Quantifying local scour at bridge piers in coarse-bed streams: insights from on-site measurements and nonlinear quantile regression

This study investigates local scour at bridge piers in coarse-bed streams, which are characterized by an armor layer of large sediment particles. Unlike traditional approaches that rely on laboratory experiments conducted in sand-filled flumes, this research leverages a wide-ranging collection of on-site measurements from coarse-bed channels to develop more representative scour prediction models. The study employs nonlinear quantile regression (NQR) to estimate scour depths at quantile levels of 0.5 (median), 0.95, and 0.99. Models developed for higher quantiles provide statistically robust safety margins essential for bridge design, significantly improving predictive accuracy over widely used empirical equations derived from sand-bed conditions. This contribution enhances the understanding of local pier scour in coarse-bed streams and advances engineering practices for safer and more resilient bridge infrastructure in these challenging environments.