When a rover receives observations from a base station (e.g. through NTRIP from a CORS network) it can perform RTK to reduce ambiguities in matching observations to get a more accurate fix.
What happens when the set of visible satellites for the rover have little overlap with the base station? Will only the common observations be used? Or will other observations be used, just with more ambiguity?
For example, how would the performance of an L1/L5 capable rover receiving L1/L2 observations from an NTRIP server compare to the same rover with only L1 observations enabled, receiving the same NTRIP observations as the other?