Modelling the Fate of Molinate in Rice Paddies of South Eastern Australia Using RICEWG

Abstract

Contamination of drainage channels and creeks with pesticides used in rice production is of concern in south eastern Australia. Of major concern is the herbicide molinate that is detected in over 25% of water samples. This pesticide has been the focus of researchers and environmental protection authorities due to continuing frequent detection off farm despite improved application methods and water management guidelines. The objective of this study was to assess the rice pesticide model RICEWQ version 1.7.2 for its applicability in simulating pesticide in runoff in south eastern Australia. The model was successfully calibrated against field data on water depths and molinate concentrations from a rice field in the Murrumbidgee Irrigation Area. It was found that the calibrated model was able to simulate the field data in the supply bay adequately; however it is not capable of modelling rice fields with multiple bays, which are much more complex than a single bay situation. Sensitivity analyses of the parameter values on molinate concentrations in ponded water, sediment and foliage were performed. Overall the application efficiency has a major impact and this impact is carried throughout the entire simulation. In ponded water the bulk density, mixing velocity, release rate for slow release formulation, pesticide solubility and water/sediment partition coefficient were relatively sensitive. In the sediment the release rate and the mixing , soil bulk density, degradation rate in the sediment, water/sediment partition coefficient and mixing velocity have large sensitivities. On the foliage only three parameters have non-zero sensitivities, the application efficiency, the wash off coefficient and the degradation rate on foliage. The calibrated model was used to investigate water and pesticide management for a single bay. It was found that water management was critical to minimising molinate runoff. Using a 41 year weather sequence for Griffith in the Murrumbidgee Irrigation Area it was found that if water levels were maintained 5 cm below the drainage outlet there was little likelihood of surface runoff occurring. Simulation of the registered label application methods and rates for molinate were undertaken. These compared application onto a dry bay, a ponded bay and application by ground rig, aerial, and Soluble Chemical Water Injection In Rice Technique (SCWIIRT) low pressure system. The greatest maximum concentrations of molinate in the ponded water occurred when molinate was applied directly onto the water. The maximum concentrations for application onto a dry bay were an order of magnitude lower than for the applications onto a bay filled with water. However, the pesticide concentrations in water declined more rapidly for the application onto a water filled bay than for application onto a dry bay. Field trials are required to assess the accuracy of these results as no data comparing ponded water and dry bay applications is available. The comparison of application methods was undertaken by adjusting the application efficiency parameter. This ranged from 60 % (assumed) for the aerial application on dry bay, to 70 % (assumed) for the ground rig, 95 % for aerial application, determined from the model calibration, and 100 % (assumed) for the SCWIIRT. The results showed that increasing the application rate by 60 % increased the period during which the water molinate concentration was above guideline level by 11 %. The results indicate that the application amount is only critical to the concentration of molinate in runoff if it occurs in about 30 days after application. The results regarding molinate concentrations in water with time and effects of different application rates suggest that poor application efficiency results in a major loss of chemical. If the application efficiency could be improved and application aimed at a target concentration then lower application rates of molinate could potentially be as effective as current label rates. This requires further research.