Empirical modeling of streambed hydraulic conductivity in ephemeral rivers: Insights from sediment characteristics in arid environments

Abstract

This study presents a pioneering investigation into the relationships between vertical hydraulic conductivity (K_v) and sediment properties through an extensive field measurement of 15 ephemeral sandy-bed rivers in Razavi Khorasan province, Iran. A robust dataset comprising 300 in situ K_v measurements and grain size analyses was collected from 2019 to 2022, including both main channel and bar morphological units. Statistical analyses, such as coefficient of variation and skewness, revealed significant deviations from normality in the data distributions, particularly pronounced for bar samples, emphasizing the limitations of conventional statistical frameworks based on normality assumptions. Applying nonparametric techniques of Kruskal–Wallis and Mann–Whitney, the study uncovered strong interdependencies between K_v and key sediment descriptors, including armor ratio, mud percentage, and grain size (d₁₀). Notably, main channel data indicated a stronger tendency towards normality relative to the bar data, reflecting the inherent disparities in sediment dynamics and hydrologic regimes across these various morphological units. The data samples were randomly split into two groups: 70% for calibration and 30% for validation to develop a novel empirical model, explicitly incorporating the interdependent effects of armor ratio, mud percentage, and d₁₀ on K_v. On the 70% calibration set, R² and RMSE were calculated to evaluate model fit. On the 30% validation set, RMSE was calculated to evaluate predictive accuracy. Thereafter, the validated data were used to assess performance of the adopted model. The model is meant for a specific range of sediment characteristics, including uniformity coefficients (5.2 < U_c < 141), characteristic grain sizes (0.05 mm < d₁₀ < 0.62 mm), mud percentages (0.01 < m_p < 7.2), and armor ratios (0.9 < A_r < 6.2), which are particularly relevant in ephemeral rivers. Finally, the proposed model yielded results outperform those of several widely adopted empirical models, reflecting the effect of mud percentage and armor ratio to provide accurate prediction of the K_v within the scope of this study.