Spray Deposition, Drift and Equipment Contamination for Drone and Conventional Orchard Spraying Under European Conditions

Abstract

In Europe, there is a growing interest in crop spraying using unmanned aerial vehicles (UAVs, drones), although current legislation imposes significant limitations on this technique. Spraying of orchard crops with drones remains particularly challenging due to the risks of spray drift and insufficient deposition uniformity. This study evaluated spray deposition within tree canopies (in two application terms), airborne and sediment drift losses, and contamination of the spraying equipment. The performance of a medium-sized drone (ABZ Innovation L10, maximum take-off weight 29 kg) was compared at flight speeds of 1.8, 2.7, and 3.6 m·s⁻¹ with that of a conventional orchard sprayer (Munckhof axial sprayer with column attachment, operating at 1.7 m·s⁻¹). A fluorescent tracer (BF7G, 1200 g·ha⁻¹) was used in all trials, with spray volume rates of 27 or 40 L·ha⁻¹ for the drone and 400 L·ha⁻¹ for the sprayer. In most cases, deposition within the tree canopy was significantly lower for the drone. Poor uniformity of spray distribution was observed, especially between the upper and lower surfaces of collector plates with attached filter papers and between the top and bottom canopy zones. Airborne drift increased significantly with higher drone flight speeds, while sediment drift decreased. At 1.8 m·s⁻¹, both drift types were comparable to those from the conventional sprayer. Drone surface contamination was several times lower than that of the ground sprayer, even when accounting for differences in equipment surface area.

1. Introduction

Drone spraying has become an important topic within sustainable and precision agriculture due to its potential for labor reduction, automation, and site-specific pesticide application. The first attempts at unmanned aerial spraying date back to the 1980s with unmanned helicopters [1]. The modern era of electric multi-rotor UAVs, such as DJI and XAG models, began between 2010 and 2015, when field trials were conducted in orchards in the USA, Japan, and China. From 2015 to 2020, UAV spraying technologies advanced rapidly, particularly in China, and the first studies appeared on spray deposition, drift losses, and the effects of nozzle type, weather, and flight parameters on spray performance.

In China, drones have proved an efficient alternative to manual or knapsack spraying [2]. However, within the European Union, drones used for pesticide application are still legally categorized as aircraft, subject to the same restrictions as aerial spraying by airplanes or helicopters. Directive 2009/128/EC [3] prohibits aerial spraying but allows derogations where no viable ground-based alternatives exist and where environmental and health risks are reduced. The applied product must be approved for aerial use, and the operator must hold a relevant certificate. Although these regulations restrict drone spraying in the EU, the growing use of UAVs in difficult terrain (e.g., mountainous or waterlogged areas) motivates continued research and the development of practices suited for spatial crops such as orchards and vineyards, where ground spraying is often inefficient or impossible.

Orchards, like vineyards, require frequent pesticide treatments and often suffer from uneven deposit distribution and significant spray losses. While Polish orchards are generally located in areas accessible to tractor-mounted sprayers, vineyards in southern Europe are often established on slopes where tractors cannot operate safely. Thus, drone-based spraying may provide a practical alternative under EU conditions. Research into UAV spraying effects in orchards and its associated risks is justified both by the growing interest among growers and service providers and by scientific needs.

Spraying with drones differs fundamentally from conventional ground-based application. A tractor sprayer moves between rows and delivers spray from the sides, whereas drones spray from above, either directly above the tree axis or along offset parallel passes [4]. Tractor sprayers rely on fans generating horizontal airflows to transport droplets, while drones generate vertical air movement via propeller downwash. This airflow, depending on the drone’s lift requirements, flight altitude, and payload, strongly influences spray deposition and drift [5,6,7]. As noted by Chojnacki and Pachuta [6], drone rotor airflow cannot be freely adjusted because it depends on the drone’s weight and flight conditions. Higher flight speeds increase airflow and penetration but may reduce uniformity of coverage [7]. In contrast, for ground sprayers, airflow and droplet trajectories can be optimized by adjusting nozzle distance and travel speed [8,9]. Flight altitude significantly affects deposition and drift; at higher altitudes, weaker downwash and longer droplet trajectories promote off-target movement [10,11].

Both laboratory and field studies have been used to evaluate drone spraying performance. Laboratory tests focus on the influence of drone design, propeller configuration, and nozzle type on droplet size and deposition patterns [6,7,12], while field trials assess deposit quality, drift magnitude, and operator exposure under varying operational and environmental conditions. However, as Wang et al. [7] highlight, the lack of standardized testing protocols limits comparison among studies. ISO standards for drift and spray distribution—such as ISO 24253-2:2015 [13], ISO 22522:2007 [14], and ISO 22369-1:2006 [15]—offer partial guidance, but they were developed for ground sprayers. The first ISO standard dedicated to UAV spraying (published in 2025 [16]) provides methods to assess horizontal spray distribution, yet further harmonization is still required.

Spray deposition within trees is often evaluated using fluorescent tracers [17] or colored dyes [4], while droplet density and coverage can be measured on water-sensitive papers (WSP) [18]. Guo et al. [18] compared different WSP placement methods in pear trees and found that pole-mounted samplers recorded 1.5–2.4 times higher coverage than leaf-mounted ones. Li et al. [19] studied three pesticides in peaches and found imidacloprid had the best deposition efficiency. Yan et al. [20] demonstrated that drones provide relatively uniform droplet distribution within citrus canopies, outperforming knapsack sprayers, especially in upper canopy layers. Increasing spray volume or reducing flight speed improves coverage in lower canopy zones, though uniformity remains a challenge. Similarly, Biglia et al. [4] reported that optimal drone parameters can achieve effects comparable to orchard sprayers, though larger spray volumes (e.g., >55 L·ha⁻¹) may be necessary.

Spray drift remains a key issue. Wind tunnel experiments showed that rotary nozzles with droplets of DV50 111–277 μm produced higher drift risk than pressure nozzles [7]. Field trials using the QuanFeng120 UAV (Anyang Quan-Feng Aviation Crop Protection Science and Technology Co., Ltd., Quanfeng, China) indicated that at flight heights below 2.5 m, 90% of sediment drift occurred within 10 m, while at 3.5 m and moderate wind speeds, drift extended to 30–45 m [21]. Thus, maintaining flight heights below 2.5 m is critical for drift reduction. Drift magnitude also depends on droplet size and nozzle type: air induction nozzles produce coarser droplets and lower drift, particularly at moderate flight speeds [4].

Equipment contamination is another concern. The design and operation of sprayers influence internal and external contamination, which affects both operator safety and environmental pollution [22,23,24]. While numerous studies exist for ground sprayers, there are no published data on drone contamination. Research on manned aircraft [25] suggests that external residues after application are minor compared to tank residues, but similar studies are needed for UAVs.

Flight speed, height, nozzle type, and drone airflow all interact to determine spraying outcomes. Higher speeds (up to 3 m·s⁻¹) can improve canopy penetration for conventional nozzles [4], while slower speeds favor deposition uniformity [19]. Drone studies typically report optimal speeds of 1–3 m·s⁻¹ and flight heights of 1–3 m [26]. Increasing speed may enhance deposition with fine droplets but raises the risk of air drift [19]. Flight height varies widely by crop: from 1–2 m in low crops like rice and soybeans to 3–6 m in orchards and up to 10–11 m in tall trees [7,19,26,27,28,29,30,31,32,33,34,35,36,37,38]. Reducing height generally improves deposition and coverage uniformity [18].

Nozzle selection also plays a critical role. Droplet size depends on nozzle design, operating pressure, and liquid properties. Both rotary and pressure nozzles are used on drones, with higher pressures or rotational speeds producing finer droplets that enhance coverage but increase drift [2,20,39]. Injector nozzles and adjuvants have been shown to reduce drift [20]. In laboratory trials, double-jet injector nozzles provided higher deposition in young cherry trees than single-jet ones, largely due to enhanced airflow interaction [6]. Drone mass and rotor thrust also influence droplet trajectories and distribution patterns, with heavier drones producing stronger downwash that increases liquid deposition in lower canopy zones [6]. This rotor-generated airflow narrows and concentrates the spray stream, improving distribution uniformity.

In summary, UAV spraying presents a promising yet complex alternative to conventional orchard sprayers. Its efficiency depends on numerous interacting factors—flight parameters, nozzle configuration, airflow, and environmental conditions. Compared to ground sprayers, drones offer flexibility and accessibility in difficult terrains and can achieve comparable deposition when properly optimized. However, issues of drift, non-uniform canopy coverage, equipment contamination, and regulatory constraints remain significant challenges. Continued research and standardization of testing methodologies are essential to enable the safe and effective integration of UAV spraying technologies into sustainable orchard management systems [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41].

