https://en.m.wikipedia.org/wiki/File:1960-07-21_First_Polari...
AFAIK submarine launched ballistic missiles aren't launched that far below the water surface and even a traditional periscope can reach 15+m
It is speculated that you could keep navigational errors bounded with gravimetric measurements. Essentially if you have a detailed pre-surveyed map of the underwater landscape, and high precision gravity measurements you can match your measurements to the undulations of the expected gravity field. Somewhat similar, at least on the math level, to how TERCOM works with cruise missiles.
Very likely they would use the best motion isolation money can buy. Also the submarine is quite slow (especially at navigation speeds) so the terrain changes under it slowly. Water currents and the motion of people and machinary happens at much higher rate, so you can probably try to eliminate the quick changes and keep the slower changes with filtering and statistics.
Most gravity-gradient measurements are differential. Common-mode accelerations can be substantially-rejected.
Also, if you're in the precision-measurement game, navigationally-relevant signals can be much larger than the precision of modern instruments. People in the prospecting business sometimes fly gravity-gradiometers in airplanes.
I suppose you can also do it with the moon. On a dark, cloudy night though? You could risk using the sonar since the launch is going to make a ruckus anyway. Compare measures against seafloor maps.
Pre GPS navigation methods include sextants, celestial navigation, and LORAN (navigation via bearings to multiple radio towers).
Sub launch ballistic missiles actually have an automated version of celestial navigation: there's a star spotting camera that compares what it sees to a reference database of the couple hundred brightest stars. So even if the missile starts out with a lot of ambiguity about where it's launched from, by the time it gets above the clouds it can correct this.
For routine operations, modern subs do carry a towed buoy which can be used to exchange radio messages and update GPS position while the sub remains deep underwater. This carries some risk of detection, but less than coming up to periscope depth.
Ballistic missile submarines need very low drift inertial navigation, because they stay submerged for a long time without outside references. The submarines provides a starting position for missiles, so missile error is added to submarine error. Size and weight aren't major issues, so they can use much larger gyros.
There's continuing interest in inertial navigation. Projects show up on DARPA.[1] The military assumes that GPS will be jammed much of the time in wartime, so they want inertial backup.
[1] https://www.darpa.mil/program/micro-technology-for-positioni...