Estimating biogeochemical rates using a computationally efficient Lagrangian approach

Abstract

Nutrient concentrations in many estuaries have increased over the past century due to increases in wastewater discharge and increased agricultural intensity, contributing to multiple environmental problems. Numerous biogeochemical and physical processes in estuaries influence nutrient concentrations during transport, resulting in complex spatial and temporal variability and challenges identifying predominant processes and their rates. Mechanistic models which require these rates to quantify biogeochemical processes become complex and difficult to calibrate as the number of processes and parameters grows, owing to the high dimensionality of the parameter space and the computational cost of simultaneously modeling the transport and transformations of constituents. We developed a modeling approach that decouples transport from transformations, enabling fast, data-driven exploration of the parameter space. The approach extracted information including water age, cumulative exposure to specific habitats, and mean water depth exposure from a hydrodynamic model. Using this information, a biogeochemical model was implemented to predict ammonium and nitrate concentrations in a Lagrangian frame. The model performed each simulation in milliseconds on a laptop computer, allowing the fitting of rate parameters for key transformations by optimization. The optimization used fixed station nitrate observations and the model was then validated against high-resolution mapping observations of ammonium and nitrate. The results suggest that the observed spatial and temporal variation can be largely represented with five transformation processes and their associated rates. Dissolved inorganic nitrogen (DIN) losses occurred only in shallow vegetated areas in the model, highlighting that biogeochemical processes in these areas should be included in DIN models.