2. Materials and Methods

2.1. Experimental Field

The experiment was conducted in 27 May and 18 September 2024 in Skierniewice, in an apple orchard located at the Experimental Orchard of the Research Institute of Horticulture–National Research Institute. The experimental plot (51°57′37.4″ N, 20°09′37.1″ E) consisted of ten tree rows (55 m in length each) oriented along the north–south axis. The plot included two buffer zones—drift areas—extending along the full length of the rows and positioned eastward and westward relative to the tree line.

Measurements were carried out in a section of the orchard established in 2016, planted with ‘Gala’ apple trees grafted on M9 rootstock. The trees were trained in a spindle canopy system and planted at a spacing of 3.5 × 1.0 m, with average tree dimensions of 3.5 × 1.4 m (height × width). The orchard was equipped with a trellis structure designed for deploying an anti-hail net at a height of 4.5 m above ground level (Figure 1a,b). During the experiment, the anti-hail net was rolled up along the rows, forming strips approximately 25 ÷ 30 cm wide located at the mounting height above the row axis.

2.1.1. Spraying

For all measurements (drift, air and sediment drift, and deposit on spray equipment), for each experimental combination, a single spraying was performed on five consecutive rows of trees, located on the outermost leeward side of the experimental plot. On the first date, canopy deposition and air drift were measured. On the second date, canopy deposition, sediment drift, and contamination—spray equipment deposition—were measured. On both dates, spraying was performed for three drone combinations (flight speeds: 1.8, 2.7, and 3.6 m·s⁻¹—combinations: Drone-1.8; Drone-2.7 and Drone-3.6) and one orchard sprayer combination (combination: Orch.Spr.-1.7). The speeds were selected to take into account a speed similar to that used when spraying with an orchard sprayer (1.8 m·s⁻¹) and speeds 50% and further 50% higher found in the literature on orchard spraying with drones. Comparing the spraying results achieved with a drone and an orchard sprayer typically used in Polish orchards was aimed at answering the question of whether the risks posed by a drone (drift and equipment contamination) are comparable to those of an orchard sprayer (it was assumed that drift is higher and contamination could be smaller) and, if acceptable, whether the quality of spray application in the trees would be sufficiently high and uniform (it was assumed that it is worse). The differences between the spraying dates were in the liquid dose for the drone (27 vs. 40 L·ha⁻¹) and the flight altitude of 8 ÷ 9 m above ground level (AGL) in May and 7 ÷ 8 m AGL in September. The change in the spray liquid dose resulted from the assumption that the trees are more developed in the second term. The change in flight altitude resulted from an improved pilot-observer position in the second date, when the pilot gained better visibility of the drone’s flight thanks to the use of a special scaffolding (Figure 1c). In this situation, the pilot did not have to—for safety reasons—respond to flight altitude fluctuations by correcting the flight altitude.

The spray liquid was a water solution of fluorescent tracer BF7G—Acid brilliant flavine 7G (Waldeck GmbH & Co. KG, Münster, Germany). This tracer is characterized by high photostability, easily dissolves in water, and is easily washed from the surfaces of leaves and artificial targets. To maintain the tracer dose per hectare, an appropriately concentrated tracer concentration was used for drone spraying. Spraying parameters are listed in

Table 1. Combinations and operating parameters of spraying equipment.

Parameter	Combination
	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
Flight/Ground Speed [m·s−1]	1.8	2.7	3.6	1.7 (6.0 km·h−1)
Flight height AGL [m]—May	8–9			N/A
Flight height AGL [m]—September	7–8			N/A
Spray volume [l·ha−1]—May	27			400
Spray volume [l·ha−1]—September	40			400
Nozzle [type] Pressure [bar]	Rotational CDA			Lechler TR 80 15 @ 6.6 bar
Nozzle number	2			16
Droplets VMD [µm]	195			ca 150
Tracer dose—BF7G [g·ha−1]	1200 g/ha

.

2.1.2. Weather Conditions Measurement

Weather conditions were measured during spraying in two ways. To provide immediate feedback, a VelociCALC Plus model 8386A-M-GB handheld temperature anemometer (TSI Inc., St. Paul, MN, USA) was used. Air temperature and humidity (RH%) were measured (once per combination) as well as the instantaneous wind speed during spraying of each row of trees (when the tractor/drone was in position opposite to the drift measurement zone). The approximate wind direction, within or outside of ±30° from the direction perpendicular to the line of travel/flight was determined based on a mechanical indicator (ribbon) placed on a mast at a height of approximately 2.5 m from the ground and observed by the person recording the wind parameters. In addition, to obtain precise information, wind speed and direction were recorded continuously. Wind conditions were measured using two 3-axis ultrasonic anemometers, Gill WindMaster, model 1590-PK-020/w (Gill Instruments Limited, Lymington, Hampshire, UK) placed on a mast. The parameters of the anemometers were: 20 Hz output rate, 0–50 m/s wind speed, 0–360° wind direction. The anemometers were mounted at two heights. On the first date, the anemometers were mounted at heights of 3 and 5 m AGL to reflect the typical measurement height of a hand-held anemometer (approximately 2.5 ÷ 3.0 m), and at 5 m—just above the hail net. On the second date, the anemometers were placed at heights of 3 and 8 m AGL to measure air velocity in the drone’s flight altitude zone. The anemometers were connected to a two-home-made channel data logger (developed by Andrzej Bartosik member of the Agroengineering Department team), battery-powered and based on an ARM microprocessor. The data logger allows for recording parameters from two anemometers, connected via a serial communication bus and to save the measured values of wind speed and direction to a file on a memory card. The mast with the anemometers was positioned each time at a distance of approx. 10 m from the sprayed orchard (on its northern side, on the extension of the third outermost row (middle sprayed row). The anemometers were directed in such a way that was consistent with the cardinal directions (direction 0° = north direction). The rows of sprayed trees grew in a direction parallel to the north–south axis, and during the experiment the wind was blowing from the east (both dates). Therefore, the 270° direction for positioning the anemometers corresponded to the direction perpendicular to the direction of the drones’ flight or the sprayer’s movement (direction of the rows).

2.2. Spraying Equipment

The experiment used a spraying drone and, as a control, an orchard sprayer with a column attachment, representing the most common technical solution in Polish orchards.

2.2.1. Drone—Technical Parameters and Spraying Parameters

The commercially available UAV was used in this study: the ABZ Innovation drone (ABZ INNOVATION, Szentendre, Hungary). The drone has a quadcopter arrangement, with a tank capacity of 10 L. The UAV utilize two rotational CDA (Controlled Droplet Application) nozzles placed directly under the rear 2 rotors. The UAV was controlled using a radio controller (universal Herelink Controller Unit, Cube Pilot Pty. Ltd., Geelong, Australia). For used drone, the flight planning software allowed for pre-programmed flight paths to be created and utilized during the testing. The drone cooperated with the Multi-band RTK station (Emlid Tech Kft., Budapest, Hungary). Drone spraying was performed after mapping the to be sprayed area (drone with camera and GPS), then a flight trajectory was determined on a geo-positioned image—along the rows—with the flight line offset by 70 cm upwind relative to the tree line axis. The upwind offset distance was chosen to allow the LiDAR-based altitude system to “see” the ground surface and then better control the drone flight altitude. The detailed specifications for the drone used are provided in

Table 2. Technical parameters of the ABZ Innovation L10 drone.

General Specifications of ABZ Innovation L10
Total weight (without batteries)	13.6 kg
Max. Take-off weight	29 kg
Dimensions	1460 × 1020 × 610 [mm]
GPS	GPS, GLONAS, Galileo
Hovering precision	±10 cm (RTK) ±2 m (without RTK)
Battery capacity	16,000 mAh
Battery voltage	44.4V
Battery weight	4.7 kg
Spraying
Per Hectare Performance	10 ha/h
Spraying system	rotational CDA
Number of nozzles	2
Droplet size (adjustable)	40–1000 µm
Adjustable working width	1.5–6.0 m
Pump type	Membrane
Maximum liquid flow	5 L·min−1
Pump operating voltage	48 V
Flight
Max. Pitch angle	30°
Max operating flight speed	7 m·s−1
Max level speed	24 m·s−1
Max flight altitude	120 m
Max tolerable wind speed	10 m·s−1
Altitude measurement	LIDAR

.

