Resiliency of hydraulic infrastructure designs in a climate hot-spot at the intersection of two climate zones

Flood damages to hydraulic infrastructure leads to devastating economic losses in the Southeast Texas/Southwest Louisiana region. This area lies at the cusp of the humid southeastern US and the more arid southern high plains and exhibits extreme climate variability. Traditional design protocols recommend the use of Log Pearson-III (LP3) based flood frequency analysis (FFA) using annual maximum flows (AMF). The reasonableness of this approach is increasingly being questioned due to intensification of flooding in recent years with possibility of nonstationary climate shifts. Using long-term historical data and an ensemble of downscaled climate data, the study demonstrates that streamflows exhibit stationary behavior and any noted shifts in precipitation lie within the high variability noted in the region. While nonstationarity is not of concern, the study also demonstrates that hydraulic designs based on LP3-AMF techniques are not resilient (defined here based on withstanding capacity and reliable operations over a design life-time). Two new resiliency indicators – Flood Withstanding Capacity (FWC) – defined as the ratio of anticipated flood quantile to the design flood quantile for a given return period and Reliability Reduction Factor (RRF) – defined as the ratio of the anticipated reliability to the design reliability over the design lifetime of the structure are developed. The level of conservatism within the study area was noted to be AMF-GEV ​∼ ​AMF-GEV (nonstationary) ​< ​LP3-AMF ​< ​GP-PDS ​< ​LP3-PDS. Data are readily available for implementation of partial duration series, the non-inclusion of this approach in Flood frequency analysis is neglecting valuable information that lead to better resilient designs. The framework and resiliency indicators developed here are generic and can be applied in any part of the world. • Two new resiliency indicators: FWC and RRF are developed to evaluate hydraulic design protocols. • Commonly recommended, Log Pearson 3 distribution does not yield conservative designs, especially for high return periods. • Partial duration series (PDS) based frequency analysis is recommended to obtain more reliable designs with available data. • Using LP3 distribution for partial duration series provides conservative designs than the Generalized Pareto distribution. • Variability of precipitation masks evidence of streamflow nonstationarity in southeast Texas/southwest Louisiana region.