You should probably first consider whether your sampling is going to be destructive or non-destructive. That will narrow down the options greatly. Let us know and im confident we can come up with an appropriate solution.
Well, taking into account that I would work on more sites, I would probably go for a non-destructive sampling or, maybe, for a protocol that would take the destructive part as lower as possible (e.g. for getting info on the ratio between surface and biomass in a given subset of species or growth forms).
I learned from a French colleague (F. Baret) how the biomass of mosses on a forest floor can be estimated. My colleague Baret and a colleague of his (Marie Weiss) introduced hemispheric photography to estimate, among other variables in a forest, LAI. Fred did the same with mosses on a forest floor in Germany once (for fun). But his approach works damn good. The LAI of the moss layer was about 2. Not negligible. You can obtain more info on the hemispheric technique and the software needed to estimate gap fraction or LAI, using the CAN-EYE software developed and still improved by Marie Weiss.
Shortly. Once you have your LAI estimate of the mosses, you can convert that to biomass by estimating specific LAI or the LAI per unit weight, by some destructive sampling of the moss species you are looking at with your camera. From LAI and specific LAI, you can get your biomass. What do you need? A hemispheric (objective) camera (Nikon), an accurate weighing device, and the CAN-EYE software.
For epiphytic species on trees, when they are flat you could try the same approach, but when they are growing in different kinds of shapes, then this turns into a completely different business, for which I don't have a solution at hand yet.