The drone flew at speeds of 1.8, 2.7, and 3.6 m·s⁻¹ at an altitude of 8 ÷ 9 m AGL on the first date and 7 ÷ 8 m on the second date. On the first date (May), the drone sprayed a spray volume of 27 L·ha⁻¹ (1/15 of the orchard sprayer dose). On the second date, for more established trees, a dose of 40 L·ha⁻¹ (1/10 of the orchard sprayer dose) was used. A droplet size of 195 µm was always used.

2.2.2. Orchard Sprayer—Technical Parameters and Spraying Parameters

As a standard equipment, the Munckhof orchard sprayer (Munckhof Fruit Tech Innovators, Horst, The Netherlands, Figure 2) was used. The sprayer featured a column attachment and 20 nozzles (10 per side).

The sprayer was attached to a orchard tractor). Lechler TR 80 015 fine-droplet hollow cone nozzles (Lechler GmbH, Metzingen, Germany) operating at a pressure of 6.6 bar, with a flow rate of 0.875 L·min⁻¹, were used for spraying. The fan gear operated at a slower speed. The fan gear ratio was selected after an orchard test to minimize liquid blowing through the tree canopy.

2.3. Field Measurements

2.3.1. Spray Deposit Measurement

Spray deposit measurements were conducted on two dates (27 May and 18 September), corresponding to the early growth phase—after flowering and full leaf development. The experiment year (2024) was unusual for Poland, with vegetation accelerated by approximately 3–4 weeks. Spraying effects for drone and a Munckhof orchard sprayer were compared—details in Section 3.2.

In-crown spray deposit was measured using the filter paper samples 2.0 cm × 4.0 cm. The quality filter paper used in the experiment (Ahlstrom Germany GmbH, Munich, Germany).

To determine the amount of spray liquid available in different zones of the sprayed trees (drift potential), six trees were selected in the last downwind row of the sprayed quarter. The trees formed three groups of two trees, randomly placed along the sprayed row. Three metal poles (10 mm in diameter) were placed in each tree, positioned in the windward, axial, and leeward layers of the tree (Figure 3). Four horizontal aluminum plates (collector plates) measuring 60 × 25 mm were rigidly attached to each pole. The collector plates were mounted at four heights (approx.: 260; 200; 140 and 80 cm from the ground level). The vertical arrangement of measurement points included the highest crown area, the lowest crown area, and two intermediate points evenly spaced between these two extremes. The plates were attached to crocodile clips placed on mounting elements attached to the mast (Figure 3a). This attachment method allowed the collector plates to be positioned horizontally (rigidly). Filter paper samples were attached to the collector plates, to their upper and lower surfaces, using paper clips. A total of 12 measurement points were placed in each tree: four in the windward layer of the tree crown (locations 9–12), four in the axial layer (locations 5–8), and four in the leeward layer (directed towards the spray drift zone, locations 1–4).

To speed up work in the orchard, filter paper was attached to the collector plates in the laboratory. The prepared plate-and-paper assemblies were simply attached in their locations. After placing all samples at measurement points the spraying was performed (see Section 2.1.1). After spraying, the filter paper samples were collected (subtracted from the collector plates) in 45 mL sealed plastic containers and stored under cover until transport to the laboratory.

2.3.2. Spray Drift Measurement

The experiment was planned as the first in a series of experiments conducted on various fruit crops, so it was assumed that current and future drift measurements (to maintain comparability of research results) would be performed for air drift—measured on masts placed 5 m from the tree line. This assumption stemmed from concerns about the lack of available space in the drift zones to arrange samples at a sufficient distance—in our studies, samples are typically placed up to 28 ÷ 30 m from the spray zone boundary. The boundary of the beginning of the spray zone (also for drone measurements) was defined by the axis of the tractor’s driving line when passing outside the last row (Figure 4).

After conducting the first series of measurements (in May), it was decided to replace the air drift measurement with sediment drift measurement. This was possible because the experiment was conducted in a facility specifically designed for sediment drift testing (see Section 2.1).

Air Drift Measurements

Air losses of spray were measured on 27 May, using special masts (Figure 5), to which were mounted horizontal crossbars, each 1.0 m long, mounted at eight heights (1 ÷ 8 m, every 1 m). Openwork spherical samples (plastic string scourers—Figure 5) with a (maximum) cross-sectional area of 40.1 cm² were attached to the end of the crossbars using special forks (Figure 5). Two masts were used, with two vertical measurement lines on each (a total of four vertical measurement lines were placed 5.0 m from the last row of trees in the orchard spray zone). After spraying, the spherical samples were placed in 1.0 L plastic jars with screw caps and stored under cover until transport to the laboratory.

Sedimentation Drift Measurements

Sedimentation drift measurements were conducted on 18th September at ground level, in Petri dishes with a diameter of 14 mm. The dishes were placed on flat metal stands placed on the ground in the drift zone in 10 rows (every 1.0 m—Figure 6). In each row (replicate), the dishes were placed at distances from the sprayed area of: 0; 1; 2; 3; 4; 5; 7.5; 10; 15; 20; 28 m (Figure 6). The boundary between the spray zone and the drift zone (distance 0.0 m) was determined based on the axis line of the orchard sprayer passing outside the orchard (spraying the outermost row from outside the orchard—Figure 4).

The dishes were carried in special carriers, each holding 11 dishes with lids (Figure 6b). After spraying (see Section 2.1.1), the dishes were collected, placed back in the carriers, and stored covered until transport to the laboratory.

2.3.3. Measurements of Spraying Equipment Contamination (Drone and Sprayer)

The measurement of pollution of the drone and the orchard sprayer with tractor was carried out indirectly, using filter paper samples attached to the surface of the equipment used (Figure 7).

On the key components of both types of spraying machines, filter paper samples with a known surface area were attached using double-sided adhesive tape. On the drone, 38 samples were mounted with a surface area depending on the available space, measuring 36 or 72 cm² (

Table 3. Location and characteristics of samples and measurement points for measuring deposition of the tracer BF7G on the ABZ Innovation L10 drone during spraying the orchard. Skierniewice, 18 September 2024.

Description of the Sample Placement Location	Samples			Drone Area Represented (cm2)
	No.	Area (cm2)	Dimensions (cm × cm)
Casing	1, 2	2 × 36	6 × 6	210.0
Tank	3–8	6 × 36	6 × 6	2287.6
Case	9–14	6 × 36	6 × 6	862.0
Propellers carriers	15–22	8 × 72	12.5 × 5.76	2968.4
Drone base (legs, crossbars)	23–34	12 × 36	6 × 6	1356.8
Nozzles holders	35–38	4 × 36	7.5 × 4.8, 6 × 6	531.2

). On the tractor and orchard sprayer, a total of 36 samples with a surface area of 64 cm² (8 × 8 cm,

Table 4. Location and characteristics of samples and measurement points for measuring deposition of the tracer BF7G on the Munckhof orchard sprayer and the cooperating Landini REX 90S tractor (Landini SpA, Fabbrico, Italy) during spraying the orchard. Skierniewice, 18 September 2024.

Description of the Sample Placement Location	Samples			Sprayer Area Represented (cm2)
	No.	Area (cm2)	Dimensions (cm × cm)
Sprayer
Fan	1–10	10 × 64	8 × 8	41,314.6
Tank	11–18	8 × 64	8 × 8	49,338.0
Sprayer wheels	19–22	4 × 64	8 × 8	5652.0
Tractor
Windows (left, right, front, rear)	23–30	8 × 64	8 × 8	3825.0
Tractor roof	31, 32	2 × 64	8 × 8	11,700.0
Tractor rear wheels	33–36	4 × 64	8 × 8	13,364.0

) were mounted. After spraying of the experimental orchard area (approx. 1000 m²) (see Section 2.1.1), the samples were removed and placed in 1.0 L plastic jars. They were then stored under cover until were transported to the laboratory.

Measurement points were located on surfaces that were easy to define and where it was possible to place filter paper samples. Other surfaces, on which no samples were placed, were measured or estimated (arcs and other curves were treated as straight lines). Due to the presence of surfaces with no samples, their area was only estimated and then expressed as a percentage of the defined surface. For example, in the case of a drone, placing filter paper samples on the propellers was not possible due to the risk of them being “blown away” during flight and the rotational movement of the propellers, and it might have hindered the generation of lift by the propellers obstructed by an “aerodynamic obstacle.” The propeller surface constituted in 30% of the defined surface.

