Developing a method to model extreme rainfall events using drought indices in northern Algeria under climate change

ABSTRACT To enhance water resource management and mitigate risks in vulnerable areas, a comprehensive understanding of rainfall dynamics, especially extreme events, is imperative. Employing the extreme values theory, this study objectively examines the characteristics of extremely dry and wet events in north Algeria, elucidating their recurrence frequency and the return levels connected to the 2-, 5-, 20-, and 100-year return periods. Two principal methodologies were utilized: the Block-maxima and minima approach (BM) and the peak-over-threshold approach (POT), which identifies values surpassing predefined thresholds like the effective drought index (EDI) or the standardized precipitation index (SPI). Analysis of 41 years’ worth of data employed the generalized extreme value model (GEV) for BM and the generalized Pareto model (GPD) for POT. These models underwent rigorous goodness-of-fit tests – Anderson–Darling, Chi-square, and Kolmogorov–Smirnov – confirming their aptness in representing the data. The findings unveil significant rainfall trends spanning decades: a decline in the 1980s and 1990s, a subsequent upsurge in the 2000s, and a subsequent decrease by the late 2010s. Moreover, a notable correlation between standardized drought indices (EDI/SPI) was observed across most surveyed stations. The GEV and GPD models emerged as optimal choices for data analysis, showcasing their compatibility with observed values.