"Norman Lynagh" wrote in message
...

Time doesn't really come into it. What matters is that the wind is changing
direction with time.

Thanks, but maybe I should have spelled out a little what was behind the
question. This is simply trying to define a rule in software whose output
indicates whether the current wind direction is backing or veering and, if
possible, whether the change is happening relatively slowly or quickly.
Inevitably this means finding a simple formula with numeric inputs, which
clearly has to be based on current mean direction compared with the
equivalent mean direction at some specified point in the recent past, eg
15/30/60 minutes previously.

Perhaps the answer is that there is no such rule in common usage? But even
so, I'd be interested in any views on what might be sensible values to use -
an initial suggestion might be to compare the current 2-min mean direction
with the value 30 minutes ago, but happy to listen to alternatives.