explains the lack of gutters; I've lived where its cold and you have gutters, winter tires, basements where you store your stuff instead of garages which keep your cars' blocks from cracking..)
Anyway if you put a thin plate --trash can lid or inverted bucket-- under a drip from the major "waterfall" you can hear it better. Because you have a bigger watershed, the roof, concentrating it, and a nice sounding board for the waterfall.
With drips, concentrated by a roof, you gain sensitivity but would be swamped by bigger outbursts. A waterfall would be hard to measure.
A computer program could do pulse (droplet) counting. How would it compare (sensitivity, response time, etc) to a traditional graduated cylinder and funnel opening? Extension cable, microphone covered by bucket, computer, pulse counter software.
Actually you could just skip the latter two and just record the sound for later analysis.
That way one teacher could help a bunch of kids ---all they need is the mic, cable, bucket. Or they could bag their iGizmo, set it on record, and put it under the bucket and on a brick.
The "splats per minute collected over an area of A square meters (in flat projection, not tiles :-)" would be a good local measure of precipitation intensity.
You can add nuances like dry tiles at first. There is some delay between rain changes and the effects at the bucket-sensor.
