MAD filter
In order to improve the behavior of the MAD filter on snow height, it would make sense to add an optional argument, giving the minimum sigma that can be used. Thus if sigma gets too small (no variations anymore), the filter could still accept the next snow fall instead of rejecting it until enough variance has accumulated (as is currently the case).
This requires an improved syntax to provide filter arguments (since this makes one more argument to MAD). This would most probably follow the new syntax for spatial interpolations.