Downwind drift from almond airblast spray applications: field measurement to aid mechanistic model development

Abstract

BACKGROUND

Off‐target drift of pesticides from orchard airblast spray applications has potential implications for human health and the environment. A field study was conducted in a commercial almond orchard to generate spray drift data for validating a mechanistic airblast spray drift risk assessment model under development. Pyranine fluorescent dye solution spray was applied in 22 trials at 935.4 L/ha with the sprayer making four passes along the third drive lane upwind from the edge of orchard. Drift was measured with artificial spray samplers distributed in an adjacent open field up to 183 m from the edge of the orchard, which were retrieved and analyzed by fluorometry to generate dye drift deposition data.

RESULTS

Drift data showed downwind deposit decay, indicating the extent of drift under different prevailing weather conditions. Terminal drift amount (in percentage of applied dose) and distances were estimated to be: 6.0 × 10⁻⁷ at 531.1 m for artificial foliage; and 4.0 × 10⁻⁶ at 232.8 m for horizontal string. Weather condition caused variation in drift among spray trials but only wind direction, wind speed, and atmospheric pressure were significant in accounting for the variability.

CONCLUSION

This new dataset provides opportunities for objective pesticide risk assessment decision‐making that will potentially impact regulations affecting the California almond industry. The data will also be useful for confirming and/or updating spray application best practices for promoting efficiency, effectiveness, and environmental sustainability. © 2025 The Author(s). Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Spray drift deposit measured in almond orchard significantly decayed downwind until termination at 531.1 m for artificial foliage sampler and 232.8 m for horizontal string samplers based on logarithmic curve fitting. Drift variability among spray trials was significantly affected by variation in weather condition (wind direction, wind speed, and atmospheric pressure).

1. INTRODUCTION

California's tree crop industries produce over three‐quarters of US fruit and nuts. Almonds are among California's largest agricultural commodities, covering over 1.3 million crop bearing acres throughout the Central Valley region. Nearly 2.6 billion pounds of nuts valued at 3.8 billion dollars were produced in 2023 (NASS statistics, 2024) to supply over 80% of the global demand for almonds. Though highly productive, the rapid growth, high density, and close proximity of almond acreage throughout the state increase susceptibility to pest and disease infections. Navel orangeworm (NOW) is the most damaging pest to harvested almonds in California.

To reduce pest populations and prevent disease outbreaks, pesticides are one critical tool used by growers. Typically used as a last resort in keeping pest pressures below economic thresholds, pesticide application is preceded by other integrated pest management (IPM) practices such as sanitation and biological control. For effectiveness against the target pest, these first‐line‐of‐defense practices are coupled with pest monitoring to ensure optimal timing of pesticide application. Most pesticide applications in California are done with airblast sprayers, but inefficiencies such as poor coverage, low on‐target deposition, and associated risk of pesticide drift from orchard operations are well known. Consequently, poor efficacy of spray application can also lead to disease and other pest outbreaks, which can reduce productivity and profitability.

Pesticide drift – off‐site translocation of spray particles during pesticide application – from airblast applications is of great concern to federal and state regulators. It is to be noted that later drift by volatilization is outside the scope of this study. In view of the regulated community, pesticide drift regulation represents potential risks to both human health and the environment. ¹ , ² This is because underestimating drift could create unpredicted consequences on human health and the environment. Regulatory concerns related to human health encompass pesticide exposure to agricultural workers, bystanders, and surrounding rural communities subjected to drift. ³ , ⁴ , ⁵ Concerns about the environment involve contamination both directly and indirectly. ¹ Overestimating drift exposure could have regulatory repercussions such as needless pesticide label restrictions ⁶ that could ban important pest control chemicals. Such restrictions could unnecessarily restrain or eliminate pest control options for pests common to perennial crop production systems, exotic pests of quarantine concern, and/or vectors of human pathogens. These restrictions could also rid out excessive areas of arable land from production, thus making it more challenging to feed a growing world population.

