1989 - 1994
Predicting Pesticide Migration Through Soils
R. de Jong, W.D. Reynolds, S.R. Vieira
and R.S. Clemente
Download Report (1500 KB pdf) [graphs & tables included]
Pesticide contamination of ground water has traditionally been considered to be due primarily to spills, and to improper storage, disposal and application practices. There is increasing evidence, however, that normal agricultural practices can also result in low-level, non-point source contamination of ground water via the downward migration of pesticides through the soil profile (Agriculture Canada, 1990). Although this type of contamination is usually well below Canadian drinking water guidelines, there are growing public concerns over potential health hazards related to long-term exposure to low levels of pesticides (Agriculture Canada, 1990). Consequently, there is a need to determine how important and widespread low-level non-point source pesticide contamination of ground water might be; what the controlling soil, land use and weather factors are; and which agricultural practices are required to mitigate and control this type of pollution at acceptable and sustainable levels.
Essential steps in obtaining this information include identification of the primary mechanisms controlling pesticide movement through the soil profile, and development of the capability to characterize and predict the pesticide movement in space and time with acceptable accuracy. Accordingly, the objectives of this study are: (i) to develop a methodology for predicting, characterizing and quantifying pesticide migration rates through the soil profile, and (ii) to conduct a test and evaluation of the methodology by applying it to atrazine migration through the soil profiles of the Grand River watershed in Southern Ontario, Canada.
The herbicide atrazine was chosen for study because it is the most commonly used pesticide for corn production in Southern Ontario; and because atrazine residues have been detected in the ground waters of many agricultural watersheds in the Great Lakes basin, particularly where there is some combination of high atrazine usage, intensive agriculture, high precipitation, coarse textured and other highly permeable soils, high water tables, and sloping topography (Millette and Torreiter, 1992). The watershed scale was chosen (rather than the field scale, for example) because it is a natural landscape unit; because the most readily available and complete soil and weather databases are applicable on the watershed scale; and because the watershed provides a convenient basis for estimating agrochemical loadings to the Great Lakes. The Grand River watershed (Fig. 1) was chosen as a test case because it is one of the largest in Southern Ontario ( 680,000 ha); it contains a large range of soil textures with a complexity of distribution that is typical for the region; the primary land use is field crop production using standard agricultural practices and "normal" rates of pesticide usage (Shelton et al., 1988); and it empties directly into the Great Lakes (Lake Erie). These features make this watershed ideal for testing and demonstrating the methodology. In addition, the results obtained for the Grand River watershed should be "characteristic" of the entire Southern Ontario region.
Ground water contamination in this study is defined as non-zero values of predicted annual mass loading of pesticide to the 90 cm depth in the soil. Non-zero mass loading was used, rather than pore water concentrations above a specified threshold, because of the need to estimate the quantities and distributions of all pesticide additions to ground water, not just the "high level" additions. The 90 cm depth was selected because it reflects the mean tile drain depth in Southern Ontario, as well as the primary rooting depth for most field crops. It was assumed that if a pesticide reached this depth, it would not be intercepted by tiles and roots, but continue to percolate downward and eventually enter the ground water.
ASSESSMENT OF THE GRAND RIVER WATERSHED PREDICTIONS
A comprehensive assessment of the accuracy and validity of the predictions is not possible due to a lack of appropriate field measurements. There are, however, sufficient field data available to get a general indication of the plausibility of the predictions; as well as an indication of the sensitivity of the predictions to the quality and quantity of input data.
i) Comparison to Ground Water Survey Data
ii) Effect of Scale and Missing Data
Although the LEACHM - pedotransfer function - geostatistics - GIS methodology is still under development, the results of this study are encouraging. The required input data was extractable, or derivable (via pedotransfer functions) from information archived in the NSDB and AWDB databases. The pedotransfer function, kriging and ILWIS manipulations were effective and sufficiently robust to accommodate small map scales and a high percentage of missing data. Predictions of potential ground water contamination by atrazine for the Grand River watershed are plausible and compare favourably with recent ground water survey results. It is consequently felt that this methodology will prove very useful in the development of agricultural practices and guidelines that maintain agrochemical inputs to the groundwater at acceptable and sustainable levels.
