I’ve discussed this before - it’s what I call the “mostly harmless” problem. Whilst [low prob x high impact] and [high prob x low impact] are the same in EV terms (so useful for insurers etc) they aren’t the same when preparedness is the key outcome. In simple terms these two axes shouldn’t be reduced to one dimension (colour) because too much valuable information is lost in the final transformation.
I like the matrix, but not the colour overlay. Perhaps retain the matrix but go with a “PxI” labelling system where
P(Probability) is Low(L), Medium(M), High(H), Very High(V)
I(Impact) is Low(1), Medium(2), High(3), Extreme(4)
So Isha would have been “upgraded” to an V4 for a limited area late last night.
Why? Because most people don’t need to make special preparations for high probability / low impact events, but they should for a low probability / high impact events. At the moment these two scenarios both end up being called Amber. That’s where the loss of valuable information occurs.