SCI-GROW Description
On this Page
- Background
- Initial Tier Screening of Pesticides for Ground Water Concentrations Using the SCI-GROW Model
- SCI-GROW: Instructions for Input Value Selection
- Comments on the SCI-GROW Model, Version 1.0
Background
SCI-GROW is a screening model which the Office of Pesticide Programs (OPP) in EPA frequently uses to estimate pesticide concentrations in vulnerable ground water. The model provides an exposure value which is used to determine the potential risk to the environment and to human health from drinking water contaminated with the pesticide. The SCI-GROW estimate is based on environmental fate properties of the pesticide (aerobic soil degradation half-life and linear adsorption coefficient normalized for soil organic carbon content), the maximum application rate, and existing data from small-scale prospective ground-water monitoring studies at sites with sandy soils and shallow ground water.
Pesticide concentrations estimated by SCI-GROW represent conservative or high-end exposure values because the model is based on ground-water monitoring studies which were conducted by applying pesticides at maximum allowed rates and frequency to vulnerable sites (i.e., shallow aquifers, sandy, permeable soils, and substantial rainfall and/or irrigation to maximize leaching). In most cases, a large majority of the use areas will have ground water that is less vulnerable to contamination than the areas used to derive the SCI-GROW estimate. For this reason, it is not appropriate to use SCI-GROW concentrations for national or regional exposure estimates.
Initial Tier Screening of Pesticides for Ground Water Concentrations Using the SCI-GROW Model
This model, hereafter referred to as SCI-GROW Screening Concentration In GROund Water) has the following essential attributes:
-
SCI-GROW only provides a screening concentration: an estimate of likely ground-water concentrations if the pesticide is used at the maximum allowable rate in areas with ground water exceptionally vulnerable to contamination. In most cases, a large majority of the use area will have ground water that is less vulnerable to contamination than the areas used to derive the SCI-GROW estimate.
-
Generally, the SCI-GROW estimate should not be multiplied by a factor to provide a "conservative" estimate. The SCI-GROW value is usually only likely to be exceeded under exceptional circumstances in a small percentage of the use area (unless, for example, the pesticide is used only in areas with sandy soils and high rainfall or irrigation). Excluding Florida, no more than a few percent of U.S. agricultural land consists of sandy soils overlying shallow ground water like those selected for reasonable worst case small-scale prospective ground-water monitoring studies (source: STATSGO soils database).
The SCI-GROW estimation procedure cannot currently be adjusted (for example, divided by a factor) to estimate a more realistic exposure level for people deriving their drinking water sources from ground water that is not especially vulnerable to contamination. It is anticipated, however, that in the future, if and when there is availability of the appropriate type of monitoring data, a similar regression model could be developed to estimate concentrations of pesticides that may occur in ground water that is less vulnerable to contamination. See below for further discussion.
-
The SCI-GROW estimate is based on environmental fate properties of the pesticide, the application rate, and the existing body of data from Agency-required small-scale prospective ground-water monitoring studies for all pesticides (in two cases, data from large-scale retrospective ground-water monitoring studies have also been used). Therefore, this estimate will change if:
-
The SCI-GROW model itself is modified based upon new data from additional ground-water monitoring studies;
-
New environmental fate data are obtained for the pesticide in question; or,
-
The maximum use rate of the pesticide in question is changed.
-
-
If and when adequate ground-water monitoring data are obtained for the pesticide of interest then the SCI-GROW estimate will be superseded by these data.
SCI-GROW: Instructions for Input Value Selection
-
Degradation half-life.
Use the average aerobic soil metabolism half-life (or median half-life if measurements are available in at least four different soils). If the compound is ionic, then use a half-life applicable to soils in which the compound is most likely to be stable and more mobile. If the degradation does not closely follow first-order kinetics, then a 100 to 50% disappearance time (or 50 to 25% disappearance time, if this is longer) may be used. More specific guidance on input for non-first order half-lives is being developed.
Field dissipation half-lives, because of the difficulty in the influence of factors other than degradation on the dissipation rate, should not be directly used in selection of the input values. However, cross checking with field dissipation half-lives is encouraged to develop a feel for the representativeness of the laboratory test soil (especially when there is only one or two soils in which degradation rates have been determined). When field dissipation half-lives are longer than the degradation half-life, this implies that there may be many soils where degradation is slower than occurred in the laboratory test soil.
If there is more than a five-fold difference in half-lives, note this, and you may also want to do a second SCI-GROW estimate with the longest half-life to see how much this adds to the uncertainty of the estimate.
Experience shows that if a pesticide is substantially subject to abiotic hydrolysis (i.e., not stable over the 30-day period that most such studies for registration are run), this generally mitigates its long-term ground-water contamination potential. If the hydrolysis data demonstrate that the pesticide in question is subject to hydrolysis, then this should be noted because it means that the SCI-GROW value may be an overestimate even for highly vulnerable ground water.
-
Soil organic carbon partition coefficient (Koc).
Use the median value (e.g., when there are values for four soils, use the average of the two central values), unless there is a greater than three-fold difference in the Koc values or other evidence that the Koc varies substantially between soils (because, for example, the compound is an acid and the degree of ionization and adsorption is strongly pH dependent). If the Koc is highly variable (that is, organic matter content is not the major factor for responsible for the variation in adsorptivity of pesticide residues between soils), then use a Koc for a sandy soil or other soil where leaching appears to be most likely.
-
Use rate.
