Rainfall varies according to a lot of different factors, which include the location of mountains and hills, bodies of water, and vegetation.
One approach is to collect data interactively with the environment. To do that keep placing rain gauges. If they are reading the same, separate them. So, by trial and error you determine a good distribution.
The answer, as is often the case for questions about optimums, depends on objectives and constraints.
How accurately do you want to measure rainfall? The more gauges the more precision, but obviously at greater cost.
How important is the spatial element, which in turn might depend on the objectives of monitoring? Is it more important to get accurate total ...