Author(s): Lucas Rogerio Gomes; Maria Manuela Portela; Luis Angel Espinosa; Martina Zelenakova

Linked Author(s): Maria Manuela Portela, Luis Angel Espinosa

Keywords: Madeira Island; Time-series data; Extreme rainfall; Alluvium; Bivariate analysis; Copula; Return period

Abstract:

Extreme rainfall events interacting with exposed and vulnerable human and natural systems can lead to disasters. Furthermore, they can be the trigger of tragic alluvium flooding, as for the historical case of the small Portuguese island of Madeira (area c.a. 740 km²), located in the North Atlantic Ocean. However, the understanding on how extreme rainfall events relate to alluvium flooding is still a challenging task. Based on rainfall data at Funchal rain gauge station (located on the southern slope of the Island) and on alluvium records that affected Funchal area, the research sought (i) to ascertain the exceptionality of the extreme rainfall events by coupling them with the rainfalls that preceded and/or followed them and (ii) to understand if such exceptionality could be an indicator of alluvium flooding occurrences. Other studies took into consideration univariate approaches of hydrological and geomorphological data related to alluvium flood events. The study presented herein solely focuses on extreme rainfall data and bivariate copulas aiming at calculating the return periods of coupled extreme rainfall events, some of them, known for having triggered alluviums floods. Extreme rainfall observations were identified based on the annual maximum series (AMS) approach to classify hourly and daily rainfall series over 34-year (from October 1980 to September 2014) and 80-year (from October 1937 to September 2017) periods, respectively. By using bivariate copula analysis, the dependence between (i) annual maximum rainfall and (ii) rainfall immediately before and/or after the corresponding annual maximum is assessed. The joint and conditional return periods of each coupled rainfall event were calculated, and their relationship with alluvium flooding events were analysed. The results of this study conclude that the bivariate copula approach may be useful and adequate to characterize the exceptionality of the extreme rainfalls, namely by allowing the assignment of return periods to extreme rainfalls coupled with those that preceded and/or followed them. Another important finding of the study is that the extreme rainfall events that triggered alluvium floods tend to have much higher return periods than those with no coupled alluvium events. Among the former events, there is the rainfall event on late February 2010, which triggered a well-documented alluvium flooding that caused significant property damage and fatalities. Additionally, the methodology applied proved that the event of February 2010 was indeed exceptional having the highest assigned return period in the analysed time span. Finally, given the susceptibility of Madeira Island to the highly occurrence of alluvium events, this work assists in (i) understanding how extreme rainfall can be related to such events, and (ii) the adequacy of bivariate copulas in hydrological studies.

DOI: https://doi.org/10.3850/IAHR-39WC2521711920221315

Year: 2022