This value is the total application rate allowed for an entire year for a major agricultural use site. If, for example, up to 3 applications of up to 1.0 lb ai/A are allowed; then the use rate for input into SCI-GROW should be 3.0 lb ai/A. When use rates allowed vary with soil texture, use the highest rate for a medium-textured (loam) or coarse-textured (sandy loam or coarser) soil.
Comments on the SCI-GROW Model, Version 1.0
-
Composition of Input Data
Currently, the model equation is based upon the fate data (aerobic soil metabolism half-life and soil partition coefficient normalized for organic carbon content) and the ground-water monitoring study results for ten pesticides. In some studies, the actual ground water concentration was significantly less or greater than the value that would be calculated with SCI-GROW. This is not surprising and does not at all reflect negatively on the performance of the model because each of these studies were conducted at different sites under very different conditions.
-
Deviations from Model Estimates
The natural variation in pesticide leaching is large, with several-fold deviations in the mass of pesticide leaching commonplace even over a small field with apparently uniform soil and having received a uniform application of the pesticide. Furthermore, tremendous variability in the mass of a given pesticide leaching in a field is introduced by variation in the timing and intensity of precipitation / irrigation events. Consequently, a regression procedure based upon multiple studies will never provide anywhere near a perfect fit. The SCI-GROW regression does, however, provide a good estimate of the kinds of ground water concentrations that most likely would occur for a particular pesticide under a use scenario which is most likely to result in ground-water contamination (the SCI-GROW concentration has usefulness as a screening or upper-bound value because the regression is based solely upon scenarios in which the vulnerability of ground-water contamination was much higher than the vast majority of agricultural use scenarios).
-
Characteristics of the Study Sites.
A detailed analysis of the soil and hydrogeologic characteristics of each site will be presented in the next version of this documentation, including the median and range of depth to ground water, sand content, clay content, organic carbon content, etc. In general, these studies were done at sites with 70 to 100 % sand content (average of 89 % in the upper 0.3 meters) throughout the profile and less than 10 % clay content throughout the profile (average of 4 % in the upper 0.3 meters). Organic matter content was generally less than 2 % in the A horizon (average of 0.9 % in the upper 0.3 meters). The depth to ground water was usually in the range of 8 to 25 feet (i.e., 2 to 8 meters with an average of 4.6 meters). The geographic location of the sites ranged all over the United States. Sufficient water was added to each site through irrigation to ensure that a year of higher than average rainfall was simulated (or, in irrigated areas, 10 or 20 % more than an average addition of irrigation water was added). An average of 17 inches (44 cm) of water reached the sites through precipitation plus irrigation during the first three months after application.
-
Application of SCI-GROW Estimates
The above discussion illustrates that there may be exceptional circumstances under which ground-water concentrations could exceed the SCI-GROW estimate. However, such exceptions should be quite rare since the SCI-GROW model is based exclusively on maximum ground water concentrations from studies conducted at sites and under conditions which are most likely to result in ground-water contamination. Furthermore, the data set used to develop SCI-GROW was biased in the sense that negative data were ignored: studies where the pesticide was not detected in ground water were not included in the data set even if the site had highly vulnerable ground water.
Acute versus Chronic Estimates. With most ground-water sources there are no known predictable seasonal and longer term trends in concentrations of pesticide contaminants. This is in contrast with surface water bodies of known dimensions and flow rates where time series trends are well described. For this reason, the highest average concentration for three consecutive samples (generally taken monthly) in a small-scale prospective ground-water study was assumed to provide a conservative (in the protective sense) estimate of chronic exposure levels. For large-scale retrospective studies, an average of the highest 5% of detections in areas with known prior use of the pesticide or similar value was used to estimate chronic exposure levels.
Additional work will be conducted to see if a better estimator of acute exposure levels can be obtained. However, based on an informal evaluation of the data, it is not expected that the acute exposure levels for ground water will be substantially different from the chronic exposure levels, at least as we have chosen to estimate them (highest 90-day average).
-
Ideas for Future Enhancements.
The current version of SCI-GROW provides a useful tool for screening out pesticides for which levels which might occur in ground water are highly unlikely to approach toxicologically significant levels. With additional ground-water pesticide residue data collected from highly characterized sites like the small-scale prospective study sites, more detailed modeling predictions of pesticide residues in ground water may be made. Evaluation of additional data may facilitate examination effects of depth to ground water, soil water-holding capacity, infiltration rates, etc. on pesticide residue levels in ground water. Data from sites with more moderate vulnerability of ground water to contamination along with regional characterization of the types of ground water used for drinking water may facilitate more general predictions of residue levels in ground water used for drinking water.
-
Model Evaluation and Validation
Some of the efforts in progress include the following:
-
A systematic comparison of SCI-GROW estimates to various ground-water monitoring data sets (for example, to answer the question can SCI-GROW be adopted to predict impacts of pesticides in vulnerable ground water/ drinking water in proportion to their use?)
-
A systematic comparison of SCI-GROW estimates to PRZM predictions of pesticide concentrations in leachate at a site with sandy soils and shallow ground water (SCI-GROW and PRZM estimates are comparable for weakly adsorbed compounds and SCI-GROW estimates are increasingly large relative to PRZM estimates for more strongly adsorbed compounds).
-
Addition of new data from recently completed small-scale prospective ground-water monitoring studies to the SCI-GROW regression data set. (These data are being incorporated into a database which will also be used for Tier 2 ground-water model development and validation). Also, an effort is being made to evaluate results from studies without detections in ground water, comparing SCI-GROW estimates for vulnerable sites with the detection limits in these studies.
-