Rainfall event analysis in the north of Tunisia using the self-organizing map

Sabrine Derouiche (1, 2) , Cécile Mallet (1) , Zoubeida Bargaoui (2)

1 SPACE - LATMOS
2 LR-LMHE ENIT - Laboratoire de Modélisation en Hydraulique et Environnement [Tunis]

During the last decade, scientific research has shown a growing interest in quantifying the impact of climate variability on rainfall and water resources. In Tunisia, the variability of rainfall is at the origin of natural disasters extremely expensive in human lives and responsible for innumerable material damages. So, the analysis of rain events characteristics is an essential step for improving our understanding of spatial and temporal variation in precipitation. For this study, 70 raingauge stations in Northern Tunisia are used over 50 years (1959-2008). It is proposed to adopt a seasonal analysis (December- January- February) with a separation of daily rainall time series into events after the determination of the minimum inter event time MIT. This data transformation give us 6 rainfall variables: (1) rainfall event number, (2) total event duration, (3) Average precipitation, (4) total precipitation, (5) Average intensity, (6) Average duration. Those variables are clustered by using the method of Self Organizing Map SOM and give 4 type of seasons.

Domaines: Météorologie

Sabrine Derouiche, Cécile Mallet, Zoubeida Bargaoui. Rainfall event analysis in the north of Tunisia using the self-organizing map. Proceedings of the 9th International Workshop on Climate Informatics, Oct 2019, Paris, France. 5 p. ⟨insu-02314437⟩