Application of the DNDC model to tile-drained Illinois agroecosystems: model calibration, validation, and uncertainty analysis

We applied the Denitrification-Decomposition (DNDC) model to a typical corn-soybean rotation on silty clay loams with tile-drainage in east-central Illinois (IL). Model outcomes are compared to 10 years of observed drainage and nitrate leaching data aggregated across the Embarras River watershed. We found that accurate simulation of NO3-N leaching and drainage dynamics required significant changes to key soil physical and chemical parameters relative to their default values. Overall, our calibration of DNDC resulted in a good statistical fit between model output and IL data for crop yield, NO3-N leaching, and drainage. Our modifications to DNDC reduced the RMSE from 9.4 to a range of 1.3-2.9 for NO3-N leaching and from 51.2 to a range of 13-23.6 for drainage. Modeling efficiency ranged from 0.25 to 0.85 in comparison with measured drainage and leachate values and from 0.65 to 1 in comparison with crop yield data. However, analysis of simulation results at a monthly time step indicated that DNDC consistently underpredicted peak drainage events. Underprediction ranged from 50 to 100 mm month(-1) following three extreme precipitation events, a flux equivalent to 0.25-0.5 of the total measured monthly flux. Our simulations demonstrated high interannual variation in nitrate leaching with average annual NO3-N loss of 24 kg N ha(-1), peak annual NO3-N loss of 58 kg N ha(-1) and low annual NO3-N loss of 1-5 kg N ha(-1).