In Section 2.4, there are presented the deposition data on the measured (defined) surfaces where the samples were placed on as well as the estimated/calculated areas of locations where no samples were placed. In addition, in the Drone-1.8 combination, the equipment was further contaminated while landed on the ground, when the drone’s surface contamination occurred additionally during the collection of the tank solution (Tank mix). It happened after finishing the spraying flight, but before taking off the samples. An attempt was made to correct the result for deposition at measurement points on the drone that were located near the rotary nozzles; however, it was not possible to obtain a reliable data. Therefore, for the Drone-1.8 combination, the contamination was defined as the maximum total one for both flight and stationary stage.

In the laboratory, the samples were flooded with 100 mL (drone) or 200 mL (sprayer) of deionized water. The difference in the amount of solvent was due to the size of the filter paper samples and the jars used for them (different volumes).

2.4. Laboratory Measurements

The analysis of the amount of marker deposited on the samples was carried out on subsequent days (after field measurements) in the laboratory. The samples were rinsed out with an appropriate amount of deionized water (20, 40, 100, 200 or 400 mL) depending on containers type and volume and samples used, and then they were shaken for 15 min on a device developed at the Department of Agroengineering. Quantitative analysis of marker concentration was conducted using a PerkinElmer LS55 spectrophotometer (PerkinElmer, Inc., Headquarter: Shelton, CT, USA; Made in: High Wycombe, UK) (Figure 8c). Prior to measurement, the spectrophotometer was calibrated using several times diluted Tank Mix and calibration curve was then obtained. The calibration curve allows the fluorescence intensity (signal) to be converted into the concentration of the fluorescent marker. An excitation wavelength of 457 nm and an emission wavelength of 512 nm were used. The measured concentration of the BF7G marker (ng·mL⁻¹) was converted into surface deposition (ng·cm⁻²)—taking into account the volume of deionized water used to rinse the samples and the surface area of the samples or dishes (filter paper: 8 cm², dishes: 151.7 cm², on drone samples: 36 or 72 cm², or on sprayer and tractor ones: 64 cm², and spherical samples: 40.1 cm²). The unit deposition obtained for different locations (trees, poles, ground, spraying equipment) was subjected to further calculations and statistical analysis.

2.5. Statistical Analyses and Other Calculations

The obtained results of deposition on masts placed within the tree crowns were developed as average values for

• deposition on the upper surfaces of the leaves—U;
• deposition on the lower surfaces of the leaves—L;
• total deposition on both leaf surfaces—U + L;
• the ratio of deposition on the upper leaf surfaces to deposition on the lower leaf surfaces—U/L.

For each of the values (U, L, U + L, U/L), the following were calculated:

• average values for entire trees;
• average values in the upper zone of trees (the two upper measurement locations on the masts, Figure 9a) (Tree Top—TT);
• average values in the lower zone of trees (the two lower measurement locations on the masts, Figure 9a) (Tree Bottom—TB);
• average values in the windward layer of the tree (Figure 9b) (Tree Windward—TW);
• average values in the leeward layer of the tree (Figure 9b) (Tree Leeward—TL).

For each of the values (U, L, U + L), uniformity indices of deposition in the tree crowns were calculated:

• the ratio of the deposition value in the upper zone of the trees to the average deposition value in the lower zone (Figure 9a) Tree Top/Bottom (T-T/B);
• the ratio of deposition in the windward zone of sprayed trees to the deposition in the leeward zone of sprayed trees (Figure 9b) Tree Windward/Leeward (T-W/L).

For airborne drift (measured on masts), the unit deposition was calculated, which was related to the (largest) cross-sectional area of the used samples—40.1 cm²—and then converted into drift values related to the applied dose of the BF7G tracer per unit area of the sprayed orchard (applied: 1200 g·ha⁻¹ = 12,000 ng·cm²). The results were processed as average loss values (in relation to the applied tracer dose) at different measurement heights (1–8 m) and as a proportion (ratio) of losses in the upper and lower zones of the masts.

Sedimentary drift was subject to similar calculations. In the first stage, the unit deposition was calculated taking into account the area of the Petri dishes (151.7 cm²) and the measured concentration of the tracer for each sample. Then, the obtained unit deposition values were converted to the drift value as a percentage of the applied dose of the BF7G tracer per unit area of the sprayed orchard (1200 g·ha⁻¹ = 12,000 ng·cm⁻²). In the next stage, the cumulative sedimentary drift for the individual measurement distances was calculated. In this final calculation, the unit deposition (as a percentage of the applied dose) and the width of the strip represented by the given samples (1.0 m or 2.5 m or 5.0 m or 8.0 m) were taken into account. The cumulative sedimentary drift was calculated assuming some simplification, that the drift value changes between subsequent drift measurement positions in linear manner. Additionally, the values of sedimentary drift were presented as a percentage of the applied dose at individual measurement distances (1.0; 2.0; 3.0; 4.0; 5.0; 7.5; 10.0; 15.0; 20.0; and 28.0 m).

To assess differences between means, a statistical analysis was performed. The obtained data were analysed using a multi-way analysis of variance (ANOVA) to determine the effect of the spraying technique (drone flight speed or orchard sprayer) on spray deposit in the trees and spray drift to the air and to the ground. All statistical analyses were performed using STATISTICA 13. The treatment means were separated by Duncan’s Multiple Range Test at the significance level p < 0.05. Statistical analyses were not conducted for deposition on the spraying equipment because measurements were performed with only one repetition for each combination.

For contamination of the spraying equipment, the total deposition on the elements subject to measurement was calculated (the sum of the products of the unit deposition on a given element and the area of that element), and additionally the total area of elements where samples were not placed was calculated.

3. Results and Discussion

3.1. Weather Conditions

The results of atmospheric wind direction and velocity measurements are presented in the figures and tables (Figure 10 and Figure 11;

Table 5. Wind speed and direction, air temperature, and relative humidity during drone and orchard sprayer applications. Skierniewice, 27 May 2024.

Parameter/Combination	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
Automatic measurement of wind direction and wind speed [m·s−1] parameters
Mean direction at 3 m AGL	293°	287°	282°	292°
Mean direction at 5 m AGL	300°	287°	286°	294°
Wind dir. consist. (R) at 3 m AGL	0.89	0.88	0.89	0.92
Wind dir. consist. (R) at 5 m AGL	0.95	0.89	0.94	0.94
Speed (10th–90th percentile) at 3 m	1.47–4.70	1.56–4.05	1.61–4.22	1.55–3.45
Speed (10th–90th percentile) at 5 m	2.03–5.90	1.84–5.38	2.12–5.06	2.16–4.56
Mean speed at 3 m AGL	2.84	2.74	2.83	2.41
Mean speed at 5 m AGL	3.93	3.46	3.75	3.29
Wind speed CV at 3 m AGL [%]	43.3	34.8	35.9	30.7
Wind speed CV at 5 m AGL [%]	37.9	38.8	30.5	27.2
Spraying duration [s]	211	107	90	318
Manual measurement of wind speed and direction, temperature and humidity
Mean speed at 2.5 m AGL Speed range (min.–max.)	2.76 0.9–5.5	3.04 0.4–5.3	3.62 1.7–5.6	2.43 1.2–3.6
Wind direction range at 2.5 m AGL	280–315°	280–310°	280–310°	280–310°
Air temperature [°C]	31.2	30.4	32.1	32.1
Relative air humidity [%]	26.5	20.8	17.9	19.4

and

Table 6. Wind speed and direction, air temperature, and relative humidity during drone and orchard sprayer applications. Skierniewice, 18 September 2024.

