engleski

The statistical analysis of wet and dry days by binary DARMA (1, 1) model in Split, Croatia

The study of the floods and droughts requires the study of the characteristics of wet and dry day sequences. Relying on empirical relative frequencies of sequence length only, it may create completely the wrong idea about probability of very long sequences which have more practical importance than the short sequences. Various statistical models give more reliable estimates. Dryness or wetness of a month can be determined using daily precipitation data. Nevertheless, in some cases these estimates may also give the wrong impression, so monitoring the number of wet and dry spells is also important. In this work the Discrete Autoregressive Moving Average model (DARMA(1, 1)) is used to describe the wet-dry day sequences in Split, middle Adriatic coast of Croatia. These estimates are also compared to the DAR(1) (First-order Markov chain) model estimates. The daily precipitation data from Observatory Split-Marjan during the 1948-2000 period is used.

wet and dry spells; DARMA (1; 1) model; DAR(1) model; middle Adriatic coast; autocorrelation coefficient

nije evidentirano