Observed Extreme Rainfall Trends Along the United States Gulf and Southeastern Coasts
Abstract
Rainfall that occurs at the tails of the precipitation distribution are considered to be the extremes, which are of critical importance for regional flood planning and infrastructure design. Previously, a stationary climate assumption had been used for computing extreme rainfall estimates. However, this assumption has since been shown to underestimate current precipitation frequency and intensity in the presence of an upward trend in precipitation. Trends in the rainfall extremes, calculated through a nonstationary lens, have been documented in observations and global climate models. This study computes extreme rainfall trends across the Gulf and Southeastern Coasts of the US and presents additional methods for addressing current uncertainties in assessing the climate-driven trends. Nonstationary generalized extreme value (GEV) models are applied to historical data (1890-2019) and fit to the log of precipitation while using CMIP5 global mean model surface temperature as the covariate. Comparisons are also made to NOAA Atlas 14 using a stationary GEV model and truncated datasets.
Trends for the composite stations vary from -30% to ~ +60% across the area of study while the pooled trends vary from -5% to +27% at the 2-yr return period and -8% to ~ +50% at the 100-yr return period. The trend estimates themselves are sensitive to various weather events and are therefore unreliable as individual estimates. The aggregate of the pooled trends yield mean trend estimates and associated 95% confidence intervals of 8.99% +/- 3.90% for the 2-yr return period and 13.43% +/- 3.91% for the 100-yr return period using a spatial statistical model. Through the comparisons to NOAA Atlas 14 extreme rainfall estimates, it can be seen that updates to the stationary estimates are necessary.
Citation
Jorgensen, Savannah K. (2022). Observed Extreme Rainfall Trends Along the United States Gulf and Southeastern Coasts. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /197368.