Well, when going to Mars it typically works like this:
Launch services deliver the bus to low-earth-orbit (LEO) for a once-around. Then the spacecraft (s/c) is launched on a course to Mars.
"Cruise" takes over at his point. They rely heavily the Nav[igation] team. Of course these guys have a pretty good idea what the gravity field is between here and Mars, but they still need telemetry-correction-maneuvers (TCMs). There are 3 scheduled TCM outbound from Earth (1-3), and 3 more (4-6) on the Mars end.
Nav uses delta-DORS to figure out where the s/c is. This is like the old LORAN system used by ships, except instead of using manmade radio transmitters placed around the globe to triangulate a position, it uses natural radio transmitters: quasars powered by black holes (this is the only know engineering application of black holes at this time). The signals' times are compared with super accurate Earth signals to compute the s/c position in the JPL barycentric coordinates system (BCS). Mars is usually a direct transfer orbit, though it can be done with a Venus flyby. Flyby's, esp. for the gas giants, require post-Newtonian or post-post-Newtonian corrections for relativistic effects.
TCM's hone in on the correct trajectory, usually with only 2--the 3rd is for margin.
Nav tracks the s/c's position during cruise. Outgassing, solar radiation pressure, and asymmetric thermal emissions are bigger effects than gravity-field uncertainty.
A few days before entry, TCM 4 tunes up the trajectory, with TCM 6 scheduled for 4 hours before entry. E -4h is the last time anyone commands the s/c, and it is a big deal to do anything out of the ordinary (like a major TCM, or a flight software new-parameter up-load).
At this point Nav really knows where the s/c is, and the deliver it to an imaginary plane called the "b-plane" for some reason. It is about 1 square km, and all they tell Entry-Descent-Landing (EDL) is that it will go through that plane somewhere within a 1 second time window.
At this point, the greatest uncertainty is "Where is Mars"--the planet's position is far less well known than the s/c's. Moreover, inertial guidance cannot track the 10g aerobraking and violent supersonic parachute deploy to better than a fraction of a km. Hence, on board terminal descent sensors (radar, maybe lidar someday), figure out where the surface is and how fast it is moving. Touch down velocity requirements can be in the cm/s range, so that the rotational speed of the surface over a 100 km landing ellipse has a much large spread than that.
In summary, gravity-field uncertainty is far from the top of the list of concerns.