Evaluation of Nonequilibrium Models for Predicting Atrazine Transport in Soils

Model evaluations are commonly undertaken using relatively sparce data, and the criteria for such evaluations are often not fully addressed. For instance, less effort has been made to evaluate models based on their prediction capability. In this study we compared several chemically and physically based nonequilibrium models for their predictions of atrazine (2‐chloro‐4‐ethylamino‐6‐isopropylamino‐ s ‐triazine) transport in a uniformly packed Sharkey soil (Vertic Haplaquept). The models compared were a nonlinear multireaction model (MRM), a second‐order model (SOM), and their coupled versions with a mobile‐immobile model (MIM), MRM‐MIM and SOM‐MIM. Model parameters were derived from kinetic batch experiments and were used to predict atrazine column breakthrough curves (BTCs) for different pore water velocities (0.52–2.17 cm h −1 ), aggregate sizes (0–2, 2–4, and 4–6 mm), flow interruptions (0, 4, and 16 d), and soil column lengths (10 and 15 cm). For each model, nine formulations reflecting various adsorption sites were evaluated. Based on root mean squared errors and sums of squared errors, considerable improvement in atrazine BTC predictions was achieved using SOM compared with MRM. Furthermore, SOM was more sensitive to different model formulations than MRM. Coupled MRM‐MIM did not improve atrazine BTC prediction. Although goodness of model predictions varied with individual columns, we concluded that SOM‐MIM with four parameters and a time‐dependent mass transfer coefficient (α) best described atrazine retention and transport in Sharkey soil.