There is a very useful trick to use:
mask = reshape(T,1,[]) > 5;
starts = strfind([0 mask], [0 1]);
stops = strfind([mask 0], [1 0]);
Now starts(K) and stops(K) together indicate indices of the beginning and end of runs of T > 5; you can calculate durations by looking at the difference between them.
Careful, though: If the data were [2 6 3] then mask would be [0 1 0] and starts would be 2 and stops would also be 2. The run starts at index 2 and ends at index 2, which is duration 1.
Now, you can filter those runs out to only find the ones of sufficient duration. Or you can modify the strfind() search. If you extend the [0 1] to [0 1 1 1 1 1] and [1 1 1 1 1 0] in the other one, you only match runs of at least 5. But when you do that you will find you need to adjust the indices returned by stops.
What do you want to do in the case of T being exactly 5 ? Should [7 8 5 7 9 8] qualify as a run of 6 days, or does it get disqualified because exactly 5 is not > 5 ?