Beside the influence of spray application parameters on on‐target deposition and off‐site drift, ⁷ , ⁸ , ⁹ the outcome of an application is also affected by the target canopy characteristics. ¹⁰ , ¹¹ To account for the total amount of spray applied in an airblast application, it can be simply solved as a summation of the amount that deposited on the target and the remaining amount that went off‐target to the ground and air. ² , ¹² , ¹³ , ¹⁴ , ¹⁵ In light of this, maximizing on‐target deposition implies minimizing drift and/or ground deposition. Drift reduction practices like matching the application equipment to the target canopy and spraying in favorable weather condition such as low wind, low temperature, and high relative humidity were suggested by Fox et al. ¹⁶ Holterman et al. ¹⁷ pointed out the importance of recognizing the influence of temporal variation in canopy characteristics on drift. The air‐in method of using two airblast sprayers in tandem with one sprayer only blasting air with no spray has also been proposed by van Steenwyk et al. ¹⁸ as showing promise for reducing drift near the edge of orchards.

The study presented here is part of a larger program of studies examining deposition and drift of pesticides applied using orchard airblast sprayers. The overarching objective of these programs when completed is to develop a mechanistic model of orchard airblast spraying that will allow regulators, pesticide applicators, and other stakeholders to simulate this activity. The calculated deposition and airborne movement of the applied pesticide can be used by regulatory risk assessors to estimate the risk to humans and the environment. A typical model output would portray deposition versus distance downwind from the application vehicle, providing input to exposure pathways of direct ingestion, concentration in water, and dermal exposure at specified downwind distances.

Models of pesticide deposition and drift generally fall into categories of full‐physics, mechanistic, and empirical (statistical). Teske et al. ¹⁹ give an overview of pesticide drift and deposition modeling. The primary audience of the model being developed in this program is the regulatory community. If the model is accepted as a tool by the regulatory community, other stakeholders will use the model to interact with that community. Regulatory models need to be approachable to non‐experts, need to use inputs that are readily available and provide consistent output quickly. Of the three types of models listed earlier, regulatory models generally fall into the categories of mechanistic or empirical. An example of a mechanistic model is AGDISP. ²⁰ , ²¹ This type of model uses basic analytical physical equations, tuned with data, to yield a prediction of spray droplet landing position and fate. The AGDISP model has around three dozen adjustable input variables but is generally run using defaults and libraries, so the user only has to supply basic information.

AGDISP was originally designed as an aerial application model and can be compared with the full physics model of aerial application published by Ryan et al. ²² The Ryan model does utilize computational fluid dynamics (CFD) in the model formulation. This type of model can yield greater insight into the process being described as compared to the simpler mechanistic models but are limited by the dense data requirements needed to achieve that reality. Finally, empirical models, such as those included in Tiers I and II of AgDrift ²³ , ²⁴ are used. These models are assemblages of data curves. Composite curves are created and predictions of deposition at distance are based on the composite curve of deposition versus distance. This type of model is convenient but suffers from limitations inherent in the data and are much less reliable when extended beyond conditions measured in the data collection program. An interesting hybrid model was created by Hong et al. ²⁵ and utilized a commercially available CFD package to generate results from a large number of spray scenarios and create a look‐up table. The model user then inputs a spray scenario and the model interpolates within the table of preexisting results to yield an answer.

Larbi and Salyani ¹³ , ¹⁴ have created a mechanistic model of orchard airblast spraying in citrus orchards. It is envisioned that this model will be used as a basis for the more general model proposed in this study. To facilitate this generalization, data has been collected under this program in an almond orchard, a grape vineyard, ²⁶ a citrus orchard ²⁷ and orchards of dormant and foliated apples. ⁶ As indicated earlier, the model will have many adjustable input variables, but the primary inputs will be application rate, variables to allow knowledge of droplet size at release, release position, wind speed, wind direction and humidity. Other critical inputs will be sprayer type and canopy and orchard architecture. It is envisioned that these last two input groups will be selected from an imbedded library. A sprayer type library might allow the user to pick by manufacturer and model and include such information as nozzle and fan position, typical operating pressures at release, typical forward speeds, and so forth. The library would automatically populate the inputs but also allow the user to modify the library values in a given scenario. Similarly, the orchard canopy library would contain typical values of orchard structure (tree and row spacing) as well as canopy architecture (canopy density with height) for a given orchard type. For instance, if a pear orchard was being sprayed, the user would call the library entry for pears and then be allowed to modify the inputs based on exact knowledge of the scenario. The library could contain entries for different pear phenologies but even the library entries are constrained by existing data. A library of this type has been developed for forests. ²⁸

