Multivariate calibration of the agricultural policy/environmental eXtender model for field scale simulation of hydrologic and agronomic outcomes

Farmers need information on the long-term effects of agricultural practices on crop yields and water balance. Models can be used to estimate these outcomes but parameter calibration with observed data is critical for accuracy of model outputs. Remote sensed surface soil moisture and evapotranspiration are becoming available, potentially enabling hydrological calibration when in-field data are lacking. Since the accuracy of these remote sensed data has not yet been assessed at the field scale, the objective of this study was to use combinations of locally measured soil moisture, evapotranspiration, crop yields and surface runoff to calibrate the Agricultural Policy/Environmental Extender (APEX) model for these variables and assess model performance. APEX parameters were calibrated for a 36-ha field in Missouri, United States, a spatial scale far smaller than those used in past multivariate calibration studies. Model performance was highest for outcomes included in the calibration process. However, the model struggled to simultaneously optimize crop yields and runoff performance, possibly due to limitations in the model's implementation of crop failure. Including soil moisture or evapotranspiration with runoff in the calibration process improved their respective performance but did not significantly change runoff performance. Calibration with evapotranspiration (excluding runoff) outperformed other non-runoff calibration combinations, implying that remotely sensed evapotranspiration data could help expand the spatial coverage of field-scale agricultural modeling in a region. This would allow the model to be applied more widely and to provide useful estimates of management outcomes to a larger number of US farmers.