Regional Frequency Analysis of Droughts Using Copula Functions (Case Study: Part of Semiarid Climate of Fars Province, Iran)

Agricultural sector of Fars Province plays a substantial role in ensuring food security in Iran. Unfortunately, in recent decade wide areas of the province have been affected by drought phenomenon. Due to significant correlation between drought characteristics and the existence of ungagged areas, in this research, a regional bivariate analysis is recruited for meteorological drought. The aims of this analysis are to shed the light upon the effects of interdependence drought characteristics on frequency of occurrence and return period, in addition, improve the results comparing to local frequency analysis in Fars Province. Drought events are analyzed via the Standardized Precipitation Index and run theory. For regional frequency analysis, the index drought procedure which is coupled with the L-moments method is employed and the bivariate distribution of drought is estimated through the copulas for a homogeneous region. The results present that the marginal distributions for duration and severity variables are exponential and 2-parameter gamma, respectively. The Gumbel copula is the most suitable copula for representing joint distribution of drought characteristics. Comparing univariate and multivariate return periods for different non-dimensional severity and durations having identical marginal cumulative distribution function amounts indicate their values are relatively close up to 20 year. However, a significant discrepancy has been revealed when univariate return periods are 50 and 100 year which have a critical role in risk analysis and water resources management. To sum up, drought severity with duration greater than 6 month was estimated in different return periods for all of stations and it is observed that drought severity in northwest is greater than the southwest in under-studied region.