Parameter/Combination	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
Automatic measurement of wind direction [°] and wind speed [m·s−1] parameters
Mean direction at 3 m AGL	234°	236°	208°	213°
Mean direction at 8 m AGL	228°	231°	206°	214°
Wind dir. consist. (R) at 3 m AGL	0.91	0.94	0.92	0.93
Wind dir. consist. (R) at 8 m AGL	0.97	0.99	0.96	0.97
Speed (10th–90th percentile) at 3 m	1.01–2.78	1.22–2.87	1.26–3.64	1.22–3.36
Speed (10th–90th percentile) at 8 m	2.09–3.79	2.45–3.52	1.96–4.58	1.60–4.40
Mean speed at 3 m AGL	1.94	2.10	2.36	2.33
Mean speed at 8 m AGL	2.96	3.00	3.17	3.18
Wind speed CV at 3 m AGL [%]	35.5	32.0	37.7	36.3
Wind speed CV at 8 m AGL [%]	22.3	13.9	30.2	31.2
Spraying duration [s]	144	104	87	327
Manual measurement of wind speed [m·s−1] and direction [°], temperature and humidity
Mean speed at 2.5 m AGL	1.90	1.82	1.48	1.43
Speed range (min.–max.)	0.9–3.1	1.2–2.5	0.5–2.4	0.9–1.7
Wind direction range at 2.5 m AGL	250–270°	250–270°	250–270°	250–270°
Air temperature [°C]	26.1	26.8	27.2	28.6
Relative air humidity [%]	42.6	43.5	40.3	36.6

), while air temperature and relative humidity values are included in the tables. To reduce the influence of extreme wind speeds, the wind velocity was expressed as the range between the 10th and 90th percentiles of the recorded data, as well as the mean value across the entire measurement period. Automatically recorded wind speeds (measured at 20 Hz frequency) differed from those obtained using a handheld anemometer by an average of 9.2% during the first measurement term and by approximately 35% during the second term, when the wind was about 0.5 m·s⁻¹ slower. Interestingly, in the first term, three out of four handheld measurements were overestimated, whereas in the second term, they were underestimated. This suggests that a more precise real-time measurement of wind velocity during spraying could be beneficial; however, continuous monitoring of wind direction is even more critical to meet ISO standards and to ensure that spraying is conducted under comparable wind direction ranges for all treatment combinations. A potential solution to this issue could involve repeating the entire experimental setup over time, although for drift measurements, identical wind strength and direction conditions cannot be guaranteed.

To assess the variability of instantaneous wind speeds, the coefficient of variation (CV) was calculated (Table 5 and Table 6). In the first measurement term, CV values ranged from 30.7% to 43.3%, while in the second term it ranged from 32.0% to 37.7%. The mean wind speeds at a height of 3 m above ground level (AGL) varied between 2.41 and 2.84 m·s⁻¹ in the first term and between 1.94 and 2.33 m·s⁻¹ in the second one. This represents a relatively narrow range, indicating that spraying operations during both terms were conducted under comparable wind speed conditions. Similarly, the range between the 10th and 90th percentiles of all wind speed measurements was narrower during the second term (1.01–3.64 m·s⁻¹) than during the first one (1.47–4.70 m·s⁻¹).

The wind direction measured using the mechanical indicator did not exceed the allowable deviation specified in ISO 22866:2005 [39], remaining mostly within ±30° from the direction perpendicular to the driving line, and no more than 30% of measurements deviated by more than ±45° from this direction. Wind direction measured automatically with a set of two 3-axis ultrasonic anemometers is presented as circular plots and numerically as the mean wind direction (angle in the 0–360° range; Figure 10 and Figure 11). The mean wind direction determined from the mechanical indicator observations (Table 5 and Table 6) differed from the automatic measurements (mean value) mainly in the second measurement term (approximately 10–20% vs. 1.0–4.6%).

For the 3-axis anemometer measurements, 0°/360° represents the north direction. Considering the row orientation of the trees along the north–south line, an easterly wind corresponds to an acceptable wind direction according to ISO 22866:2005, ranging from 240° to 300°. The permissible deviation (>±45°) corresponds to directions outside the 225–315° range. Most combinations were conducted under wind conditions within the standard requirements. Only during the second measurement term (Figure 11c,d, Table 6) it was measured that the mean wind direction deviation was above the ISO [39] limit for the drone at the highest flight speed and for the orchard sprayer.

Since the analysis of wind parameters recorded by the ultrasonic anemometer system is performed after spraying (sometimes with a delay of several days), these data can only be used to interpret the obtained results, mainly for drift measurements and, occasionally, for deposition within the trees. Readings from the mechanical indicator (placed nearby, approximately 7 m) did not provide a basis for suspending spraying. Acquisition of a more precise mechanical wind indicator is planned to allow real-time precise control of wind direction.

Wind direction stability was described using the wind direction consistency coefficient R (mean resultant length R, Equation (1); Table 5 and Table 6), where a value close to 1 indicates stable wind direction, and a value close to 0 indicates complete instability. In the first measurement term, R values ranged from 0.89 to 0.92 at 3 m AGL and from 0.89 to 0.95 at 5 m AGL (Table 5), indicating greater stability of wind direction at higher elevations. A similar trend was observed during the second measurement term, in which the upper sensor was mounted at a higher position (8 m AGL, Table 6).

$$R=\sqrt{{\overline{x}}^2+{\overline{y}}^2}$$

(1)

where

• $\overline{x}$—mean value of the cosines of the measured wind direction angles;
• $\overline{y}$—mean value of the sines of the measured wind direction angles.

Certain discrepancies between manual and automatic wind speed and direction measurements indicate, first, the need to increase the frequency of manual measurements. They also may suggest that a planned assessment should be conducted to evaluate the spatial variability of these parameters at different locations around the orchard during spraying operations.

3.2. In-Tree Deposition

Due to the large number of analyses of variance (26 ANOVAs for

Table 7. Mean total deposition on both collector plates surfaces (U + L) in trees (ng·cm⁻²) for different tree zones: entire trees, upper (TT) and lower (TB) zones, windward (TW) and leeward (TL) layers, and deposition ratios between zones (T-T/B, T-W/L) for drone spraying at three flight speeds and the Munckhof orchard sprayer. Skierniewice, 27 May and 18 September 2024.

Tree Zone	Term	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
The whole tree	1	1380.5 a	1992.8 a	1607.2 a	2722.4 b
	2	3798.6 c	1573.4 a	1507.2 a	3736.4 c
Lower tree zone TB	1	629.8 a	1293.9 b	797.5 ab	2280.9 c
	2	2640.2 c	890.9 ab	903.3 ab	2570.7 c
Upper tree zone TT	1	2131.1 a	2691.7 a	2416.8 a	3164.0 a
	2	4957.0 b	2255.9 a	2111.2 a	4902.2 b
Leeward layer TL	1	1160.3 a	763.1 a	910.7 a	2539.1 b
	2	3156.2 b	1203.7 a	1207.3 a	3162.2 b
Windward layer TW	1	2118.1 a	4432.2 cd	3186.0 a–c	3723.4 bc
	2	5953.8 e	2393.6 ab	2185.2 ab	5256.0 de
Ratio T-T/B	1	3.40 c	2.30 a–c	2.96 bc	1.40 a
	2	1.97 ab	2.55 a–c	2.45 a–c	1.94 ab
Ratio T-W/L	1	2.71 a	8.89 b	3.78 a	1.47 a
	2	2.65 a	2.20 a	2.17 a	1.67 a

Means for tree zone categories (separated by horizontal lines) marked with the same lowercase letter (a–e) do not significantly differ for Duncan multiple range test p < 0.05.

,

Table 8. Mean deposition on upper collector plates surfaces (U) in trees (ng·cm⁻²) for different tree zones: entire trees, upper (TT) and lower (TB) zones, windward (TW) and leeward (TL) layers, and deposition ratios between zones (T-T/B, T-W/L) for drone spraying at three flight speeds and the Munckhof orchard sprayer. Skierniewice, 27 May and 18 September 2024.

Tree Zone	Term	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
The whole tree	1	1253.0 a	1905.8 a	1473.6 a	1251.8 a
	2	3472.0 b	1433.2 a	1337.4 a	1797.2 a
Lower tree zone TB	1	521.1 a	1227.6 cd	662.9 ab	1185.9 b–d
	2	2154.0 e	757.6 a–c	802.0 a–c	1356.0 d
Upper tree zone TT	1	1985.0 a	2584.1 a	2284.4 a	1317.6 a
	2	4790.0 b	2108.9 a	1872.8 a	2238.3 a
Leeward layer TL	1	1073.6 a	693.1 a	764.7 a	1277.1 a
	2	2826.2 b	1094.4 a	1021.8 a	1512.1 a
Windward layer TW	1	1901.9 a	4313.2 b	2983.7 a	1623.0 a
	2	5597.4 b	2190.4 a	1995.7 a	2374.7 a
Ratio T-T/B	1	4.1 c	2.3 ab	3.6 bc	1.2 a
	2	2.4 ab	2.8 a–c	2.5 a–c	1.7 a
Ratio T-W/L	1	2.6 a	9.6 b	4.9 a	1.3 a
	2	3.0 a	2.2 a	2.3 a	1.8 a

Means for tree zone categories (separated by horizontal lines) marked with the same lowercase letter (a–e) do not significantly differ for Duncan multiple range test p < 0.05.

