Peak over Threshold Analysis of Heavy Precipitation in Texas

Heavy and extreme precipitation is a particular problem in Central Texas along the I-35 corridor in what is known as Flash Flood Alley. Both the geography and meteorology of the region provide a unique set of conditions which lead to dangerous flash flooding. Limestone bedrock close to the surface limits infiltration of rainwater, which promotes runoff into narrow, deep incised streams, which can rise rapidly. The Gulf of Mexico provides a rich source of moisture for showers and thunderstorms. The Balcones Escarpment, a feature which bisects the state into the Coastal Plain and the Hill Country, runs in a north-south line between San Antonio and Dallas, and may promote orographic lift. This lift serves as a focusing mechanism for heavy and extreme precipitation. One anticipated impact of climate change in Texas is an enhancement of the drought-flood cycle, which will have important consequences for both the agricultural sector and the safety of the population living along the heavily populated flash flood alley. This analysis will explore temporal and geographic variation in heavy and extreme precipitation in Texas over the last century.

More than one-hundred years of archived precipitation data from the National Weather Service will be used to complete a peak-over-threshold extreme value analysis of precipitation events in the state of Texas. The peak-over-threshold analysis allows multiple extreme events per year, which is appropriate for the region under consideration. Cities in the study will be from all sectors of the state: the Panhandle, far West Texas, Central Texas, Southeast Texas, and the Rio Grande Valley. Data will be examined in overlapping 30-year segments. Each segment will be subject to a declustering algorithm to ensure independence of precipitation events. The 95th percentile of each thirty-year dataset will be used as the threshold for heavy and extreme events. Events whose daily precipitation totals exceeded the threshold will be included in the analysis. A Generalized Pareto Distribution (GPD) will be fit to the exceedance data. Both GPD fits and extreme value curves will be examined across the study period 1900-2015 for each location.