Statistical detection and no-detection of rainfall change trends and breaks in semiarid Tunisia—50+ years over the Merguellil agro-hydro-climatic reference basin

Abstract

Rainfall variability is an important feature of semiarid climates with major effects on hydrology, and beyond on key water-dependent societal aspects. Eventual changes in rainfall variability are a strong driver of change of hydrological processes, resources, and hazards, up to catchment signatures and spatial arrangements. We deal with observed precipitations and subsequent statistical coefficients available from a network of 15 rainfall gauges over and around the Merguellil catchment (1175 km2), with series ranging up to the 1961–2013 period. We look for eventual annual trends and breakpoints with a set of methods: Mann Kendall test, Pettitt test, Hubert segmentation procedure, Buishand U statistic, and Lee Heghinian Bayesian procedure. The results underline oscillation of dry and wet periods; several studied rain gauges (Tella, Oueslatia foret, Majbar, Kesra foret, Henchir Bhima, and Haffouz DGRE) denote significant trends in annual precipitation. Some breaks are detected but they are not synchronous. These methods reveal the variability of rainfall regimes in the semiarid region and provide a synoptic view of detection and no-detection of symptoms of change. This work gives opportunities to water stakeholders and climate experts in understanding the relationships between climate variability and water availability.