Performance assessment of nitrate leaching models for highly vulnerable soils used in low-input farming based on lysimeter data

P. Groenendijk, M. Heinen, G. Klammler, J. Fank, H. Kupfersberger, V. Pisinaras

    Research output: Contribution to journalArticleAcademicpeer-review

    33 Citations (Scopus)


    The agricultural sector faces the challenge of ensuring food security without an excessive burden on the environment. Simulation models provide excellent instruments for researchers to gain more insight into relevant processes and best agricultural practices and provide tools for planners for decision making support. The extent to which models are capable of reliable extrapolation and prediction is important for exploring new farming systems or assessing the impacts of future land and climate changes. A performance assessment was conducted by testing six detailed state-of-the-art models for simulation of nitrate leaching (ARMOSA, COUPMODEL, DAISY, EPIC, SIMWASER/STOTRASIM, SWAP/ANIMO) for lysimeter data of the Wagna experimental field station in Eastern Austria, where the soil is highly vulnerable to nitrate leaching. Three consecutive phases were distinguished to gain insight in the predictive power of the models: 1) a blind test for 2005–2008 in which only soil hydraulic characteristics, meteorological data and information about the agricultural management were accessible; 2) a calibration for the same period in which essential information on field observations was additionally available to the modellers; and 3) a validation for 2009–2011 with the corresponding type of data available as for the blind test. A set of statistical metrics (mean absolute error, root mean squared error, index of agreement, model efficiency, root relative squared error, Pearson's linear correlation coefficient) was applied for testing the results and comparing the models. None of the models performed good for all of the statistical metrics. Models designed for nitrate leaching in high-input farming systems had difficulties in accurately predicting leaching in low-input farming systems that are strongly influenced by the retention of nitrogen in catch crops and nitrogen fixation by legumes. An accurate calibration does not guarantee a good predictive power of the model. Nevertheless all models were able to identify years and crops with high- and low-leaching rates.
    Original languageEnglish
    Pages (from-to)463-480
    JournalScience of the Total Environment
    Publication statusPublished - 2014


    • long-term experiments
    • hydraulic-properties
    • hydrological models
    • standardized assessment
    • nitrogen dynamics
    • simulation-models
    • organic-matter
    • porous-media
    • drain flow
    • water


    Dive into the research topics of 'Performance assessment of nitrate leaching models for highly vulnerable soils used in low-input farming based on lysimeter data'. Together they form a unique fingerprint.

    Cite this