Developing this essential mechanistic model, which will account for actual application parameters pertinent to modern almond orchard settings, necessitates the collection of new data. Therefore, in support of developing and validating the proposed mechanistic model for estimating pesticide exposure due to drift, this current study sought to generate such data from almond airblast applications based upon a US Environmental Protection Agency (USEPA)‐approved protocol. ²⁹ , ³⁰ As an offshoot of providing a more objective assessment of downwind pesticide exposure from drift, the new dataset will also provide a basis for confirming and/or developing new spray application best practices that will promote efficiency, effectiveness, and environmental sustainability in almond production. This will enhance extension education trainings for updating or upgrading the knowledge and skills of growers and pesticide applicators toward more effective airblast spray applications. ³¹

2. MATERIALS AND METHODS

A study was done in a California almond orchard in accordance with a USEPA approved data collection protocol ⁶ , ³⁰ to assess airblast spray drift. Setup of the field experiment was performed from late‐May to early‐June of 2021 – entailing canopy characteristics measurement, drift sampling structures and weather instrumentation installation, and spray equipment calibration. The experiment was accomplished from mid‐June to June ending. Details of the study are provided subsequently.

2.1. Study site characteristics

The study site was a commercial ‘Independence’ almond orchard having merged canopies along tree rows with intermittent touching or nearly touching canopies across tree rows. The orchard met the basic site requirements of having a row length of ≥ 152 m (500 ft) and an adjoining open field (bare ground or low vegetation cover) extending to ≥ 183 m (600 ft) downwind (Fig. 1). With reference to the edge of the orchard, the direction away from the orchard was considered downwind (positive) while the direction into the orchard was taken as upwind (negative). The tree/orchard characteristics are summarized in Table 1. The canopy profile and foliage density of a random sample of 15 trees in the orchard were measured using a plant canopy analyzer (LAI‐2200C; LI‐COR, Inc., Lincoln, NE, USA), an extendable pole, and a measuring tape to represent the entire orchard.

Figure 1.

Sampling transect layout showing required placement of different sampling stations for various sampler types. Wind direction was monitored at Met 2 weather station and used to trigger the start of spraying in each trial.

Table 1.

Attributes of almond orchard used in study

Attribute	Tree and orchard characteristics
Crop type/variety	‘Independence’ variety
Tree/row height, m (ft)	5.5 (18.0)
Canopy width, m (ft)	6.5 (21.3)
Leaf area density, a m2/m3	1.3 ± 0.26
Row spacing, m (ft)	6.6 (21.5)
Tree spacing, m (ft)	4.6 (15.0)
Row direction	E–W
Downwind direction	N → S
One‐way length of sprayer path, b m (ft)	152.4 (500.0)

^a Measurement based on a random sample using a plant canopy analyzer (LAI‐2200C; LI‐COR, Inc., Lincoln, NE, USA).

^b Four passes of length considered one run in the study.

2.2. Field setup

The field setup (Fig. 1) was made up of artificial spray sampling structures distributed along four equally spaced transects [18 m (60 ft) between transects, T1–T4] from within orchard (upwind) up to 183 m (600 ft) away from the edge of the orchard (downwind). The collection stations had various combinations of artificial samplers with different collection efficiencies – flat card (C), artificial foliage (AF), and string collectors [horizontal string (HS) and vertical string (VS)]. For instance, studies show that AF samplers (cut out of artificial Christmas tree) have greater efficiency in capturing fine droplets over C samplers. ⁶ , ³² The individual collection stations of similar distance were considered subsamples. For example, the 61 m (200 ft) downwind collection stations from transects T1, T2, T3 and T4 were designated subsamples 1, 2, 3 and 4, respectively, for the 61 m downwind distance. Towers with vertical string collectors were stationed at 7.6 m (25 ft) and 22.9 m (75 ft) downwind from the edge of the orchard. These had the basic station at the base, strings at H, 1.5H and 2H, where H = tree/vine canopy height. The sampling height of the basic station was about 0.91 m (3 ft).

