A global ECMWF data set comprising ten years of daily routine T63 anal
yses is employed for the derivation of regional and temporal tropopaus
e fold statistics. Due to the small scale of folds relative to grid re
solution it was found to be convenient to identify them by joint occur
rences of potential vorticity maxima and upper tropospheric Q-vector d
ivergences as indicators. The applicability of the proposed method is
shown by diagnostic studies of episodic mesoscale model simulations, w
ith 40-50 km resolution as a basis for reference and comparison with t
he global data set. Tropopause folds are a frequent and rather uniform
ly occurring phenomenon of the extratropics pole-ward of 30 degrees la
titude. On the global scale, and based on daily counting, there are ab
out 11 000 tropopause fold days per year. In order to account for the
different folding depth and intensity a four step classification is pr
oposed to define a tropopause fold activity by weighted sums, in terms
of four intensity classes. Tropopause fold activity in the Northern H
emisphere is nearly twice as high and less homogeneously distributed a
s in the Southern Hemisphere, where only 62% of the tropopause fold fr
equency is estimated. The frequency of tropopause folds is significant
ly smaller in the summer than in the other seasons.