Resolving Sequential Storm-Driven Pollutant Pulses using Novel Hydrochemical Measurements within a Reactivity-Hydrodynamic workflow.

Within river deployments that combine Acoustic Doppler Current Profiler (ADCP) hydrodynamics with high-frequency water quality sensing now offer unprecedented detail regarding instream physical processes, chemical mixing, and nutrient transformations, but a key barrier is translating complex, high-volume datasets into process-based interpretations.  Here, we present the Reactivity Index – Hydrodynamic Index (RI - H) workflow, an approach that combines standard hydrodynamic and water-quality sensor data into diagnostic behavioural classes describing hydrochemical behaviours, demonstrated at the Kennet-Thames confluence (Reading, UK). 

The workflow was developed and tested using a remote-controlled moving-boat platform (ArcBoat) equipped with a SonTek M9 (ADCP) and a YSI EXO2 multiparameter sonde.  We collected simultaneous near-surface measurements of velocity, nutrients (NH₄⁺-N, NO₃⁻-N), fluorescent dissolved organic matter (fDOM), turbidity, and specific conductivity across a quasi-synoptic transect design (upstream controls, repeated cross-sections, and diagonal transects) on three days. The study reach was segregated into 14 spatial zones to monitor the chemical and physical changes from upstream end-members (of the Rivers Kennet and Thames), how the end-members interact and evolve downstream of the confluence, and the shift in chemical and physical behaviour under different hydrological conditions. 

RI and H were derived from solute concentrations and flow velocities. Plotting the two indices enabled each observation to be classified into one of five process-based behavioural categories (e.g., Retentive, Reactive, Low-energy depletion, Attenuating, and Conservative). Across three campaigns (including a rainfall-impacted survey), zonal contrasts in RI were consistently strong (Kruskal–Wallis ε² = 0.28–0.81; p < 0.0001), identifying zones with distinct behavioural signatures (nitrate reaction or ammonium retention zones). Extending the same logic to turbidity yielded complementary particulate-transport classes (Local Input, Advective Input, Sediment Deposition, Advective Dilution and Conservative mixing), demonstrating that the workflow was applicable for solutes and particulates. 

Our high-frequency transect sampling captured the hydrological and biogeochemical response to sequential rainfall events on 26 February 2025. Following morning rainfall, we identified a pollutant pulse characterised by elevated NH₄⁺ and fDOM, indicative of sewage or wastewater influence in the River Kennet, which diluted progressively downstream. A late-afternoon high-intensity rain and hail event triggered a distinct second wave, marked by a sharp spike in nitrate NO₃^- and turbidity, characteristic of surface run-off. The rapid succession of these pulses reveals differing pollutant sources and pathways activated under varying rainfall intensities. Statistically strong spatial contrasts in reactivity persisted even during this dynamic event (Kruskal-Wallis ε² > 0.28 - 0.76 for all solutes). This outcome demonstrates the workflow can resolve within-event functional shifts, translating sensor data into a real-time diagnostic of a river's response to rainfall. The RI – H framework provides a standardised approach by enabling event-scale diagnosis of solute and sediment behaviour that cannot be resolved by fixed or point-based monitoring alone. By classifying how rivers transport and process materials in space and time using deployable sensors, the workflow offers a diagnostic, process-informed water quality assessments relevant to better understanding pollutant dispersal, chemical transformations and biota in fluvial systems.