Details of one type of the structures used for holding one each of C, AF, and HS samplers at the same time is also shown in Fig. 2 which was placed at 15, 30, 61, 122 and 183 m in each sampling line. Another type which excludes the string collector was placed at 0, 3, 8, 23, 46, and 76 m downwind, as well as within the orchard at −9 and −20 m which had only C collectors. A final type of structure was a ‘tower’ that held a VS collector which was placed at 8 and 23 m downwind, shown in Fig. 3. The VS samplers were pre‐marked to be cut after each trial at H, 1.5H and 2H (where H = tree canopy height) to obtain three samples.

Figure 2.

Different sampling structures used in study for various horizontal sampler combinations: (a) holds flat card, artificial foliage, and horizontal string; (b) holds flat card and artificial foliage; and (c) holds flat card or artificial foliage.

Figure 3.

Views of the field setup highlighting installed vertical string sampling structures.

2.3. Weather instrumentation and measurement

Two weather stations having sensors at different heights were installed to record meteorological data: one placed upwind inside the orchard and another about 183 m (600 ft) downwind in the open field. A summary of the sensors used at the weather stations is provided in Table 2. The instruments were all connected to data loggers (Zentra ZL6; METER Group, Inc., Pullman, WA, USA; and CR1000/CR1000X; Campbell Scientific, Logan, UT, USA) and powered by 12 V solar batteries which were maintained by solar panels. The meteorological data were retrieved from all the sensors at the end of the spray trials described later and processed in Microsoft Excel.

Table 2.

Instrumentation used to collect meteorological parameters. Wind direction was monitored at Met 2 weather station and used to trigger start of spraying for each trial ^a

MET 1: Inside orchard		MET 2: Outside orchard
39.3 m (129 ft) upwind		183.8 m (603 ft) downwind
Height AGL, m (ft)	Sensors	Height AGL, m (ft)	Sensors
0.9 (3)	S1, S2	0.9 (3)	S1, S2
2.7 (9)	S1, S2	1.8 (6)	S1, S2
5.5 (18)	S1, S2	3.0 (10)	S1, S2
11 (36)	S1, S2	9.1 (30)	S1, S2

^a S1 = All‐in‐one weather sensor (ATMOS 41; METER Group, Inc., Pullman, WA, USA); S2 = three‐dimensional ultrasonic (81000; R.M. Young Co., Traverse City, MI, USA).

2.4. Application equipment and parameters

A power‐take‐off (PTO) powered conventional airblast sprayer (GB‐36 PTO; Air‐O‐Fan, Reedley, CA, USA) connected to a John Deere 5100E (Deere & Co., Moline, IL, USA) was used in the study as the application equipment. It was calibrated by measuring and determining ground travel speed, assessing the air profile, measuring spray nozzle flowrates, and adjusting sprayer settings to obtain a target application rate at a given operating pressure. Fan air velocity at the fan outlet was also measured. The application rate was selected based on grower standard and typical of the region. The number and positions of open nozzles used for the spray application were chosen to direct the spray onto the target canopy with minimal waste. Table 3 shows a summary of the parameters used in the study while Table 4 provides further details of the nozzle type and configuration.

Table 3.

Spray application parameters used in studies ^a

Application parameter	Setting
Nozzle type b	TXR ConeJet® Hollow Cone
Number of open nozzles per side	9
Uppermost nozzle angle (deg, w.r.t. vertical)	28
Lowermost nozzle angle (deg, w.r.t. vertical)	78
Travel speed, km/h (mph)	3.6 (1.7)
Operating pressure, bar (psi)	11.7 (170)
Sprayer output per side, L/min (gpm)	15.83 (4.18)
Adjusted application rate, L/ha (gpa)	935 (100)

^a Sprayer was of Air‐O‐Fan brand, Reedley, CA, USA.

