Integration of a detailed biogeochemical model into SWAT for improved nitrogen predictions—Model development, sensitivity, and GLUE analysis

• Published in: Ecological modelling 2007-05, Vol.203 (3), p.215-228
• Main Authors: Pohlert, T., Huisman, J.A., Breuer, L., Frede, H.-G.
• Format: Article
• Language: English
• Subjects: Animal, plant and microbial ecology; Biological and medical sciences; Denitrification; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Lysimeter; Methods and techniques (sampling, tagging, trapping, modelling...); Mineralisation; N-budget; Nitrification; Sensitivity analysis; Uncertainty estimation
• ISSN: 0304-3800; 1872-7026
• DOI: 10.1016/j.ecolmodel.2006.11.019

In this study, the Soil and Water Assessment Tool (SWAT) was extended with algorithms from a detailed nitrogen turnover model to enhance the model performance with regard to the prediction of nitrogen leaching. The new model, which is further referred to as SWAT-N, includes algorithms for decomposition, growth of nitrifying bacteria, nitrification, nitrificatory as well as denitrificatory N-emissions, N-uptake by plants and N transport due to water fluxes. The model was tested with a lysimeter dataset of a long term fertilisation experiment including crop rotation conducted in Eastern Germany. A regression-based global sensitivity analysis was employed to test the impact of the new implemented parameters on the sensitivity of various model output variables. The rate coefficient for decomposition, the pH-value, and the porous fraction from which anions are excluded were identified as the most important parameters controlling nitrogen leaching and gaseous nitrogen emissions. A generalised likelihood uncertainty estimation (GLUE) was conducted afterwards to calculate conditioned prediction intervals for each simulated time step. A maximum model efficiency after [Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models. Part 1. A discussion of principles. J. Hydrol. 10, 282–290] of 0.4 could be achieved for the simulation of monthly nitrogen leaching. It is concluded, that the implemented algorithms enhance the model performance of SWAT, since the previous SWAT version failed to accurately simulate nitrogen leaching at the investigated site.