| Popis: |
The limited number of very wet or dry events in the historical rainfall series make it difficult to estimate their occurrence probabilities. For this reason, the scientists have to resort to stochastic models, by assuming a given structure of the underlying hydrological series. In this study, a stochastic procedure for modeling monthly data was applied to rainfall series registered in five regions of southern Italy (Campania, Apulia, Basilicata, Calabria and Sicily regions). The model adopts an autoregressive process for the residual correlative structure of monthly rainfall data, previously normalized and deseasonalised. Through a Monte Carlo technique, based on the proposed model, synthetic data were generated for each rain gauge. Then, extreme dry and wet periods and their occurrence frequencies were estimated at various time scales by applying the Standardized Precipitation Index (SPI) to the synthetic data. The results clearly show greater probabilities of dry conditions than wet conditions. These outcomes are more evident when long time scales are considered. |