^b All nozzles were TeeJet® nozzles, Spraying Systems Co., Wheaton, IL, USA.

Table 4.

Nozzle configuration of sprayers used in studies

Nozzle position (clockwise)	Nozzle characteristics
	Nozzle size	Angle w.r.t. vertical, deg	Mean flow rate, L/min (gpm)
1	TXR80028VK	28.0	2.04 (0.54)
2	TXR8003VK	34.3	2.12 (0.56)
3	TXR8003VK	40.5	2.10 (0.55)
4	TXR8003VK	46.8	2.16 (0.57)
5	TXR80028VK	53.0	1.95 (0.51)
6	TXR80028VK	59.3	2.01 (0.53)
7	TXR80017VK	65.5	1.13 (0.30)
8	TXR80017VK	71.8	1.16 (0.31)
9	TXR80017VK	78.0	1.16 (0.31)

2.5. Spray application trials

With the sprayer making four passes along the third drive lane upwind from the edge of the orchard (two passes per direction of row), delivering spray from both sides of the sprayer, the immediate row of trees on each side was sprayed as one trial (i.e., replication). The length of row sprayed in one pass was 152 m (500 ft). Live wind direction data from the ATMOS 41 instrument at 1.8 m at Met 2 was monitored to dictate spray start times. Twenty‐two trials were performed, of which 18 were treatment trials (trials 2–13, and 15–20) and four were blank trials (trials 1, 14, 21, and 22). The treatment trials entailed the application of pyranine fluorescent tracer dye (sodium 8‐hydroxypyrene‐1,3,6‐trisulfonate) solution (at a target dye concentration of 2000 mg/kg) while the blank treatments consisted of applying just water with no dye in identical fashion as the treatment trials. Figure 4 shows the sprayer at the start of a pass during one of the treatment trials. A complete set of C, AF, HS, and VS samplers was first installed before running each trial and carefully removed after that into prelabeled zipper bags and packed in a cooler. Sprayer tank samples were collected at separate times during the experiment and also kept in a cooler. The spray drift and tank samples were carried to the laboratory and temporary stored in a refrigerator until they were analyzed.

Figure 4.

Spray application in progress inside orchard during a trial.

2.6. Sample analysis

In the laboratory, the drift samples were each analyzed by fluorometry to first obtain the mass of dye that deposited on each sampler type. This value was divided by the projected surface area of the sampler to obtain dye drift deposition expressed in ng/cm² and further in percentage of applied rate (or dose). The projected area of the AF samplers was evaluated by image analysis. ³⁰ , ³² The analysis was undertaken by transect and by replication, where all samples for T1 were analyzed for all replications before T2, T3, and T4. The sample analysis is described in much more details in Larbi et al. ²⁹ , ³³ Fluorometric analysis was also used to determine the actual concentrations of the tank samples, which was combined with the actual spray application rate for each trial to normalize the spray drift data.

2.7. Statistical analysis

The experimental data were consolidated and processed in Microsoft Excel. Scatter plots of airborne drift captured with the VS samplers and downwind drift deposition (sedimentation) evaluated with the C, AF, and HS samplers were obtained from the processed deposition data. Mean plots of the airborne data at the two downwind distances were added for visual comparison. A two‐way analysis of variance (ANOVA) with downwind distance and sampling height as independent variables was performed on the airborne drift data at a significance level of 0.05 in SigmaPlot 12.5. A mean curve was added on the C drift deposition data to indicate the trend in deposition from within the orchard (upwind) up to the 183 m downwind. Logarithmic curves were fitted to the AF and HS drift deposition data. Based on the fit curve equations, drift termination distance and final value of deposition at termination were estimated. All the weather data downloaded from the weather sensors were processed in Microsoft Excel. Time series plots of all the weather parameters from each spray trial were generated for analysis. Summary statistics were obtained and averaged for comparison among the trials. The composite influence of the weather parameters (solar radiation, wind direction, wind speed, air temperature, atmospheric pressure, and relative humidity) on mean drift deposition from the five downwind sampling locations common to all horizontal samplers (i.e., 15.2, 30.5, 61.0, 121.9, and 182.9 m) was analyzed by a multiple linear regression (MLR) in SigmaPlot 12.5 at a significance level of 0.05. The MLR output was used to recognize redundant and non‐influential factors. Interactions among the factors were not analyzed.

