|Wauchope, Robert - Don|
|Chandler, Laurence - Larry|
Submitted to: Agricultural Systems
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/25/1997
Publication Date: N/A
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. This paper reports on application of a new model -- RZWQM, or the Root Zone Water Quality Model -- developed by a team of ARS researchers to take advantage of two decades of research in agricultural hydrology and agricultural chemical environmental fate. We compared the model's performance at predicting runoff from the surface of a field when compared with two year's actual data, including an intensive series of simulated rainfall "events". The results showed that the model is capable of excellent predictions, but it is more complex and sensitive to local conditions than older models. Th research also revealed shortcomings in the model that will require modification.
Technical Abstract: The capability of a water quality model to correctly predict the partitioning of rainfall between infiltration and runoff is fundamental to its predictions of water contamination by chemicals. We evaluated the new model RZWQM (v3.1) -- the ARS Root Zone Water Quality Model -- for its performance as a predictor of surface runoff as compared to measured runoff fin a carefully controlled field site. Model-required minimum inputs of soil properties were supplied and the other inputs were estimated by the model based on an extensive soil properties database. In 1992 and 1993, two 15 m X 45 m corn field plots with 3% slope on Tifton loam sand received 6 severe simulated rainfall events each year with each event consisting of a 2.5 cm/h rainfall for 2 h. Runoff was monitored continuously using an edge of field trough and flume. Measured runoff from each event was compared with that predicted by the model in both uncalibrated and calibrated modes for daily and annual values. Model performance criteria included graphical comparison and statistical analyses including means, ratios of means, root mean square error (RMSE), and paired difference t-tests. The uncertainty in measuring saturated hydraulic conductivity accounted for the majority of the uncertainty in predicted runoff - the uncertainty in soil spatial variability caused by tractor wheel tracks and surface crusts/seals made significant contributions to predicted runoff as well. Simulations of tillage managements and crusts/seals by RZWQM enabled the model to effectively predict runoff over a wide range of conditions of field studies. With minimum inputs of the soil properties measured on site, RZWQM generally under-predicted runoff in uncalibrated mode, but when calibrated to the site RZWQM could accurately (within 20%) predict runoff.