Spatio-temporal modeling of urban extreme rainfall events at high resolution

Abstract

Modeling precipitation and its accumulation over time and space is essential for flood risk assessment. We here analyze rainfall data collected over several years through a microscale precipitation sensor network in Montpellier, France, by the OMSEV observatory. A novel spatio-temporal stochastic model is proposed for high-resolution urban rainfall and combines realistic marginal behavior and flexible extremal dependence structure. Rainfall intensities are described by the Extended Generalized Pareto Distribution (EGPD), capturing both moderate and extreme events without threshold selection. Based on spatial extreme-value theory, dependence during extreme episodes is modeled by an r-Pareto process with a non-separable variogram including episode-specific advection, allowing the displacement of rainfall cells to be represented explicitly. Parameters are estimated by a composite likelihood based on joint exceedances, and empirical advection velocities are derived from radar reanalysis. The model accurately reproduces the spatio-temporal structure of extreme rainfall observed in the Montpellier OMSEV network and enables realistic stochastic scenario generation for flood risk assessment.

Figures (15)

• Figure 1: Location of the 20 rain gauges of the OMSEV network over the Verdanson river catchment in Montpellier
• Figure 2: Location of the $20$ rain gauges of the OMSEV network (red dots) over the COMEPHORE reanalysis (blue grid).
• Figure 3: Estimations of variogram parameters with maximum likelihood optimization on $r$-Pareto simulations with $49$ sites and $24$ time observations. The true parameters are indicated by red crosses. Here random advection by replicate (episode) is considered.
• Figure 4: Estimations of variogram parameters with maximum likelihood optimization on $r$-Pareto simulations with $49$ sites and $12$ time observations. The true parameters are indicated by red crosses. We use a random advection by replicate (episode) with fixed $\eta_1=0.5$ and $\eta_2=1.6$.
• Figure 5: Quantile-quantile plot of the EGPD fitting with bootstrap confidence intervals on four rain gauges of the OMSEV network with site-specific left-censoring.