,

Table 9. Mean deposition on lower collector plates surfaces (L) in trees (ng·cm⁻²) for different tree zones: entire trees, upper (TT) and lower (TB) zones, windward (TW) and leeward (TL) layers, and deposition ratios between zones (T-T/B, T-W/L) for drone spraying at three flight speeds and the Munckhof orchard sprayer. Skierniewice, 27 May and 18 September 2024.

Tree Zone	Term	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
The whole tree	1	127.45 a	86.95 a	133.51 a	1470.66 b
	2	326.61 a	140.17 a	169.89 a	1939.29 c
Lower tree zone TB	1	108.75 a	66.33 a	134.61 a	1094.96 c
	2	486.19 b	133.29 a	101.31 a	1214.61 c
Upper tree zone TT	1	146.15 a	107.57 a	132.41 a	1846.35 b
	2	167.03 a	147.05 a	238.47 a	2663.97 c
Leeward layer TL	1	86.6 a	70.0 a	146.0 a	1262.0 b
	2	330.0 a	109.2 a	185.4 a	1650.1 c
Windward layer TW	1	216.2 a	118.9 a	202.4 a	2100.4 b
	2	356.3 a	203.2 a	189.5 a	2881.2 c
Ratio T-T/B	1	2.05 ab	1.77 ab	1.38 ab	1.88 ab
	2	0.36 a	1.24 ab	2.35 c	2.21 c
Ratio T-W/L	1	3.58 a	2.38 a	1.98 a	1.73 a
	2	1.33 a	2.01 a	1.48 a	1.73 a

Means for tree zone categories (separated by horizontal lines) marked with the same lowercase letter (a–c) do not significantly differ for Duncan multiple range test p < 0.05.

and

Table 10. Ratio of deposition on collector plates surfaces (L) in trees for different tree zones: entire trees, upper (TT) and lower (TB) zones, windward (TW) and leeward (TL) layers, for drone spraying at three flight speeds and the Munckhof orchard sprayer. Skierniewice, 27 May and 18 September 2024.

Tree Zone	Term	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
The whole tree	1	18.73 c	24.89 c	19.33 c	1.51 a
	2	22.47 c	11.24 b	11.75 b	1.48 a
Lower tree zone TB	1	7.25 ab	21.65 c	12.83 b	1.92 a
	2	9.96 b	6.71 ab	12.99 b	1.76 a
Upper tree zone TT	1	30.22 d	28.13 d	25.83 cd	1.09 a
	2	34.98 d	15.77 bc	10.50 ab	1.21 a
Leeward layer TL	1	19.63 bc	10.61 ab	8.00 a	1.96 a
	2	23.78 c	11.57 ab	9.14 a	1.23 a
Windward layer TW	1	22.27 bc	51.53 e	37.57 de	1.07 a
	2	29.92 cd	11.96 ab	12.83 ab	1.30 a

Means for tree zone categories (separated by horizontal lines) marked with the same lowercase letter (a–e) do not significantly differ for Duncan multiple range test p < 0.05.

), only the significance of the main factors (treatment combination and application term) and, in some cases, their interactions are discussed. For entire trees, the factor “sample location” was also considered, although it was partially included in the analyses for deposition uniformity parameters: T-T/B, T-W/L, and U/L (abbreviations explained in Section 2.4).

For total deposition (U + L) in entire trees, the application term (df = 2), combination (df = 4), and sample location within the tree (df = 12) were highly significant (p < 0.0000). Similar results were observed for the upper (TT) and lower (TB) tree zones when analyzed separately. The ratio of U + L deposition in the upper zone to that in the lower zone (T-T/B) was significant for the treatment combination, but the application term was not statistically significant. For several deposition assessment parameters, similar ANOVA results were obtained, indicating that the treatment combination was significant, whereas the application term was not. The May and September application terms differed in tree growth stage and, to a minor extent, in wind speed and direction during spraying (see Section 3.1) or flight height. Statistical analyses showed that the application term was not significant for the following:

• Total deposition (U + L): in the windward layer (Table 7);
• Deposition on upper surfaces (U): for the ratio of deposition in the upper zone to deposition in the lower zone (T-T/B) and for deposition in the windward layer (TW) (Table 8);
• Deposition on lower surfaces (L): for the T-T/B ratio and for the T-W/L ratio (the combination was also not significant, similarly for deposition in the windward zone) (Table 9);
• U/L uniformity index: in the upper tree zone (TT), the lower tree zone (TB), and in the leeward layer (TL) (Table 10).

In the May application term, for the whole trees the mean total deposition on both leaf surfaces (U + L) was higher for the orchard sprayer, ranging from 1.37 times more compared to the drone at a flight speed of 2.7 m·s⁻¹ up to 1.97 times more for the drone at 1.8 m·s⁻¹—statistically significant differences only with the orchard sprayer (Table 7).

In the second term, a statistically similar level of U + L deposition as for the orchard sprayer was achieved by the drone at the lowest flight speed (Drone-1.8). This may indicate the possibility of achieving comparable deposition, as well as similar uniformity between the upper and lower tree zones and between the windward and leeward layers, when spraying with the drone at lower flight speeds compared to the ground-based orchard sprayer. This is consistent with Biglia et al. [4] opinion after trials in vineyards, who suggested that it is possible to obtain similar spraying effects using drones (best combinations–settings) as for orchard sprayer, but they point out the need to increase the liquid doses. Higher flight speeds resulted in for drone deposition more than twice as low as that of the orchard sprayer. Similar trends were observed for the upper and lower tree zones analyzed separately.

Due to the partially one-sided nature of drone spraying (the drone flight trajectory was offset by approximately 70 cm “upwind” relative to the axis of the sprayed trees), the ratios of total deposition (U + L) in the windward zone to deposition in the leeward zone of the sprayed trees were calculated. Relatively high uniformity of this parameter was observed for the orchard sprayer, with ratios ranging from 1.47 to 1.67, whereas the lowest uniformity was recorded for the drone spraying at 2.7 m·s⁻¹, with other for drone ratios ranging from 2.20 to 8.89 (Table 7).

Values of deposition and uniformity indices for upper leaf surfaces (U) and lower leaf surfaces (L) are presented in Table 8 and Table 9. Of particular note is the more than 11-fold lower deposition on lower collector plates surfaces (mean for entire trees) for all drone spraying combinations (86.95–133.51 ng·cm⁻²) compared to the orchard sprayer (1470.66 ng·cm⁻²). This confirms earlier our assumptions regarding the difficulty of covering lower leaf surfaces when spraying plants with a downward-directed spray stream from a drone flying at 8 ÷ 9 m above ground level.

In general, previous hypotheses regarding the effect of the vertical direction of drone spraying on the deterioration of deposition uniformity within sprayed trees of the quite big size (height × width = 3.5 × 1.4 m) were confirmed, particularly between the upper and lower tree zones (T-T/B) during the first term, similarly as results of coverage measurements achieved by Guo et al. [18]. It should be noted, however, that spraying was conducted under unfavorable conditions (support frame of the net and rolled-up anti-hail net) and performed from a height of 8 ÷ 9 m above ground (4.5 ÷ 5.5 m above tree canopies) with a flight trajectory offset “upwind” by approximately 70 cm. Under more favorable conditions (lower flight height, closer to the tree canopies), better deposition and uniformity parameters within the tree canopy may be achievable.

The use of drones with higher payload capacity is also relevant. Drones are categorized by manufacturers and the agricultural sector into: low payload (5 ÷ 10 kg), medium payload (10 ÷ 30 kg), and high payload (30 ÷ 100 kg). For spraying orchards with larger, mature trees, high-payload drones (30 ÷ 100 kg) appear to be most suitable, as the stronger airflow from the rotors can improve deposition uniformity within the tree. This should be one of the directions for research on drone applicability for orchard spraying: evaluating differences in effects achieved by drones with varying payload capacities.

Furthermore, extensive research on drone spraying is required, considering factors such as spraying method (single or double pass per row), liquid dose (e.g., from 1/15–1/20 to 1/10–1/5 of the dose used by the orchard sprayer), flight height (1 ÷ 2 m above the tree canopy up to 5 ÷ 6 m), tree leafing stage (including leafless periods), tree size (including nurseries), and spray droplet size VMD—from small to large/extremely large).

