Zusammenfassung
The pluviometric regime of Catalonia (NE Spain) is analyzed from the point of view of empirical dry period lengths by considering 78 rain stations for an average of 34 years. Two possible statistical models are tested. The first model is the exponential distribution that offers reasonable results for moderate or long sequences of dry days. The other possibility comes from Markov chains of first or second order, with two or four states quantifying precipitation amounts, that are used to give better results for all the range of sequences. The Kolmogorov-Smirnov test has been applied with the aim of verifying the fit between empirical probabilities of the sequences and theoretical probabilities given by the exponential distribution and the Markov chains. It is noticeable that the Markov chain of second order is many times the distribution either satisfying the test criteria or the closest one to its fulfilment. The exponential distribution satisfies better the test criteria only for a few gauges and we have to keep in mind that for a remarkable number of pluviometric stations, none of the proposed models accomplish the test. In spite of these limited results, the Markov chains are employed to quantify important aspects which can not be studied by the exponential distribution. These aspects include return periods for a new dry or wet episode and the stationary probabilities for different precipitation amounts quantifying the states of the chains, among other aspects. As a global achievement and remembering the limited success of the test, the results depicted by Markov chains of first and second order are employed to show a differentiated behaviour among the Pyrenees and Pre-Pyrenees areas, the Central Basin, the Littoral and Pre-Littoral areas, the Transversal chain and the southern Mediterranean coast. © 1998 Royal Meteorological Society
Nutzer