NEW TECHNOLOGIES AND BENEFITS
A new methodology, which consists of solute transport modelling in combination with geostatistical analyses and a geographic information system, was developed for predicting, characterizing and quantifying non-point source contamination of ground water due to the migration of agrochemicals through the soil profile. The methodology can be used to develop inventories of "pollution potential" and to develop agricultural land use practices and guidelines that mitigate and control agrochemical inputs to the ground water at acceptable and sustainable levels. It can also be used to predict the impact of changes in land management practices and land use, using so-called "what if" scenarios. The methodology can be applied to a range of agrochemicals, and to virtually any landscape unit (e.g. plot, field, watershed, region), providing the necessary data are also available at that scale.
IMPLICATIONS FOR THE GREAT LAKES ECOSYSTEM
Predictions were made of atrazine migration through the soil profiles of a "representative" agricultural watershed within the Great Lakes Basin (the Grand River watershed, Southern Ontario). The results indicate that potential non-point atrazine contamination of ground water is highly variable spatially, generally low level, and determined by complex interactions among several soil, weather, crop and solute transport factors. Continuous corn cropping with annual applications of atrazine is predicted to produce a low, but steady, loading of atrazine to the ground water, and hence to the Great Lakes.
TECHNOLOGY TRANSFER POTENTIAL
Although the solute transport modelling-geostatistics-geographic information system methodology still requires further development and testing, the preliminary results indicate that it should be useful for: pesticide licensing and usage guidelines; land use planning and management; and the development of agricultural practices that control agrochemical pollution of ground water. Consequently, potential users of the methodology include scientists, planners, managers and policy makers.
REQUIREMENTS FOR FURTHER DEVELOPMENT AND TESTING
Several important factors were not considered in the Grand River watershed application, including land use patterns, crop rotations, annual variation in water table depth, topography, and the simultaneous transport of several agrochemicals and metabolites. These factors are likely to be important in the Great Lakes Basin, and should be taken into account. Most watersheds, especially those in the Great Lakes Basin, have substantial non-agricultural areas (e.g. 25 % of the Grand River watershed is used for non-agricultural activities) and this will obviously affect the amounts and distributions of agrochemical inputs to the ground water. Land use and crop rotations not only affect water movement and water content distributions in the soil profile (through crop water use), but also determine the type, amount, timing and frequency of application of fertilizers and pesticides. In humid regions, the depth to the water table can vary from virtually zero at spring thaw to 3 m or more in late summer.
Thus, the distance agrochemicals must travel to enter the ground water varies substantially throughout the year. Run-off and run-on of water, solutes and sediment due to variations in topography have a strong impact on the amount and spatial distribution of water and agrochemical entry into the soil. Any particular agricultural practice (e.g. continuous corn cropping) is likely to contribute a variety of agrochemicals and metabolites to the ground water (e.g. fertilizer nitrate, atrazine plus its main metabolite d-ethyl atrazine, metolachlor, etc.), rather than a single chemical. Except for topography, the methodology in its present form can account for all of the above factors through adjustments and additions to the various input data files. A run-off - run-on based routine that accounts for topography has not been developed.
The representation of "bypass flow" of solute in LEACHP (Eq. 5-7, Appendix A) is simplistic and may be inadequate in soils where extensive bypass flow occurs (e.g. see Table 8, Woodslee field site, Reynolds et al., 1994). Improved, soil property based, representations of bypass flow should be developed and added to LEACHP so that early arrival of agrochemicals to the water table can be detected.
The laboratory column studies suggest that the current form of LEACHP may overestimate the concentration of atrazine in solution, possibly due to an underestimate of effective atrazine dissipation rates (see Reynolds et al., 1994 for details). Further investigations of pesticide - soil interactions should therefore be conducted so that more accurate representations of pesticide transformation and dissipation can be incorporated into LEACHP. Only a very cursory assessment of the accuracy and uncertainty of the predictions has thus far been attempted. Major sources of uncertainty that require further investigation include:
i) NSDB database
ii) Accuracy and precision of the pedotransfer functions.
iii) Use of the dominant soil type in the landscape polygons.
iv) Effect of map scale.
Sunday, August 27, 2017 09:57:57 AM