Reducing equifinality using isotopes in a process‐based stream nitrogen model highlights the flux of algal nitrogen from agricultural streams

Abstract

The fate of bioavailable nitrogen species transported through agricultural landscapes remains highly uncertain given complexities of measuring fluxes impacting the fluvial N cycle. We present and test a new numerical model named Technology for Removable Annual Nitrogen in Streams For Ecosystem Restoration (TRANSFER), which aims to reduce model uncertainty due to erroneous parameterization, i.e., equifinality, in stream nitrogen cycle assessment and quantify the significance of transient and permanent removal pathways. TRANSFER couples nitrogen elemental and stable isotope mass‐balance equations with existing hydrologic, hydraulic, sediment transport, algal biomass, and sediment organic matter mass‐balance subroutines and a robust GLUE‐like uncertainty analysis. We test the model in an agriculturally impacted, third‐order stream reach located in the Bluegrass Region of Central Kentucky. Results of the multiobjective model evaluation for the model application highlight the ability of sediment nitrogen fingerprints including elemental concentrations and stable N isotope signatures to reduce equifinality of the stream N model. Advancements in the numerical simulations allow for illumination of the significance of algal sloughing fluxes for the first time in relation to denitrification. Broadly, model estimates suggest that denitrification is slightly greater than algal N sloughing (10.7% and 6.3% of dissolved N load on average), highlighting the potential for overestimation of denitrification by 37%. We highlight the significance of the transient N pool given the potential for the N store to be regenerated to the water column in downstream reaches, leading to harmful and nuisance algal bloom development.

Nitrate Pathways, Processes, and Timing in an Agricultural Karst System: Development and Application of a Numerical Model

“…While the magnitude of flux from South Elkhorn (2.10 ± 0.66 t N·km −2 ·yr −1 ) and Cane Run (1.85 ± 0.21 t N·km −2 ·yr −1 ) are comparable, the dynamics controlling the timing of flux vary considerably. At coarse resolution (Figure a), the karst‐dominated system of this study and fluvial‐dominated karst system in Ford et al () exhibit high similarity in terms of their mean temporal trend. The similarity is reasonable given that the agricultural land surfaces in the two systems apply similar nutrient treatments seasonally and that the two systems experienced the same rainfall distributions.…”

“…However, by including only hydrologic sets that also produce satisfactory N model results and represent behavior processes, the remaining sets ( n = 419) indicate that 42% of water is discharged by epikarst, 39% by phreatic zone, and 19% by quick flow. The utility of multiple response variables to reduce equifinality has been noted in other systems such as in surface streams using stable isotopes (Ford et al, ; Ford & Fox, ), in watershed‐scale models using remote‐sensing data (Silvestro et al, ), and in vegetation zones using carbon data representative of different time scales (Carvalhais et al, ). We add to this list with an application of equifinality reduction to water flow dynamics in an agricultural karst system using an N data set and numerical modeling.…”

“…Numerical modeling uncertainty analysis was performed on an institutionally shared high‐performance computing cluster (DLX2/3) with 4800 processor cores, 18 TB of RAM, and 1 PB of high‐speed disk storage. Uncertainty in the hydrologic and N models was assessed with the generalized likelihood uncertainty estimation method (Figure ), which has been applied to water resources modeling to assess the equifinality of model parameter sets (Beven, ; Ford et al, ). The generalized likelihood uncertainty estimation methodology is initiated by assuming a prior distribution (e.g., uniform) for model parameters.…”

“…Timing of N exports from karst was analyzed to assess temporal delivery of NO 3 − to downstream waterbodies. We performed an analysis using results from a recent fluvial‐dominated, immature karst study in Ford et al () to provide comparison with karst‐dominated NO 3 − results from the present system. The South Elkhorn (62 km 2 ) drains southwest Lexington, KY, while the Cane Run system (58 km 2 ) drains the northern portion of the city.…”

“…While the reservoir approach is attractive, issues with model suitability and equifinality (defined as the existence of multiple “acceptable” representations of an environmental system; Beven, ) must be addressed through rigorous evaluation of the model's process‐representation capability (Hartmann et al, , ; Hartmann, ). Additionally, equifinality and uncertainty can be reduced through coupling multiple data streams (e.g., Ford et al, ). Therefore, in this paper, we develop a robust reservoir modeling approach for karst water and N capable of representing field processes and data streams.…”