3. RESULTS AND DISCUSSION

The results of the California almond airblast spray drift study, including weather conditions, spray drift, and effects of weather parameters on drift are provided.

3.1. Weather conditions

The summary weather conditions recorded during the experiment by the all‐in‐one weather instruments installed at 2.7 m high at Met 1 weather station and at 1.8 m high at Met 2 station are shown in Table 5. The data denote mean weather condition from 2 min to start of spraying until all samples were retrieved following spraying in each trial. The average values of the parameters ranged as follow: 95.3–956.9 W/m² for solar radiation; southeast–northwest (SE–NW) for wind direction; 0.4–1.3 m/s for wind speed; 17.6–31.6 °C for air temperature; 1.5–2.6 kPa for vapor pressure; 99.5–100.7 kPa for atmospheric pressure; and 35.7–85.1% for relative humidity.

Table 5.

Summary weather data for the treatment trials based on the all‐in‐one weather sensors installed at a height of 2.7 m (9 ft) at Met 1 and 1.8 m (6 ft) at Met 2

Trial number	Solar radiation, W/m2		Wind direction, deg		Wind speed, m/s		Air temperature, °C		Vapor pressure, kPa		Atmospheric pressure, kPa		Relative humidity, %
	Met 1	Met 2	Met 1	Met 2	Met 1	Met 2	Met 1	Met 2	Met 1	Met 2	Met 1	Met 2	Met 1	Met 2
2	938.0	917.9	S	ESE	0.31	1.37	24.8	26.6	1.60	1.34	100.6	100.7	51.1	38.6
3	308.5	676.8	WNW	NW	0.35	2.06	28.6	31.2	1.92	1.45	100.4	100.5	49.0	31.9
4	384.4	656.9	SW	SW	0.27	1.26	23.5	25.2	1.59	1.32	100.4	100.4	55.3	41.1
5	39.8	153.0	NW	NW	0.22	1.44	21.7	24.7	2.26	2.09	99.6	99.6	87.0	67.2
6	205.9	308.5	WNW	W	0.37	1.34	27.4	31.3	2.95	2.27	99.5	99.5	80.9	49.7
7	41.3	237.4	W	NW	0.14	1.54	17.9	20.6	1.91	1.75	100.0	100.1	93.3	72.3
8	444.41	765.97	W	SE	0.36	1.00	28.1	28.8	2.18	1.90	100.1	100.1	57.6	48.1
9	37.6	222.4	SSE	NW	0.16	1.24	16.6	18.6	1.71	1.65	100.1	100.1	90.7	76.9
10	588.9	599.2	SW	W	0.28	1.11	21.5	23.9	1.83	1.61	100.1	100.2	71.4	54.4
11	704.1	823.9	W	W	0.36	1.47	26.1	27.2	1.93	1.63	100.1	100.1	57.2	45.2
12	984.6	929.3	W	W	0.38	1.51	28.0	30.0	1.82	1.37	100.0	100.0	48.2	32.3
13	904.7	890.1	W	WSW	0.40	1.29	30.3	32.6	1.89	1.44	99.9	99.9	43.8	29.3
15	686.9	816.2	WSW	SW	0.33	1.15	25.4	26.3	1.86	1.66	100.1	100.2	57.4	48.5
16	983.7	926.2	WNW	NW	0.35	1.76	27.7	29.5	2.01	1.69	100.1	100.1	54.0	41.1
17	923.4	827.4	W	WNW	0.42	1.68	30.2	33.0	1.92	1.39	99.9	100.0	44.9	27.6
18	70.6	346.7	SE	SE	0.22	0.57	18.5	20.6	1.73	1.61	100.4	100.4	81.4	66.5
19	830.9	843.7	W	W	0.32	1.21	24.9	26.9	1.92	1.70	100.4	100.4	60.9	48.1
20	877.4	842.5	W	NW	0.45	2.25	28.9	31.2	2.01	1.59	100.2	100.2	50.6	35.1

Figure 5 shows summary compass roses of all the wind speeds and directions during the treatment trials. Consistent with other studies, ⁶ the data indicate that overall wind direction for the treatment trials varied during the experiment despite aiming to start from southwards based on monitoring at Met 2 weather station.

