From Pattern to Process at a River Confluence: A Process-based Reactivity-Hydrodynamic Framework from High-resolution Synoptic Monitoring

High-resolution, water quality observations reveal complex instream transformations and mixing of solutes dependent on the source and the prevailing flow conditions, with both concentration and dilution effects apparent. Here we explore river confluence behaviour using a combination of an Acoustic Doppler Current Profiler (ADCP) and multiparameter water quality sonde to resolve solute and particulate behaviours of tributary mixing. We hypothesise that confluence-scale functional heterogeneity is spatially persistent, but that hydrological forcing systematically alters the predominance of source- and sink-type behaviours spatially across the confluence. To test this, we implemented an innovative quasi-synoptic field campaign at the Kennet-Thames River confluence (Reading, UK). Using an uncrewed moving-boat platform (ArcBoat), we collected simultaneous, high-resolution data on velocity, nutrients (NH₄⁺, NO₃⁻), fluorescent dissolved organic matter (fDOM), and turbidity on three days. Each time, the reach was subdivided into 14 fixed spatial zones, allowing reproducible analysis across the three hydrologically distinct campaigns: higher winter flow (12 Feb 2025), a rainfall event pulse (26 Feb 2025), and low spring baseflow (14 Mar 2025). 

We evaluated solute behaviour using a Reactivity Index (RI) based on conservative mixing and integrated it with a Hydrodynamic Index (H) to classify each observation into process-informed categories (e.g., Reactive, Retentive, Low-energy depletion, Attenuating, and Conservative). Extending the same logic to turbidity yielded complementary particulate-transport classes (Local Input, Advective Input, Sediment Deposition, Advective Dilution and Conservative mixing).

Zone-wise analysis revealed exceptionally strong and persistent spatial structuring of functional classifications across campaigns (Kruskal–Wallis ε² = 0.28–0.81, p < 0.0001 for solute RIs). Across the fortnightly transition from event to baseflow conditions, the Kennet-influenced pathway exhibited a coherent regime shift in both dissolved and particulate classifications: during the rainfall snapshot, NH₄⁺ enrichment and turbidity input classes dominated, whereas under baseflow the same corridor shifted toward attenuation-depletion and dilution-deposition dominance.

These results demonstrate that confluence function is organised into distinct functional zones. Hydrological forcing does not erase these zones but alters the predominant process, driving downstream branch-wide (Kennet-influenced, middle mixing corridor, and Thames-influenced branches) switches from source to sink dominated regimes. Because confluences integrate signals from contrasting tributary sub-catchments, this approach provides a transferable workflow for translating high-resolution synoptic patterns into process-based diagnostics that complement fixed-station monitoring and help locate source- versus sink-dominant behaviour within river catchments.