Calibrating the Root Zone Water Quality Model

Abstract

AbstractTechniques to reliably calibrate computer models are needed before the models can be applied to help solve natural resource problems. The USDA‐ARS Root Zone Water Quality Model (RZWQM) is a comprehensive simulation model designed to predict hydrologic and chemical response, including potential for ground‐water contamination, of agricultural management systems. RZWQM Version 3.2 was calibrated and evaluated at sites in Iowa, Minnesota, Missouri, Nebraska, and Ohio as part of the Management Systems Evaluation Areas (MSEA) project and at a site near Sterling in northeastern Colorado. Soil horizon description and a description of the physical and hydraulic properties of the soil were required to initialize the model. Calibration for nutrient cycling involved adjusting the model coefficients for mineralization, infiltration, and denitrification. Initial N pool sizes were estimated using medium to long‐term computer simulations. Maximum N uptake rate, plant respiration, specific leaf area, and the effect of age at the time of propagule development and senescence were used to calibrate the plant production and yield component. To match the observed results for soil water, N, and plant growth, an iterative approach for calibrating the model was followed. When done methodically, total biomass estimates were within 5%, yield estimates were within l0%, and N uptake was within 20% of field measurements. Calibration of the C and N dynamics module produced results that were generally within 20 to 50 kg ha−1 of measured values for soil profile NO−3‐N. Independent evaluations of the calibrated model focused on four indicator output variables related to plant growth—total biomass, yield, N uptake, and N in the soil profile. Predictions matched the observed data in most cases. The crop model is very sensitive to plant N content. Even small errors in simulating N uptake levels can result in substantial errors in estimates of yield and total aboveground biomass. The model predicted biomass and yield well on irrigated and most dryland management systems and adequately simulated crop variables at various positions along the landscape.