An important aspect of UAV orchard spraying is the possibility for the operator to directly observe the drone’s flight trajectory due to flight height and limited visibility caused by the trees, enabling immediate reaction in case of unexpected changes in speed or trajectory. Possible solutions include observation towers, allowing the operator’s line of sight to be at least 1.0 m above the sprayed tree row, or monitoring the drone’s flight via an onboard camera. Such an observation station (construction scaffolding) was acquired and used during the September measurement term.

The problem highlighted by Wang et al. [7] regarding the difficulty of comparing results from different experiments can be minimized by expressing the obtained deposition on sprayed plants as recovery, which refers to the applied dose. A similar approach was applied by Godyń et al. [42], where the authors used this method to compare treatments within their own experiment, which involved different doses and concentrations of the fluorescent tracer. In the present study, all deposition values should be divided by the applied BF7G tracer dose expressed in ng·cm⁻² (i.e., 12,000) and multiplied by 100% to express recovery. Ultimately, to present recovery as a percentage, each deposition value should be divided by 120 and such data may be used to be compared with other normalized data of similar experiments.

Given the observed deviations from the planned flight parameters, the development of methods to verify the accuracy of flight and spraying parameters during drone-based spraying experiments appears useful, if not essential. These parameters include flight height and speed (no issues were observed with flight direction), as well as the actual liquid output and applied droplet size (maintaining nozzle parameters: rotations for CDA or pressure for pressure-based nozzles). Some information—such as GPS/GNSS coordinates, horizontal and vertical speed, pump performance, and liquid flow rate—is recorded in a control file generated after the flight, often referred to as the Flight Log, Flight Record Data, or Operational Data Log. However, there remains a need to verify the accuracy of these records and their values at different time intervals during the flight (not only the mean values).

The uniformity of deposition between the upper and lower collector plates surfaces (U/L) was significantly worse for all drone-based treatments, with the lowest (the best for drone) value of 6.71 (in the lower tree zone) and the highest of 34.98 (in the upper tree zone), compared to the orchard sprayer, for which this index ranged relatively low (quite good), from 1.07 to 1.96 (Table 10). This is partly due to significantly lower deposition values on the lower collector plates surfaces for the drone, compared to the orchard sprayer, both for entire trees and across all separately analyzed zones (Table 9), whereas deposition on the upper leaf surfaces in the upper tree zone remained at a statistically similar level (and in one combination even higher) than for the orchard sprayer (Table 8).

To assess the influence of the orchard itself and additionally the anti-hail net, it is recommended to perform measurements in an orchard with smaller trees and in an orchard without obstacles for spraying, such as the supporting structure and the (even folded) anti-hail net.

3.3. Spray Drift

3.3.1. Airborne Drift

The ANOVA revealed a significant effect of the combination and the height of the sample position on the mast, with a statistically insignificant effect of the repetition (here: the vertical measurement line). This may suggest a relatively high vertical uniformity of the drifted liquid cloud, i.e., the vertical distribution of drifted liquid, measured at the same height. The mean coefficient of variation between repetitions at the same heights ranged from 29.51% to 38.79%.

Average losses (% of dose applied) measured on the masts were the lowest for the orchard sprayer (6.88%), while for drone spraying it ranged from 14.47% at 1.8 m·s⁻¹ to 41.08% at a flight speed of 2.7 m·s⁻¹ (

Table 11. Air drift at eight individual heights and for the upper and lower parts o the masts, as well as the drift ratio for the upper and for the lower part of the mast for drone spraying at three flight speeds and the Munckhof orchard sprayer. Skierniewice, 27 May 2024.

Height from the Ground [m]	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
8.0	0.94 a	2.04 a	8.15 ab	2.74 a
7.0	1.71 a	2.48 a	11.32 ab	3.90 a
6.0	3.18 a	7.07 ab	16.53 ab	4.43 a
5.0	6.03 a	14.18 ab	26.83 bc	6.49 a
4.0	7.95 ab	62.09 fg	48.61 d–f	7.80 ab
3.0	14.88 ab	75.82 g	41.97 c–e	9.69 ab
2.0	35.06 cd	101.41 h	56.09 ef	10.16 ab
1.0	46.02 d–f	63.52 fg	48.56 d–f	9.86 ab
Mean (1–8 m)	14.47 a	41.08 b	32.26 b	6.88 a
Upper mast (5–8 m)	2.97 a	6.44 a	15.71 c	4.39 a
Lower mast (1–4 m)	25.98 b	75.71 d	48.81 c	9.38 a
Ratio Upper/Lower	0.14 ab	0.09 a	0.33 bc	0.54 c

Means for tree zone categories (separated by horizontal lines) marked with the same lowercase letter (a–h) do not significantly differ for Duncan multiple range test p < 0.05.

). Lower drift (32.26% for the combination with the highest flight speed–3.6 m·s⁻¹) than for a speed of 2.7 m·s⁻¹ seems to be related to the rearward deflection of the spray liquid stream (relative to the flight direction) at this flight speed and, consequently, to the extension of the drift path relative to the measurement line length. Analysis of the automatic wind direction measurements confirms that atmospheric wind direction should not have an impact on the drift magnitude. For the tested combinations, the average deviation of the wind direction from the direction perpendicular to the flight/sprayer direction was within relatively narrow limits of 12–23° (282–293°, Section 3.1, Table 5).

Greater losses were measured in the lower part of the masts, particularly for the drone spraying combinations (3.11 ÷ 11.75 times more than in the upper part). The greatest uniformity of this parameter was observed for the orchard sprayer (only 2.14 times more). In this experiment, losses were initially measured on the masts, which was related to the need to ensure repeatable conditions for subsequent experiments in other crops (e.g., berry bushes or other orchards). This resulted from justified doubts about the possibility of finding experimental locations with sufficient space to conduct sediment drift measurements (at least 28 ÷ 30 m from the sprayed plants). Because the analysis of spray distribution on the masts (airborne drift) suggested the need to clarify how far the liquid visible in the results at the lower part of the masts was carried, so sediment drift measurements were performed in September.

3.3.2. Sedimentation Drift

ANOVA for total drift up to individual measurement distances showed high significance for both combinations and distances, but the interaction of these factors was insignificant (p = 0.3535). However, for drift at individual distances, both factors and their interaction were highly significant.

The highest total drift (totaling up to 28 m) was observed for the drone spraying at the lowest flight speed (1.8 m·s⁻¹,

Table 12. Sediment drift up to 28 m from the spray zone boundary: cumulative drift for individual measurement distances as a percentage of the applied dose [%] for drone spraying at three flight speeds and the Munckhof orchard sprayer. Skierniewice, 18 September 2024.

Distance [m]	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
1	9.8 a–d	5.9 a	6.6 ab	9.5 a–c
2	17.1 c–g	10.6 a–d	12.0 a–e	16.2 c–g
3	22.9 g–j	14.3 b–f	16.4 c–g	21.2 f–j
4	27.9 i–n	17.6 d–h	19.6 e–h	25.4 h–k
5	31.7 k–q	20.4 f–i	21.7 f–j	28.6 j–o
7.5	38.2 q–u	25.7 h–m	25.3 h–l	33.7 m–r
10	42.4 s–v	29.2 j–p	27.8 i–n	36.9 p–t
15	47.5 v–x	33.5 l–r	31.2 k–q	41.2 r–v
20	50.6 w–x	36.2 o–t	33.1 k–q	43.5 t–w
28	54.4 x	38.6 q–u	34.6 n–s	45.2 u–w

Means marked with the same lowercase letter (a–x) do not significantly differ for Duncan multiple range test p < 0.05.

), i.e., for the combination that achieved the highest spray deposit inside the sprayed trees (ex aequo with the orchard sprayer, Table 7). At the same time, the lowest sediment drift was observed for the remaining drone flight speeds (2.7 and 3.6 m·s⁻¹, Table 12, Figure 12).

Drawing clear conclusions from the relationship between air drift and sedimentation effects may be subject to some error, as both measurements were performed separately (at different times). Increasing flight speed, especially over the first outermost row, undoubtedly influences the deviation of the liquid stream from the spray target, which are trees, and increases air drift (measured only 5 m from the last tree line). This is especially true for sprayers with a relatively low maximum takeoff mass and weaker airflow from the propellers. The reduction in sedimentation drift (at all measurement distances) with increasing speed may be the result of increased settling of liquid drifted from rows deeper in the orchard (on trees and on the ground between trees). This hypothesis will be verified in subsequent studies.

