Estimation of return dates and return levels of extreme rainfall in the city of Douala, Cameroon

Abstract

The problem of extreme phenomena with a more precise estimation of their return periods for early warnings, notably to preserve the safety of populations and properties, arises all over the world. This work develops another aspect in the estimation of Return Levels (RLs) and Return Periods (RPs) of extreme precipitation in particular and natural risk in general. In particular, it gives the Return Dates (RDs) with their Confidence Intervals (CIs). The RPs, the RLs and their CIs for extreme rainfall were also investigated. These estimates were made by approaching the peak over a threshold chosen by the Generalized Pareto Distribution (GPD). The CIs of RPs and RLs were determined by the Delta method. The daily rainfall data used were obtained from the data of the synoptic report for the period 2011 to 2021 for the Douala weather station (more details can be found on http://www.ogimet.com/guia.phtml.en). To validate the methods used, real cases of floods occurred in Douala city were considered: for example, a local press compiled flood dates and mentioned that a flood occurred on the April 16, 2013 in this city. Following the data of synoptic report, the corresponding amount of rainfall was around 150 mm. The results obtained have shown a RD on the August 12, 2014. The confidence intervals of return levels and return dates determined by the Delta method were [131.66, 168.456] and [June 23, 2014, January 02, 2015], respectively. These results are in agreement with the data of synoptic report since the rainfall amounts was 132.2 mm (belonging to the confidence interval of return levels), on the August 11, 2014 (belonging to the confidence interval of return dates). These predictions of RDs and RLs with their CIs, at reasonable time scales, can help for efficient management of floods and thus, improve early warnings for safety of populations and goods.