I've forgotten exactly, but water flow is supposed to be a combination of the expansion and compression of the ground reducing and enhancing the water flow temporarily, and the shaking releasing some water. Scientists used to think bedrock was involved, but since the Loma Prieta earthquake and work by Stuart Rozjacker (sp?) at Duke, it looks like the water only comes from the sedimentary rocks.

At least that's the current thinking. You might check work by Michael Manga at UCBerkeley.