GLEAMS, Opus, PRZM2β, and PRZM3 Simulations Compared with Measured Atrazine Runoff

Abstract

High‐intensity storms that occur shortly after chemical application have the greatest potential to cause chemical runoff. We examined how effectively current chemical transport models GLEAMS, Opus, PRZM2β, and PRZM3 could predict water runoff and runoff losses of atrazine [6‐chloro‐N‐ethyl‐N′‐(1‐methylethyl)‐1,3,5‐triazine‐2,4‐diamine] under such conditions, as compared with observations from a controlled field runoff experiment. The experiment was conducted for 2 yr using simulated rainfall on two 14.6‐ by 42.7‐m plots within a corn (Zea mays L.) field on Tifton loamy sand (fine‐loamy, kaolinitic, thermic Plinthic Kandiudults) under conventional tillage practices. For each plot‐year, atrazine was applied as surface spray immediately after planting and followed by a 50‐mm, 2‐h simulated rainfall 24 h later. A similar preapplication rainfall and four subsequent rainfalls during the growing season were also applied. Observed water runoff averaged 20% of the applied rainfall. Less runoff occurred from freshly tilled soil or under full canopy cover; more runoff occurred when nearly bare soil had crusted. Observed total seasonal atrazine runoff averaged 2.7% of that applied, with the first posttreatment event runoff averaging 89% of the total. GLEAMS, Opus, PRZM2β and PRZM3 adequately predicted water runoff amounts, with normalized root mean square errors of 29, 29, 31, and 31%, respectively. GLEAMS and PRZM3 predicted atrazine concentrations in runoff within a factor of two of observed concentrations. PRZM2β overpredicted atrazine concentrations. Opus adequately predicted atrazine concentrations in runoff when it was run with an equilibrium adsorption submodel, but significantly underestimated atrazine concentrations when it was run with a kinetic sorption submodel.