Harold W. Thistle

TECHNICAL ADVANCES IN MODELING AERIALLY APPLIED SPRAYS

Since the 1989 publication of the original modeling approaches assembled into AGDISP, significant advances have been made in the development and validation of spray drift models. The latest version of AGDISP (version 8.08) now embodies the latest refinements to the computational engine first developed for NASA, improved by the USDA Forest Service, and implemented by the Spray Drift Task Force and the U.S. Environmental Protection Agency into a regulatory version. This article updates the previous summary of AGDISP, builds on a recent publication summarizing the regulatory version of the model, and includes the most recent modeling additions of atmospheric stability effects, vortical decay, simple terrain features, plant canopy, riparian barriers, and the aerial release of dry materials.

Rotary atomizer drop size distribution database

Wind tunnel measurements of drop Size distributions from Micronair A U4000 and A U5000 rotary atomizers were collected to develop a database for model use. The measurements varied tank mix, flow rate, air speed, and blade angle conditions, which were correlated by multiple regressions (average R-2 = 0.995 for A U4000 and 0.988 for AU5000). This database replaces an outdated set of rotary atomizer data measured in the 1980s by the USDA Forest Service and fills in a gap in data measured in the 1990s by the Spray Drift Task Force. Since current USDA Forest Service spray projects rely on rotary atomizers, the creation of the database (and its multiple regression interpolation) satisfies a need seen for ten years.

Building canopy retention into agdisp: Preliminary models and results

AGDISP uses canopy structure and collection efficiency to model canopy interception but assumes droplets that intercept a leaf surface are retained at that point and do not bounce or shatter. This behavior is not the case for many plant species, particularly species that are moderately to very difficult to wet, for which canopy deposition would be overestimated with this assumption. This article summarizes the initial implementation of process-based models for spray droplet bounce and shatter within an experimental build of AGDISP in order to predict spray retention within plant canopies. Spray simulations were run for formulations ranging in surface tension, applied to three species (wheat, canola, and capsicum) with different wettabilities and leaf orientations, and using the AGDISP ground model with and without the retention model, to evaluate the impact of the retention model on canopy deposition within AGDISP. The model outputs were also compared to previously determined tracksprayer results. The current AGDISP canopy deposition (interception) model was unable to fully account for differences in retention due to the spray formulation used or species studied. Incorporation of the process-driven models for bounce and shatter allowed these differences to be predicted. Over the three species and four formulations studied, there was good agreement (Pearson’s correlation coefficient = 0.9308, p = 0.0000, indicating an almost certain correlation) between predicted and experimentally determined retention. Considerable work is still required to make this approach practical. However, this article illustrates that the ability to model retention is important for many crops (and weeds) and that the approach outlined herein is an effective adjunct to the current AGDISP interception model.

Atmospheric Dispersion from a Point Source in Four Southern Pine Thinning Scenarios: Basic Relationships and Case Studies

An atmospheric tracer experiment using SF6 was designed to assess changes in the dispersive environment in the trunk space of a southern pine forest through four thinning regimes. Over 6000 mean half-hourly tracer samples were collected and analyzed along with a high-frequency time series of tracer concentration sampling at 1 Hz over the course of the trials. The experimental plot was thinned in four stages, from a state of dense boles and thick understory, >32.1 m2 ha-1 (140 ft2 ac-1) basal area, to a final basal area of 16.1 m2 ha-1 (70 ft2 ac-1) with understory removed. In this study, plume dilution in the unthinned stand followed previously observed relationships; however, thinned stands deviated from expectations. The in-canopy atmosphere was largely coupled with the above-canopy atmosphere once a basal area of 23.0 m2 ha-1 (100 ft2 ac-1) or lower was attained, so additional tree removal had less effect in changing the in-canopy environment. Dispersion exhibited surprisingly strong, near-linear relationships with stand density.

Performance Evaluation of Dense Gas Dispersion Models

Abstract This paper summarizes the results of a study to evaluate the performance of seven dense gas dispersion models using data from three field experiments. Two models (DEGADIS and SLAB) are in the public domain and the other five (AIRTOX, CHARM, FOCUS, SAFEMODE, and TRACE) are proprietary. The field data used are the Desert Tortoise pressurized ammonia releases, Burro liquefied natural gas spill tests, and the Goldfish anhydrous hydrofluoric acid spill experiments. Desert Tortoise and Goldfish releases were simulated as horizontal jet releases, and Burro as a liquid pool. Performance statistics were used to compare maximum observed concentrations and plume half-width to those predicted by each model. Model performance varied and no model exhibited consistently good performance across all three databases. However, when combined across the three databases, all models performed within a factor of 2. Problems encountered are discussed in order to help future investigators.

