Critical evaluation of models used to study agricultural phosphorus and water quality

• Published in: Soil use and management 2013-03, Vol.29 (s1), p.36-44
• Main Authors: Vadas, P. A., Bolster, C. H., Good, L. W.
• Format: Article
• Language: English
• Subjects: AGNPS model; Agricultural management; Agronomy. Soil science and plant productions; algorithms; Biological and medical sciences; cattle manure; Computer simulation; desorption; environmental protection; equations; Experiments; fertilizers; Fundamental and applied biological sciences. Psychology; grazing; inorganic phosphorus; leaching; Management; manure spreading; Mathematical analysis; Mathematical models; model validation; modelling; nonpoint source pollution; Phosphorus; runoff; soil; Soil (material); soil nutrients; Soil science; Studies; United States; Water quality
• ISSN: 0266-0032; 1475-2743
• DOI: 10.1111/j.1475-2743.2012.00431.x

There is an increasing demand to evaluate all agricultural systems where nonpoint source phosphorus (P) pollution is a priority. Because experiments cannot evaluate all possible interactions between management and natural processes that impact P in the environment, computer models are necessary. This article uses research examples to discuss issues related to integrating experimental and model development research. Model development often follows an evolution from a perceptual model of qualitative understanding to a conceptual model of equations to a procedural model of computer code. Integrating experiments and models is an efficient way to show if conceptual models are incorrect and design hypothesis‐driven experiments to develop alternative models. A model can be perceptually correct but conceptually incorrect, such as some P Indexes in the US. Translation of conceptual equations into procedural model equations can vary between models, such as how the same soil P model varies in EPIC, SWAT and AnnAGNPS. Commonly used P models often do not reflect current science, such as in simulating soil total P and inorganic P sorption and desorption, surface‐applied manure and fertilizer and direct P loss in run‐off, grazing cattle dung dynamics and contributions to P in run‐off, and subsurface P leaching and transport. Thus, models may be applied in situations for which their algorithms are inadequate. Validating models using more than one output parameter, over multiple years and with diverse scenarios can help identify model weaknesses. Ultimately, a framework of integrated experimentation and model development can advance agricultural P science and environmental protection beyond the point that either can achieve alone.