Concept art of lunar bases tends to show spherical or cylindrical structures, but they suffer from one problem: radiation. (Both of the gamma / particle and heat kinds). The lunar environment has lots of solar and cosmic radiation. Nights also last for two weeks, during which a badly insulated thing will freeze.
If you bury your moon base under a layer of regolith, you can avoid both of these problems. You don’t have to bring heavy shielded modules from earth, or heat during the night with nuclear batteries. Regolith is thermally well insulating.
NASA seems to catch onto this a little in some clearly low budget “alternative configuration” posted at Nasawatch, but it only goes halfway, putting some shallow “berms” around cylindrical structures, for shielding.
In reality, lunar bases (if crews are to spend many lunar nights there) would probably be completely buried.
Burying might actually be easy: a small automated/remotely controlled snow blower style rover vehicle might be able to do it slowly with the help of just solar power. Since there is no air, tiny amounts of regolith can be thrown large distances. A thin wheel with whiskers spinning rapidly would throw the sand to the wanted direction. A lunar day is 336 Earth hours. Even if the “regolith lobber” robot can not survive the night and is expendable, it could manage to move significant amounts of regolith. A sub-MER size rover with large solar cells could throw perhaps 20 grams of regolith per second, or 72 kg in an hour. That’s 7 tons if there is 100 hours of efficient sand throwing time.
Say, landing at 50 hours from dawn, setting up 50 hours (survey area, lower rover from lander, unfold solar cells etc), operating for 150 hours (which includes maneuvering 50 hours and sand lobbing 100 hours), and finally stopping at a low sun angle 50 hours before dusk.
Such concepts are probably unlikely to work in an atmosphere, though I would be happy if proved wrong. The 2009 lunar regolith excavation challenge is coming up, after all…