My intuition suggests that rainwater that has spent less time on the roof won't be worse than rainwater that had spent more time there, but this intuition could

My intuition suggests that rainwater that has spent less time on the roof won't be worse than rainwater that had spent more time there, but this intuition could certainly be wrong. Regarding the landscape catchment, if it's not in-situ stone that is laid bare and chiseled into shape it might simply be an economy of scale thing, use roofing commodity instead of reinventing the wheel. And if it's in-situ limestone, a major contributor to the decision for stepped/terraced might still be ease of maintenance access. The chemical qualities seem more like a convenient coincidence to me. Perhaps an enabling quality in absence of which they'd have long switched to desalination, but even that would not make it a causal "why". Particularly the part about UV working better with the reflections white making photons pass twice seems doubtful to me, because nighttime rain is probably not entirely unheard of (yeah, I might be wrong about this)