Figure 5.

Compass roses of all wind speeds and directions during treatment trials at: (a) Met 1 at 2 m height; and (b) Met 2 at 1.8 m height.

3.2. Off‐target spray drift

The amount of spray missing the target tree rows that remain airborne at the edge of the orchard represents the drift potential beyond the edge of orchard. This is a function of the tree canopy structure and foliage density, spaces in‐between trees, and the space underneath the canopy skirt. The downwind movement of this off‐target spray beyond the edge of the orchard is what constitutes drift. Drift collected downwind using the four different spray deposit samplers is discussed later. The data show a general decline of drift deposition over downwind distance under the different prevailing weather conditions; this can primarily be due to downwind dispersion of the spray particles. However, variation in the drift data is observed among the different treatment trials possibly a result of variation in the prevailing weather conditions among the experimental trials.

3.2.1. Vertical string (VS) samplers

The VS samplers were used to capture airborne spray drift at the downwind distances of 7.6 and 22.9 m in each experimental trial. The strings were carefully divided into three separate predetermined sections during retrieval from the towers. Drift captured by the bottom section corresponds to the amount of drift emerging directly from the far side of the canopy having not been intercepted by mainly the canopy and trunk. The middle section corresponds to spray drift coming from immediately above the canopy while the upper string section corresponds to spray reaching much higher above the canopy. Figure 6 shows the airborne drift captured. Mostly, the amount of airborne drift captured at the closest distance was higher compared to further from the spray release (P = 0.011) based on a two‐way ANOVA, as a result of dispersion and deposition of material as the spray moved downwind. The overall difference in the airborne drift amount at the three different vertical sections was also significant (P ≤ 0.001) although there was no significant difference between the middle and upper sections (P = 0.802).

Figure 6.

Spray drift deposit representing airborne drift collected at downwind distances of 7.6 and 22.9 m on vertical strings, at midpoints of H = 0–5.5 m, 1.5H = 5.5–8.3 m, and 2H = 8.3–11 m (where H is tree height).

3.2.2. Flat card (C) samplers

Spray deposition collected on C samplers upwind within the orchard (which is not drift) and downwind beyond the edge of the orchard (which is drift) is shown in Fig. 7 in linear plot (Fig. 7(a)) and log‐linear plot (Fig. 7(b)). Overall, the plots show decline of drift from the 0‐m downwind distance, which is muted in the linear plot due to high deposition at −16.4 m (−53.6 ft) downwind, that is, ‘‐ Row 2’. An enhanced visualization of the data is seen in Fig. 7(b). Based on a three‐way ANOVA, drift significantly declined over downwind distance (P = <0.001), significantly differed among the transects (P = 0.006), and significantly varied among the 18 treatment trials (P ≤ 0.001). Downwind drift was essentially significant only up to about the first 30 m based on pairwise comparisons. As previously mentioned, differences among the trials were probably due to differences in the prevailing weather condition during the trials. Moreover, the difference among the transects could be because of differences in particle transport dynamics.

Figure 7.

Spray dye deposit collected on plastic card samplers versus downwind distance (0 = edge of orchard) inside and outside of the orchard: (a) linear axis and (b) log‐linear axis.

