The dynamics of daily weather types according to Schuepp's classification i
n the Alpine region is investigated by means of seasonal Markov chain model
s. A legit model is employed for the transition probabilities of the Markov
chain. The parameters follow from a maximum likelihood estimation. A 1st-
and a 2nd-order Markov chain model are compared and found to yield very sim
ilar results. Model predictions are compared with counted frequencies of se
asonally-averaged joint probabilities for the occurrence of weather types a
t subsequent days, monthly-averaged probabilities of a change of the weathe
r type from one to the next day, and daily averages of the probabilities of
occurrence of the different weather types. All these predictions coincide
with the observations within the statistical limits. The only large deviati
ons occur in the tails of the lengths distributions of uninterrupted episod
es of the two most frequent weather types.