Are our dynamic water quality models too complex? A comparison of a new parsimonious phosphorus model, SimplyP, and INCA‐P. Issue 7 (4th July 2017)

• Record Type: Journal Article
• Authors: Jackson‐Blake, L. A.; Sample, J. E.; Wade, A. J.; Helliwell, R. C.; Skeffington, R. A.
• Is Part Of: Water resources research. Volume 53:Issue 7(2017)
• Page Start: 5382
• Page End: 5399
• Publication Date: 2017-07-04
• Subjects: model -- complexity -- simplicity -- phosphorus -- dynamic process‐based -- water quality; Hydrology -- Periodicals; 333.91
• DOI: 10.1002/2016WR020132
• Languages: English
• ISSNs: 0043-1397

Abstract

Abstract: Catchment‐scale water quality models are increasingly popular tools for exploring the potential effects of land management, land use change and climate change on water quality. However, the dynamic, catchment‐scale nutrient models in common usage are complex, with many uncertain parameters requiring calibration, limiting their usability and robustness. A key question is whether this complexity is justified. To explore this, we developed a parsimonious phosphorus model, SimplyP, incorporating a rainfall‐runoff model and a biogeochemical model able to simulate daily streamflow, suspended sediment, and particulate and dissolved phosphorus dynamics. The model's complexity was compared to one popular nutrient model, INCA‐P, and the performance of the two models was compared in a small rural catchment in northeast Scotland. For three land use classes, less than six SimplyP parameters must be determined through calibration, the rest may be based on measurements, while INCA‐P has around 40 unmeasurable parameters. Despite substantially simpler process‐representation, SimplyP performed comparably to INCA‐P in both calibration and validation and produced similar long‐term projections in response to changes in land management. Results support the hypothesis that INCA‐P is overly complex for the study catchment. We hope our findings will help prompt wider model comparison exercises, as well as debate among the water quality modeling community as to whether today's models are fit for purpose. Simpler models such as SimplyP have the potential to be useful management and research tools, building blocks for future model development (prototype code is freely available), or benchmarks against which more complex models could be evaluated. Plain Language Summary: Catchment models may be useful tools for managing water quality, for example for exploring how water quality may change in the future under different land management, land use or climate. However, models are only useful if they capture the right processes, otherwise there is a risk of management decisions being based on unreliable information. There is now a growing awareness that many catchment water quality models used today are too complex. This makes it difficult, time‐consuming and expensive to set models up, and reduces the reliability of their predictions. We have therefore developed a new, simple model to predict phosphorus concentrations in rivers, one of the biggest causes of troublesome algal blooms in fresh waters. The simple model was compared with one of the standard models in common use and was found to perform as well, despite being substantially simpler to set up and use. This supports the idea that current water quality models are too complex, and that modelers need to put more effort into assessing whether they are using appropriate tools. Key Points: We developed a new parsimonious dynamic catchment phosphorus model, SimplyP SimplyP performed as well in calibration, validation, and scenarios as a well‐established, substantially more complex model Results support the hypothesis that water quality models are too complex and suggest wider simplification exercises may be warranted