Guidance on the Use of the Linked AGDISP-Gaussian Extension Models for Estimating Far Field Drift of Pesticides in FIFRA Ecological Risk Assessments
March 2, 2010
SUBJECT: Guidance on the Use of the Linked AGDISP-Gaussian Extension Models for Estimating Far Field Drift of Pesticides in FIFRA Ecological Risk Assessments
FROM: /s/ Donald J. Brady, Ph.D.,Director, Environmental Fate and Effects Division (7507P)
TO: EFED Managers and Staff
I am issuing this technical guidance on the use of the linked AGDISP-Gaussian extension models for estimating far field drift of pesticides in pesticide risk assessments, consistent with senior science staff recommendations. This guidance is effective immediately.
The Lagrangian model of AGDISP predicts near-field spray drift up to 800 meters (0.5 miles) downwind from the application site. The Gaussian Extension of AGDISP model can be used to predict pesticide spray drift up to 20 kilometers (12.8 miles) downwind from the application site. In response to concerns by some Environmental Fate and Effects Division (EFED) scientists, the EFED senior scientists re-evaluated the use of the Gaussian Extension. The concerns expressed centered on the model's lack of consideration of physical barriers (e.g. trees and topographic features, etc.) and meteorological conditions (e.g., crosswinds, humidity, etc.) beyond the application site affecting far-field drift of pesticides. The Environmental Fate and Effects Division (EFED) senior science staff has re-evaluated the use of the Gaussian Extension Model in AGDISP (Version 8.13, 2004) for ecological risk assessment. The senior science staff review noted three issues:
there is significant uncertainty in estimating pesticide deposition at far-field distances (> 0.5 miles),
the estimation of far-field drift distance under the current modeling assumptions is conservative and does not take into account potential physical barriers, and
there is a lack of validation beyond 2 miles of the linked AGDISP-Gaussian Extension models for estimating far-field pesticide drift.
Use of the linked AGDISP-Gaussian Extension models for estimating far field drift are to be used only for aerially applied mosquito adulticides or those pesticides aerially applied using a ASAE droplet size spectrum of very fine to fine, with a minimum pesticide release height of 50 feet above the ground with 10 feet above the plant canopy. Aerial spray conditions covered by the guidance include forests and tree orchards (walnut, pecans, etc). These use scenarios are most closely related to the parameters that have been validated and for these use scenarios the uncertainty relative to potential physical barriers at lower application heights or larger droplet sizes, will be minimized.
DISCUSSION OF THE ISSUES
Accounting for Mass Applied
The Gaussian plume equations and dispersion coefficients used in AGDISP are similar to those used in Industrial Source Complex (ISC) model. The AGDISP equations however did not appear to account for plume depletion. The Industrial Source Complex allows for the use of a depletion factor in its algorithms to account for the mass that has already been deposited. The lack of a depletion factor would result in conservative estimates of pesticide concentration. To further explore this issue the senior science staff contacted Dr. Harold Thistle, the AGDISP model developer, who informed them that AGDISP does account for the mass of applied pesticide.
Appropriate Uses of the Gaussian Extension
In a conversation with EFED senior scientists, Dr. Thistle, stated the Gaussian Extension model was initially integrated into the AGDISP model to estimate drift of Bacillus thuringiensis (Bt) bacteria from spray applications to forests. He stated there is expected to be high uncertainty associated with far-field estimates of pesticide concentrations from what is traditionally thought of as spray drift. Additionally, Dr. Thistle stated the Gaussian Extension model in AGDISP should not be used for ground spray applications.
Validation of the AGDISP-Gaussian Extension Models
Gaussian plume models, in particular the ISC, are effective out to 50 kilometers (31.1 miles), regardless of whether the release is from the ground or a huge power plant stack. Because EFED is considering wind speeds of 10-15 mph, pesticide transport times of 0.8-1.2 hrs are expected over a distance of 20 km. Eddy effects are expected to play a substantial role downwind, either depositing the material more quickly or reintroducing it into the atmosphere for deposition further away. Teske and Thistle, 20041 validated the AGDISP-Gaussian Extension models for level terrain with little ground cover in the high desert of Utah where straight line winds and few competing topological factors exist. They found reasonable agreement between the predicted concentrations from AGDRIFT-Gaussian Extension models and observed pesticide deposition data from 0.1 km (0.062 mi) to 10 km (6.2 miles). Similarly, Woods et al., 20012 further validated the Gaussian extension model for pesticide deposition data from 0.01 km (0.006 mi) to 1 km (0.62 mi). In pesticide ecological risk assessments, Gaussian extension modeling assumes no wind barriers exist from the application site to 20 kms (12.4 mi) downwind. Additionally, meteorological variables (crosswinds, humidity, etc.) are also not considered in the Gaussian Extension modeling. These factors are expected to reduce far-field pesticide drift. Therefore, the current Gaussian extension modeling is expected to be extremely conservative in assessing potential far-field spray drift. There are limited data validating the AGDISP-Gaussian Extension models for assessing the impact of wind barriers such as plants and other topographic features on far-field drift and deposition of pesticides.
lTeske, M.E. and H. W. Thistle. 2004. Aerial Application Model Extension into the Far Field. Biosystems Engineering. 89(1): 29-36.
2Woods, N., I. Craig, G. Dorr , B. Young. 2001. Spray Drift of Pesticides Arising from Aerial Application in Cotton. J. Environ. Qual. 30:697-701.