High-frequency, in situ sampling of field woodchip bioreactors reveals sources of sampling error and hydraulic inefficiencies.

Woodchip bioreactors are a practical, low-cost technology for reducing nitrate (NO 3) loads discharged from agriculture. Traditional methods of quantifying their performance in the field mostly rely on low-frequency, time-based (weekly to monthly sampling interval) or flow-weighted sample collection at the inlet and outlet, creating uncertainty in their performance and design by providing incomplete information on flow and water chemistry. To address this uncertainty, two field bioreactors were monitored in the US and New Zealand using high-frequency, multipoint sampling for in situ monitoring of NO 3 –N concentrations. High-frequency monitoring (sub hourly interval) at the inlet and outlet of both bioreactors revealed significant variability in volumetric removal rates and percent reduction, with percent reduction varying by up to 25 percentage points within a single flow event. Time series of inlet and outlet NO 3 showed significant lag in peak concentrations of 1–3 days due to high hydraulic residence time, where calculations from instantaneous measurements produced erroneous estimates of performance and misleading relationships between residence time and removal. Internal porewater sampling wells showed differences in NO 3 concentration between shallow and deep zones, and "hot spot" zones where peak NO 3 removal co-occurred with dissolved oxygen depletion and dissolved organic carbon production. Tracking NO 3 movement through the profile showed preferential flow occurring with slower flow in deeper woodchips, and slower flow further from the most direct flowpath from inlet to outlet. High-frequency, in situ data on inlet and outlet time series and internal porewater solute profiles of this initial work highlight several key areas for future research. Image 1 • In situ, short term monitoring of nitrate removal for two field woodchip bioreactors. • High frequency measurements reduced uncertainty of performance. • Misleading relationships between flow and removal caused by lagged nitrate peaks. • Vertical and longitudinal gradients of nitrate, organic carbon, and dissolved oxygen. • High frequency solute tracking revealed zones where preferential flow was occurring. [ABSTRACT FROM AUTHOR]