railroad constructed fills that were as high as 30-35 (or more) feet; and elsewhere a cut that was just as deep. And a tunnel -- on a grade and with an interior curve! they managed to hold the maximum grade to 2.5% with a few short stretches of 2.6%. But the steepest grades were actually in some relatively unimposing canyons located elsewhere on the railroad.
Brian Norden