Evaluation of the transition probabilities for daily precipitation time series using a Markov chain model
Liana Cazacioc, Elena Corina Cipu · 2005
The Markov models are frequently proposed to quickly obtain forecasts of the weather ”states” at some future time using information given by the current state. One of the applications of the Markov chain models is the daily precipitation occurrence forecast. There is tested a Markov chain model with two states for the daily precipitation in summer and winter seasons of 1961-1990 at several stations in Romania. The states of the Markov chain are precipitation occurrence and precipitation non-occurrence, that is wet and respectively dry days. There are computed the sets of conditional (or transition) probabilities for first-order, second-order and third-order Markov chain. To find the most appropriate model order among the dierent orders of the Markov chains for the daily precipitation series, the Bayesian information criterion (BIC) was used.