Empirical modeling of salinity intrusion length in alluvial estuaries at high water slack: a regression-based approach using global datasets

Abstract

Salinity intrusion in alluvial estuaries poses increasing threats to water quality, ecosystems, and coastal livelihoods under climate change and anthropogenic pressures. This study develops an empirical predictive model for salinity intrusion length at high water slack (HWS). The proposed model uses regression analysis on 42 observations from 22 global estuaries, including the Limpopo, Maputo, Mekong, and Thames systems. Four regression models (two dimensional and two dimensionless) were derived by correlating intrusion length with tidal range, freshwater discharge, estuary geometry, and density gradients. The optimal dimensionless model, excluding average depth to reduce input uncertainty, achieved a coefficient of determination (R²) of 0.9864, root mean square error (RMSE) of 1.5 km, and mean absolute percentage error (MAPE) of 4.3%, outperforming established benchmarks by 75-94% in relative error. Sensitivity analysis identified width convergence length (Sobol index S₁= 0.62) and relative tidal range (S₁=0.28) as dominant controls, with roughness exerting negligible influence. The model’s unitindependent formulation enables rapid, accurate assessments in data-scarce regions, supporting adaptive water management amid projected 15% to 50% intrusion increases by 2050. This work provides a robust, parsimonious tool for estuary managers and a benchmark for integrating empirical and hydrodynamic modeling in dynamic coastal systems.