Daily mean water level regression model for the Sacramento-San Joaquin Delta, USA

Water levels in deltas and estuaries vary on multiple timescales due to coastal, hydrologic, meteorologic, geologic, and anthropogenic factors. These diverse factors increase uncertainty of and may bias relative sea level rise (RSLR) estimates. Baranes et al. (in review) evaluates RSLR in San Francisco Bay and the Sacramento-San Joaquin Delta, USA by applying a physics-based, nonlinear regression to 50 tide gauges. We estimate the spatially varying controls on daily mean water level for water years 2004-2022. Results show that median river flow causes water level variations of 10 mm-1.3 m, and high summertime pumping rates lower water level by up to 0.35 m. High (95th percentile) coastal water level perturbations, tidal-fluvial interaction, and wind forcing cause water level variations of -50-50, 0-70, and 0-60 mm, respectively. Removal of these interfering factors greatly improves RSLR estimates, narrowing 95% confidence intervals by 89-99% and removing bias due to recent drought. Results show that RSLR is spatially heterogeneous, with rates ranging from -2.8 to 12.9 mm y-1 (95% uncertainties <1 mm y-1). RSLR exceeds coastal SLR of 3.3 mm y-1 in San Francisco at 85% of stations. Thus, RSLR in the Delta is strongly influenced by local vertical land motion and will likely produce significantly different future flood risk trajectories. Datasets and scripts provided here can be used to run the regression model at Delta gauges.