When you say you're not convinced that it's possible to be really sure you've captured all of the benefits, do you mean captured their exact value? Or a hard upper bound of such? Or a soft upper bound with a given confidence interval? Or even just enumerated all of the benefits?

I think calculating an upper bound with confidence interval for the value of a tree in a city would be a really interesting exercise. To do it properly you'd have to account for everything from tangible benefits (shade, reduced CO2) to vague generalities such as residents who "feel happier living in a city with lots of trees". I think it'd be possible, but definitely not trivial.