|Wauchope, Robert - Don|
|Chandler, Laurence - Larry|
Submitted to: Transactions of the ASAE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/6/1997
Publication Date: N/A
Citation: Interpretive Summary: Computer simulation models have become necessary tools used for estimating the risk of water pollution by pesticides and fertilizers. They are the only means of allowing decision-makers to examine all together the multitude of interacting factors -- of weather, soil and crop situations, agricultural practices and pesticide properties -- that determine the pollution potential of a given pesticide or fertilizer use. These models are complex, and only as good as they make realistic predictions. This paper reports a comparison of water runoff (loss of surface water from edge of the field) predictions of three models with field data from a controlled simulated-rainfall experiment conducted at Tifton in 1992-1994. Two of the models are well established and one is a newer model with more sophisticated programming. The results showed that the newer model was generally more accurate, but at a cost of more complex user input requirements.
Technical Abstract: We compared GLEAMS, OPUS, and PRZM-2 model runoff predictions with runoff measured in a carefully controlled field site used for chemical runoff studies. In 1992 and 1993, two 15 m X 45 m corn field plots with 3% slope on Tifton Loamy sand (fine-loamy, siliceous, thermic Plinthic Kandiudult) received 24 severe simulated rainfall events, each consisting of a 5-cm rainfall in 2 h. Measured runoff from these events was compared with that predicted by each model for daily, monthly and annual values. Graphical comparison with an initial moisture condition-II curve number of 85 showed that GLEAMS and OPUS predicted runoff reasonably well while PRZM-2 generally over predicted runoff. Paired difference t-tests with the above curve number showed that at 95% confidence interval there was a significant difference between measured and model-simulated runoff for OPUS and PRZM-2 (P<0.0001), but not for GLEAMS (P=0.809). Means, ratio of means, and RMSE showed that model performance decreases following the order of GLEAMS > OPUS > PRZM-2. All predictions were sensitive to moisture condition-II curve number. OPUS could simulate runoff reasonably well if it was calibrated to the site. Simulation of tillage managements and crusts/seals by OPUS enabled it to effectively predict runoff over a wide range of conditions of field studies. We concluded that GLEAMS and OPUS could effectively simulate runoff in the system while PRZM-2 over predicted runoff.