There may be one speculative idea that rather suffers from the coincidence problem that might hide the water:

If the dust grains contain rust and hydrogen peroxide and get lofted high enough into the upper atmosphere to benefit from the proper light frequencies from the sun, the hydrogen peroxide can serve as a catalyst with the photoelectric energy to reduce the rust to a less oxidized state. If this dust with the hydrogen peroxide still present happens upon a cloud of methane, then the hydrogen peroxide will oxidise the methane and the available water can assist in returning the reduced iron oxide to a more completely oxidised state. This would leave the destroyed methane cloud dry at the expense of oxidising any iron grains to completion. If the same dust cloud, now without the reduced iron oxide, meets up with another methane cloud, the peroxide will still destroy the methane but the grains no longer have a need for the water and it will appear in the spectra as a byproduct.

The coincidence problem is that there would have to be an order or queing where the dust cloud attacks the methane cloud. The dry destruction would have to be first and then the same dust cloud would attack another methane cloud leaving the water signal extant. There would have to be quite a number of futher observations that track both the required dust cloud and its interactions with the methane clouds to witness the first dry then wet rhythm to test this idea before it could be taken too seriously. From observations of hydrogen peroxide spectra at different elevations, there are bits and pieces that might help to suggest this scenario, but not enough information to form a well connected theory.