The spatial distribution of rainfall extremes and the influence of El Niño Southern Oscillation

Abstract

Extreme rainfall does not occur in spatial isolation. Rainfall occurs in a region, and within that region nearby locations are likely to experience similar impacts due to spatial dependence. While univariate extreme value models provide the easiest statistical modelling approach to rainfall extremes, practitioners and researchers adopting statistical models without spatial dependence are liable to underestimate potential impacts. To minimise the adverse impacts of extreme rainfall, an understanding of the extreme precipitation field is required. To highlight how a spatial model with dependence compares with univariate models of extremes, a max-stable model is fitted to the daily annual maximum rainfall in a case study region of South East Queensland, Australia. This case study region was selected as it can be used to illustrate how climate drivers, such as El Niño Southern Oscillation (ENSO), can affect the extreme precipitation field and subsequently the distribution of spatial random variables. In adopting a max-stable model it is possible to produce simulations of the daily annual maximum rainfall field. These simulations can be used to inform urban planning strategies. This includes showing that the probability of a historical flash flood occurring was much higher given the strong La Niña phase of ENSO compared with an El Niño phase. The results presented aim to shift the dialogue from a univariate discussion to a discussion about how models of spatial extremes with dependence can be used to better understand the probability of extreme rainfall events and account for the influence of ENSO.