Interannual Variability of Streambed Hydraulic and Solute Transport Properties Quantified Using Mixed‐Effects Models

ABSTRACT

Streambed sediment transport, scour and deposition create spatiotemporal variability in streambed properties, contributing to uncertainty in hyporheic zone characterization. Accurate statistical characterization of this variability is further complicated by spatial dependence among measurements, which must be explicitly accounted for. To examine the potential effects of mobile sediments on streambed characterization, this study evaluated year‐to‐year changes in streambed properties including point measurements of streambed hydraulic conductivity ( K _v ), hydraulic head gradients, seepage flux, and dissolved mercury ( Hg ) stream concentration and mass flux (of Hg into the stream) along a reach of East Fork Poplar Creek in Tennessee, USA, over three consecutive years under baseflow conditions. Anomalously high streamflow events were recorded prior to the 2019 measurements, contributing to year‐over‐year changes in streambed properties. Mean log‐transformed K _v and inverse‐square‐root‐transformed Hg concentration differed significantly among years, with 2019 measurements having a lower mean than those from 2017 and 2018. A comparison between a one‐dimensional spatial mixed‐effects model and a model assuming independence showed that ignoring spatial autocorrelation led to downwardly biased standard error estimates and inflated statistical significance. Standard errors (SEs) from models that accommodate spatial autocorrelation can be substantially larger than those from models that assume independence. This study highlights both the temporal sensitivity of streambed properties to hydrological disturbance and the importance of using statistical frameworks that explicitly account for spatial dependence when evaluating changes in hyporheic exchange and contaminant transport.