In reply to Webster:
Looks like we both read that graph very differently.
In actual fact if you read the graph properly (you can read the exact dates by hovering the mouse over the graph) it was only below freezing for 24 hours over new year and then it only got down to -2 momentarily before heading straight back up. The air was only colder than the ground for less than 3 days yet in those 3 days the turf temperature dropped from 5.86 deg to under 2.73 deg, highlighting my point of how quickly it can change given the right conditions and it wasn't even that cold!
You can also see how quickly the turf temperature rose around the 25th of December over 3 degrees in a day - these mild wet conditions are always coming through north wales during the winter and I think it casts further doubt on to your theory regarding how important the prior weather is for devils kitchen to come in to nick once really cold weather hits. Sure it may take a day or 2 longer than normal if it's been really mild but I can't see it making much more difference than that tbh.
Insulating the turf with snow can make a much bigger difference however water will still freeze to ice when exposed to very cold air. How much this will affect the system if the ground is unusually warm and then it snows heavily is anyone's guess, although if its really cold water ice does always start forming in my experience anyway. If the ground really was that warm it would be melting the snow pack from the bottom and that would also cool the ground / ground water down.
Post edited at 16:52