3.2.3. Artificial foliage (AF) samplers

A summary of the spray drift collected on the AF samplers is provided in Fig. 8. Figure 8(a) is a linear plot while Fig. 8(b) is a log‐linear plot, both fitted with logarithmic curves. Behaving the same way as the C samplers drift data, drift captured by the AF samplers also significantly declined over downwind distance (P ≤ 0.001), significantly differed among the transects (P = 0.019), and significantly varied among the 18 treatment trials (P ≤ 0.001). The mean downwind distance where drift completely decayed was estimated as 531.1 m (1742.5 ft) and the minimum drift value in percentage of applied dose at that distance was estimated to be 6.0 × 10⁻⁷. This estimation is in consideration of drifting dye particles (non‐water component) beyond the point of complete droplet evaporation.

Figure 8.

Spray drift deposit on artificial foliage samplers with downwind distance: (a) linear axis and (b) log‐linear axis.

3.2.4. Horizontal string (HS) samplers

Spray drift data obtained using the HS samplers are provided in Fig. 9 as a linear plot (Fig. 9(a)) and a log‐linear plot (Fig. 9(b)). Just like the C and AF samplers, HS drift data values also declined significantly over downwind distance (P ≤ 0.001), significantly differed among the transects (P = 0.020), and significantly varied among the 18 treatment trials (P ≤ 0.001). HS drift completely decayed at a mean downwind distance of 232.8 m (763.8 ft), extrapolated from the logarithmic curve fitting equation, with a final drift value of 4.0 × 10⁻⁶% of applied dose.

Figure 9.

Spray drift deposit on horizontal string samplers with downwind distance: (a) linear axis and (b) log‐linear axis.

3.3. Meteorological influence

The overall effect of the weather conditions prevailing during the field trials are described later. Table 6 provides the output of the MLR which denotes that the overall mean drift deposition (dependent variable) can be predicted from a linear combination of the six independent weather variables (P ≤ 0.001). However, only wind direction (P ≤ 0.001), wind speed (P ≤ 0.001), and atmospheric pressure (P ≤ 0.002) appear to account for the ability to predict overall mean drift deposition. The others are not necessary for explaining overall mean drift deposition that occurred in the study, partly due to the presence of multicollinearity among the independent variables.

Table 6.

Output of multiple linear regression based on the all‐in‐one weather sensor installed at a height of 1.8 m (6 ft) at Met 2 for evaluating the effect of weather condition on overall mean drift deposition

Variable	Coefficient	Standard error	t	P	VIF
Constant	2.875	0.918	3.132	0.003
Solar radiation	−1.03E−05	1.03E−05	−1.001	0.32	4.826
Wind direction	−0.000153	0.000032	−4.772	<0.001	2.782
Wind speed	0.0491	0.00622	7.902	<0.001	3.447
Air temperature	−0.00141	0.00119	−1.179	0.243	15.817
Atmospheric pressure	−0.0284	0.00884	−3.209	0.002	4.031
Relative humidity	−0.0000858	0.000362	−0.237	0.813	17.884

4. CONCLUSION

Off‐target drift of pesticides from orchard airblast spray applications has potential implications for human health and the environment. Assessments of airblast spray application in an almond orchard denote that particles could drift beyond 183 m downwind of the orchard being sprayed. Variation of airborne drift over vertical flux profile of the plume was significant (P ≤ 0.001) and the amount of airborne drift captured decreased significantly (P = 0.011) from 7.6 to 22.9 m downwind because of dispersion and particle deposition as the spray moved downwind. Drift deposit significantly decayed downwind (P < 0.001) until termination. Drift deposition completed at 531.1 m for AF samplers and 232.8 m for HS samplers based on logarithmic curve fitting with respective final estimated drift values of 6.0 × 10⁻⁷ and 4.0 × 10⁻⁶% of applied dose. Consistent with other studies, greater drift deposition values were obtained from both AF and HS samplers than C samplers at common distances downwind. Variability in weather condition between spray trials caused significant corresponding variation in the drift data (P ≤ 0.001) but multicollinearity among some variables led to only wind direction, wind speed, and atmospheric pressure being significant (P ≤ 0.019) in accounting for the variability. The generated dataset, which is not intended for direct comparison with existing regulatory models, will be employed in developing and validating a mechanistic model for estimating drift from almond orchard airblast applications. The model will account for canopy characteristics and weather parameters.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest in this article or associated research.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.