I have a small objection to the term "catchable" hatch, since the hatch isn't really what you're catching. It just sounds awkward, and to be more accurate you'd need to say "hatch during which the fish are catchable," which sounds even more awkward. I'd rather stick with "fishable" and just be clear on what we mean by it.
Also, the characteristics of the hatch are just one of many factors influencing whether or not the trout are catchable, and different trout can respond differently.
Of course, "fishable" taken literally could apply to
anything unless the river is frozen over and you lack an ice auger. But let's go with the common, idiomatic definition that
a fishable hatch is one in which trout under normal conditions would feed actively on the hatching species. It's tempting to load the definition up with other little caveats and details, but I don't think it's necessary. I think we all know what we mean by the term.
A serious question was behind my rambling though. Determining whether a catchable hatch is occurring comes from experience, instinct, and lucky guesses. Is it possible at all to put numbers on it based on stream size? Or are the numbers the closest one can come to explaining the gut feeling?
That is a really good question. My academic background in mathematical modeling tells me you can
theoretically put numbers on just about anything, but sometimes it's prohibitively hard.
This question is more about trout behavior than about the insect hatch, and that requires a
lot more variables to model, even for a simple and generalized result. It also involves some tough decisions and unknowns about exactly what the trout is optimized for and how effectively it can distinguish prey from debris.
I worked pretty hard on a similar question (when should trout feed selectively?) a couple years ago for my final project in an ecological modeling class at Cornell. I had to strip the problem to the bare minimum parameters, but I got some pretty interesting quantitative results. They wouldn't work to actually predict how real trout behave, but they give some good (and I think correct) insights into how the important factors interrelate. I'll email you a PDF of the paper I turned in, which should be good food for thought if nothing else. (If anyone else would like to see it, just ask.)
Edit: I just posted this and saw Gonzo already referred to the paper I just described. These are the 'very interesting calculations.' He's right that technically my paper only models the first seasonal encounter with the hatch, but more importantly it doesn't consider past or future encounters at all, meaning it's probably just as bad for the first hatch as it is for the others. It doesn't get quite that nuanced.