Abstract
Rainfall extreme value analysis provides information that has been crucial in characterizing risk, designing successful infrastructure systems, and ultimately protecting people and property from the threat of rainfall-induced flooding. However, in the Houston region recent events (such as the unprecedented rainfall wrought by Hurricane Harvey) have highlighted the inability of existing analyses to accurately characterize current climate conditions. Specifically, there has been little research investigating how spatial patterns of extreme precipitation have shifted through time in the Texas Gulf Coast region, which has led to mis-characterization of existing intensity–duration–frequency curves. This study investigates spatiotemporal trends in extreme precipitation in southeast Texas using a statistical approach for peaks-over-threshold modeling that employs a generalized Pareto distribution. Precipitation data from over 600 rain gauges across the region are analyzed in 40-year time windows to evaluate shifts in distribution parameters and extreme rainfall levels through time. Spatial analysis of these trends focuses on highlighting regions with increasing, stationary, and decreasing extreme rainfall through time. Results demonstrate heterogeneity in spatiotemporal trends across the entire study region, but significant increases in extreme rainfall over the Houston urban area. Spatial analysis of these trends focuses on how extreme rainfall has changed within different watersheds. Return level estimates of extreme rainfall values are also compared to the current standards for Harris County. Results from this study identify areas that have experienced significant shifts in extreme rainfall, and can help inform where design standards may be inaccurate or outdated.
Highlights
Extreme weather-related events can cause significant economic, social, and environmental disruption
Flood risk has increased substantially during the previous 30 years, growing from approximately $2 bn/year in 1985 to $4 bn/year in 2015. While this increase has been driven in large part by socioeconomic growth in flood prone areas (Changnon et al 2000; Pielke and Downton 2000; Jongman et al 2012; Visser et al 2014), there is a rapidly growing body of literature that suggests that the frequency and intensity of extreme precipitation events in the U.S is increasing due to anthropogenic climate change resulting in larger flood hazards, more exposure, and less lead time prior to an event (IPCC 2012; USGCRP 2018)
Extending the generalized extreme value (GEV) to include four or twelve maxima per year can help alleviate some of our concerns for there being more than one extreme event per year; we still prefer the generalized Pareto distribution (GPD) approach so there is not a defined number of events we are including
Summary
Extreme weather-related events can cause significant economic, social, and environmental disruption. The intensity–duration–frequency (IDF) curves used to generate design storms are updated infrequently (see, e.g., Weather Bureau Technical Paper No 40 (Hershfield 1961), NOAA Technical Memorandum NWS HYDRO-35 (Frederick et al 1977) and NOAA Atlas 14 (Perica et al 2018)), and the analyses assume that the annual maximum time series are stationary over the period of record (Perica et al 2018) This is problematic for flood risk management since small changes in the upper tails of extreme precipitation distributions can have disproportionate impacts on the spatial extent of floodplains, especially in extremely flat or low-lying watersheds like those in southeast Texas. The shape parameter determines the shape of the distribution, such as if it is symmetric or skewed one way or another
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have