Advancing understanding of in-river phosphorus dynamics using an advection–dispersion model (ADModel-P)

Abstract

• Advance the understanding of in-stream phosphorus transformation processes. • Use easily attainable field data to generate empirical relations of process rates. • Extensive scope for transferability to other rivers. • At which times rivers are net sources or sinks and which processes might dominate. • Adsorption-desorption and resuspension influence investigated phosphorus the most. The objectives of the present research are (1) to predict phosphorus compounds transport along river stretches at high spatio-temporal resolution by developing an original approach based on advection–dispersion modelling (ADModel-P); (2) to advance the understanding of in-stream phosphorus transformation processes, and (3) to explore their relation to controlling factors (water temperature, seasonality and water flow). For a case study of the River Swale (UK) modelling results, in agreement with results based on experimental data, show that resuspension is the largest contributor to the variability of organic phosphorus, while adsorption–desorption are the largest contributors to the variability of soluble reactive phosphorus. Additionally, simulations reveal that conversion of inorganic to organic forms is important. In-channel sinks appear more important than sources for Soluble Reactive Phosphorus (SRP) during 80% of the time, while there is no clear evidence that Organic Phosphorus (OP) sinks or sources are dominant except the beginning of spring (around 20% of the total time). The findings are valuable because they advance knowledge regarding: (1) which in-stream processes are important; (2) values associated to transformation rates for mineralization, sedimentation, resuspension, uptake, adsorption – desorption; (3) at which times rivers are net sources or sinks and which source/sink processes might be dominant. ADModel-P is a robust model which has the benefits of simple field data requirements (compared to more complex models) and less assumptions (e.g. compared to simple models assuming perfect mixing in reaches) but without the drawbacks of lack of process representation to enable confidence in predictions to change. There is extensive scope for transferability to other rivers: rate constants can be estimated from easily attainable information on water temperature, seasonality and water flow.