Gaunty,
Yep, there cerrtainly is Stuff Out There that can help with such decisions. Modern ground proximity warning systems (GPWS) take data from the aircraft nav/flight control system (i.e. defining flight path), predict future flight path based on that data, and compare the prediction with a digital model of the world ahead. You make some assumptions in predicting future flight path - e.g. that the pilot will continue the current manoeuvre for the predicted period - and also calculate a "recovery" manoeuvre based on known aircraft capability (for example, a wings level pullup manoeuvre at full back stick, i.e. on the AOA limiter at low speed or the g limiter at higher speed. If the wings aren't level, include time to roll wings level. Add on a time for generating a suitable warning and for the pilot to react to it). In an ideal, non-risk averse, non-safety/hazard analysis world, the pilot can rip around the sky with great abandon until he gets the warning, then roll wings level and pull, and shortly afterwards zip off to complete his aeros sequence. Of course, the sensible thing is to add another "advance" to the generation of the warning, so the pilot knows, for example, that if he continues his current manoeuvre for, say, another 3 seconds he'll generate the recovery warning and have to take full recovery action.
However, the picture gets murkier when 1) you consider possible error sources and 2) the dreaded hazard analysis gets involved. 1) is defined by navigation and flight control system data accuracies (actually, pretty good nowadays, and well inside the nosie levels needed for sensible predictions), by the accuracy of the digital terrain model (scope there for errors from spot spacing and integrity) and in the assumptions made for pilot reaction time, generation and presentation of warnings, and the like. Mix this lot with a design engineer who fears being sued by the widow if his allowances or data are wrong, and you end up with pessimistic assumptions, which translates into kit that gives you warnings when the flight path is obviously and credibly safe: "Cry Wolf" syndrome. That's disasterous, because when the warning IS genuine and vital, the pilot may not respond to it in time or at all. You therefore have to juggle the settings to minimise nuisance warning while still providing a worthwhile level of warning - and that's quite possible.
As for 2) that's even more disasterous. The probability of non-flight safety critical (Class 2) software failing ( to the hazard men) is 1, which means if any of your system - data inputs, GPWS box itself, warning generator, cockpit displays, audio - goes through Class 2 software, it must be considered unsafe (e.g. it might not give a warning when it should) and cannot therefore be used to maintain safe operation. It ignores the pragmatic reality that it does NOT fail every time - even Bill Gates' stuff doesn't - and that it will therefore provide a major improvement in operating safety for the majority of operating time. I've lost friends and collagues in Jaguars, Tornados, Phantoms and Lightnings all of whom would have been saved by this kit had it been there, and my heat level rises when someone suggests it's not safe to use because of a software engineering philosophy!
OK, end of rant, time for a cup of hot chocolate and a lie down in a quiet room.