VALDRIFT—A Valley Atmospheric Dispersion Model

Abstract VALDRIFT (valley drift) is a valley atmospheric transport, diffusion, and deposition model. The model is phenomenological—that is, the dominant meteorological processes governing the behavior of the valley atmosphere are formulated explicitly in the model, although in a highly parameterized fashion. The key meteorological processes treated are 1) nonsteady and nonhomogeneous along-valley winds and turbulent diffusivities, 2) convective boundary layer growth, 3) inversion descent, and 4) nocturnal temperature inversion breakup. The model is applicable under relatively cloud-free, undisturbed synoptic conditions in which the winds in the valley are predominantly along the valley’s axis. The model is configured to operate through one diurnal cycle for a single narrow valley. The inputs required are the valley topographic characteristics, pollutant release rate as a function of time and space, wind speed and direction as functions of time measured at one height, lateral and vertical turbulent eddy di...

Modification of Droplet Evaporation in the Simulation of Fine Droplet Motion Using AGDISP

The current version of AGDISP incorporates a well-established representation of the aircraft vortices, wind speed, evaporation, canopy, and spray physical properties. However, several modeling areas have not been re-examined in several years, and it seems appropriate to investigate the algorithms involved with parsing the drop size distribution, the impact of the evaporation rate on droplet size, the behavior of the integration time step, and the approximate effect of spray cloud relative humidity.

Pheromone Movement in Four Stand Thinning Scenarios: High Frequency Plume Observations

An atmospheric tracer experiment using SF6 was designed to assess changes in the dispersive environment in the trunk space of a southern pine forest through four thinning regimes. The experimental plot was thinned from dense boles and thick understory (>140 ft2 (13 m2) basal area) in four stages with the final basal area being 70 ft2 (6.5 m2 ). Observations indicate that thinned stands are less susceptible to bark beetle attack and one possible reason could be that the plumes of pheromone the insects use for signaling cannot achieve the intended purpose in the thinned stand. The pheromone mechanism could be altered through direct dilution of the gaseous plume, through a reduction in spatial coherence of a plume making it more difficult to follow back to a source or through loss of the plume vertically from the stand environment through lofting due to surface heating and loss of overhead containment. This work is part of a larger program to improve the success of forest insect management strategies using pheromone. Data analysis is on-going but the high frequency tracer plumes examined in this study show that the plumes are more dilute and much less coherent spatially in the thinned stands. The plumes are filamentous in nature in all cases but wander (whip) across a larger volume of space in the thinned stands.

Spray Droplet Trajectory Length in Ambient Vortice

The length of droplet trajectories is important in applied deposition modeling of aerial sprays. The length of the trajectory determines the time over which evaporation will act on a droplet as well as the likelihood a droplet will encounter a canopy element. In this paper vortical motion in the atmosphere is parameterized by analyzing time series of wind speed data. Resultant discrete vortices are then defined and fit to a Lamb Vortex. Individual droplets are numerically dropped into the vortices and trajectory pathlength calculated. Trajectories of 50µm drops are nearly circular while 350µm drops are only slightly deflected by the lowest energy vortex analyzed here.

A Technical Review of MULV-Disp, a Recent Mosquito Ultra-Low Volume Pesticide Spray Dispersion Model

Abstract The authors of a recently published paper summarized the development of a regression model for ground-based ultra-low volume applications, suggesting that their model was sufficiently verified that it could be used extensively for mosquito control. These authors claimed that their statistical model was superior in its predictive capability to the extensively developed and Environmental Protection Agency–validated AGDISP mechanistic model. In this technical review, the assumptions, reduction and interpretation of data, and conclusions reached with regard to their model are discussed, and explicit misstatements and incorrect mathematical relationships are pointed out. Two published versions of the model regression equation give substantially different results without explanation. Petri dish collection was used for very small droplets, with no mention of collection efficiency. Meteorological data were misused based on manufacturer’s specification of instrument accuracy. We strongly disagree with many of the model results and show that the model misrepresents the actual behavior of aerosol sprays applied in the manner tested.