The whole Starship system is built around two fundamental ideas:
- Robotic refueling and
- A large fleet of cheap, aggressively simple, mass-produced, mostly identical vehicles driving cost down. Even Super Heavy is essentially just a strengthened, lengthened Starship with more engines and no nosecone.
So, yes, the Starship will need to bring enough fuel to both land (no aerobraking on the Moon!) and launch. But, remember that the Artemis architecture only has the lunar landers going from the Lunar Gateway to the surface of the Moon and back. They don't have to get there all the way from Earth (except for the first time, of course), and they never have to get back to Earth. They only go up and down from lunar orbit to lunar surface and back.
So, the Lunar Starship (Moonship?) only needs enough to fuel to de-orbit and land, and then launch and get back to orbit. And it does all of this in the Moon's much lower gravity. And while the lack of aerobraking means spending more fuel on landing, it also means spending less fuel due to no air resistance on launch.
If they have to refuel every time before they go back down, then so be it. Refueling is what Starship does. Remember, for a Mars mission, they already expect to refuel 5–10 times in LEO, so clearly, refueling is something they're not scared of.
Using hydrogen makes no sense because of fundamental idea #2. A vehicle like Starship is essentially just two things: a giant tank with engines attached at the bottom. For hydrogen, you obviously need completely new, completely different engines. And you need completely different tanks. Not only do the sizes of the tanks need to be different because the volume rations are going to be different for hydrolox vs. methalox, but probably also the materials need to be different, since liquid hydrogen is much colder than liquid oxygen or methane.
So, if a vehicle essentially consists of tanks and engines, and you need to design, develop, manufacture, test, qualify, human-rate completely new engines and completely new tanks from scratch, the answer to "why don't they just make a methalox and a hydrolox Starship" should be clear: it simply wouldn't be a Starship, it would be a new vehicle with the same amount of development resources, costs, engineers, and the same development time as the existing Starship.
This is the exact opposite of "a large fleet of mass-produced, mostly identical vehicles driving cost down".
Now, you might say: "Wait, isn't the Lunar Starship already very different?" Well, not really. It will have no aerodynamic control surfaces and no heat shields, but those are externally mounted "appendages". You can just … not mount them. And it will be painted white, but that is literally just a paint job.
It will have the nosecone with the airlock(s) and the crane in it, but the nosecone is a fairly simple piece that is mostly separate. (They will have to build different nosecones anyway, even if they don't win the HLS contract, since those are more or less the only difference between Cargo Starship, Crew Starship, and Tanker Starship anyway) And it will have the additional landing thrusters, but those are going to be "Raptor-derived", so they will still share some commonality with the Raptor as well as with the hot-gas thrusters, which will also be "Raptor-derived".
So, the Lunar Starship is more a straightforward modification of the Crew Starship than a completely new vehicle.
So, in the end there are two major reasons for using liquid methane:
- Designing, developing, and building a one-off completely different vehicle just for HLS makes no economic sense.
- Refueling is in Starship's DNA, so who cares if you have to refuel every time?