Abstract: The following paper deals with the modelling of a 30-year-long daily rainfall time series (1965?1994) from 39 weather stations in the Basque Country region in Northern Spain. Markov chain models are used to explore the temporal rainfall patterns in the area and to characterize the persistence of daily rain in terms of the order of the discrete Markov process. The Markov models?theoretical probabilities?are fitted to the empirical probability of wet spells lasting one more day than a specific length in the time series under the assumption of three different values (1, 5 and 10 mm) of a minimum threshold. Markovian orders that first fit the empirical distributions are used to map the theoretical persistence of daily rainfall in the Basque Country area through the use of geostatistical interpolators. Markovian orders have been confirmed as remarkable stochastic models for fitting empirical distributions of wet spells in this area, ultimately being shown as a spatial representation of theoretical persistence.

Authorship

Download reference