However, the analysis of the automatically recorded wind direction values indicates that for the Drone-3.6 combination the wind direction changed somewhat and was deviated from the direction perpendicular to the drone’s flight direction more than for the flight at a speed of 2.7 m·s⁻¹ and 1.8 m·s⁻¹ (at a height of 3 m AGL by 62° vs. 34° and 36°, Table 6). Therefore (due to trigonometric relationships) this fact could have a significant impact on the measurement of lower values (due to the actually longer distance that the drifted spray had to travel and, consequently, increased “extinguishing” of sedimentation drift in the measurement direction). For a wind direction angle of 34° from the perpendicular direction of travel, the actual distance for drifted liquid droplets is 20% greater (33.8 m vs. 28 m to the furthest measurement point). For an angle of 62°, that distance is 113% greater, reaching 59.6 m. For the orchard sprayer, the wind direction was even more deflected (by 67°), which could have even greater impact on the measured sediment drift values (lowering that values). These observations will be taken into account in subsequent experiments, which are planned for the coming years. However, normalizing sediment drift measurement results to account for the atmospheric wind direction during spraying and make appropriate corrections can be difficult due to the extended (in time) spraying process and subsequent spraying of rows further and further from the spray/drift zone boundary, as well as nonlinear drift changes with increasing distance (drift and distance relationships are usually exponential [43]). In this experiment, each subsequent flight/pass was increasingly distant from the spray/drift zone boundary (5 rows of trees were sprayed); therefore, the impact of drift from subsequent rows on the final effect was increasingly smaller. Literature reports that drift from the two outermost rows of orchard trees accounts for up to 70% of the total drift [44] and certainly at least for the majority [45]. In grapevines, drift from the first outermost row can account for up to 91% of the total drift, while that from the second row accounts for only 6% [46].

3.4. Contamination of Spraying Equipment

Spraying equipment contamination was measured in September, along with measurements of in-tree deposition and sediment drift. Cumulative deposition on spraying equipment was calculated as the sum of the absolute deposition on the individual measured components (nanograms per unit). The resulting total value was expressed in milligrams. The drone area subject to sample placement was 0.82 m², while the sprayer area subject to measurement was 9.63 m² (11.7 times larger,

Table 13. The deposition of the BF7G tracer [mg] on an ABZ Innovation L10 spraying drone and on a Munckhof orchard sprayer and cooperating Landini REX 90S tractor after spraying the area of 1000 m² with 120 g of the BF7G tracer, for drone spraying at three flight speeds and the Munckhof orchard sprayer at one speed. Skierniewice, 18 September 2024.

Parameter	Combination
	Drone-1.8	Drone-2.7	Drone-3.6	Orch.Spr.-1.7
Equipment contamination [mg]	46.48 + at standstill	11.98	6.85	1001.33
Equipment area with samples [m2]	0.82			9.63
Area without samples [% of Area with samples]	+30% (without propellers) +78% (with propellers)			+8.3%
Tractor contamination [mg]	N/A			2.77
Tractor area with samples [m2]	N/A			6.33
Tractor area without samples [% of with …]	N/A			+100%

). Because there were also surfaces where no samples were placed (see Section 2.3.3) including these additional areas (without samples) would require increasing the total drone area—for the calculation of total deposition—by 48% (excluding the propellers area) or by 78% (including the propeller area), and the orchard sprayer area by 8.3%. Furthermore, the tractor areas on which samples were placed totaled 6.33 m² and samples were not placed on the 6.3 m² area (mainly the engine).

Total drone contamination was increased 3.9–6.8-fold by turning on the nozzles at a standstill (46.48 vs. 11.98 or 6.85 mg, Table 13). For the orchard sprayer, the total contamination after spraying was 83.6–146.2 times greater (1001.33 mg vs. 11.98 or 6.85), while on the tractor it was only 2.77 mg (at least 2.5 times less than with the drone). At the tracer dose sprayed onto the experimental plot (120 g per approx. 1000 m²), deposition on the drone (contamination) ranged from 0.0387% to 0.0057%, and on the sprayer it was 0.834% of the applied dose. The total contamination, after adding the “no-sample areas” should be increased.

Since most of the liquid contaminating the drone was deposited on the drone’s legs/base (56 ÷ 75%—depending on flight speed/combination) and propeller carriers (4 ÷ 33%), and relatively little on the nozzle carriers (3 ÷ 19%), this may indicate (in combination with lower overall contamination vs. sprayer) a lower risk to drone operators during spraying and preparation. For the orchard sprayer, the majority of the spray liquid contamination was measured on the fan housing located at the very end of the sprayer (69%) and the rear part of the fan attachment (30%). Therefore, special attention should be paid to the operation of nozzles mounted in the air outlet of the fan attachment. Therefore, the drone seems to be safer for its operator than orchard sprayer.

The adopted methodology involved comparing deposition on surfaces represented by placing filter paper samples on them. To approximate the deposition value for the entire drone surface, a formula is required to calculate deposition also on surfaces “not covered” with filter paper samples. It can be assumed that the deposition on these surfaces, taken together, will have an average value similar to the average deposition on the surfaces on which the samples were placed. It is also possible to individually assign each surface to the average deposition from adjacent elements. However, deposition on propellers, whose surface area is relatively large, requires a customized approach due to the high torque of these elements (e.g., washing off sediment after a spray flight).

Measuring the contamination of the drone (and sprayer), although samples were not placed on some surfaces (see Section 2.3.3), allowed for obtaining information on the differences in contamination levels of such equipment (drone vs. sprayer). If more accurate results are needed, or to verify the results obtained in this experiment, a washing method can be used that involves washing the surfaces of the spraying equipment while simultaneously collecting the washings and then performing a quantitative analysis for the tracer content (or PPP content), as such measurements were performed for equipment mounted on manned aircraft [25]. In the case of a drone, it is necessary to determine whether washing the entire equipment with water will not cause damage, e.g., to the control electronics. The professional users of drones say, that it is no problem and drone may be washed with water totally (oral information).

The obtained measurement results allow us to formulate information regarding drone contamination during spraying and, additionally, during standstill periods when the nozzles are activated—intentionally or unintentionally. Due to limited access to the drone (the drone was loaned free of charge but for a limited time), repeating the measurements was impossible under the experimental conditions.

Equipment contamination experiments should be performed with increasing spray rates or varying concentrations of the agent to find the “saturation point” for equipment deposition/contamination. Lan et al. [24] conducted laboratory studies on the adsorption of various PPPs on new or artificially and naturally aged foils (polyethylene and microplastics) and showed that adsorption on the tested surfaces proceeds differently, depending on the condition of the surface and the type of PPP. They noted that adsorption is initially rapid, and after reaching a certain level, the rate stabilizes at a slower rate. This indicates that sprayer contamination studies should be conducted or verified with specific PPPs, and tests with a fluorescent tracer can provide information on the possible proportion of pesticides deposited on spraying equipment during spraying.

4. Conclusions

Overall, for the apple orchard (tree size 3.5 × 1.4 m), spray deposition obtained with the medium-sized drone (ABZ Innovation L10, maximum take-off weight 29 kg) was lower and less uniform than with the conventional orchard sprayer (Munckhof column sprayer). In particular, the uniformity of deposition between the upper and lower leaf surfaces (U/L) and between the upper and lower tree zones (T-T/B) was significantly poorer for the drone. In May, total deposition (U + L) on trees was 1.4–2.0 times higher for the orchard sprayer than for the drone at different flight speeds. In September, similar deposition levels for the sprayer were achieved only at the lowest drone flight speed (1.8 m·s⁻¹). Faster drone flights resulted in significantly lower deposition.

Airborne drift at 5 m from the sprayed row was statistically similar for the slowest drone speed and the sprayer, although numerically about twice as high for the drone. At higher drone flight speeds, airborne drift increased up to 5–6 times more than for the sprayer. In contrast, sedimentation drift decreased with increasing flight speed. Similar cumulative drift levels up to 28 m were recorded for the slowest drone speed and the sprayer. Wind direction variability during spraying likely affected drift measurements, indicating the need for the more accurate and more precise real-time wind monitoring during future trials.

Drone contamination was 7–12 times lower than that of the orchard sprayer when normalized for both equipment types surface area. The drone contamination increased about fourfold when spraying was initiated on the ground.

To improve result comparability, future work should verify the accuracy of drone-logged parameters (flight altitude, speed, spray rate, droplet size) from the flight itself and from flight record data.

Further research should include drones with higher payload capacity influencing stronger air movement, which may improve spray penetration and uniformity within trees. Other factors to be examined include flight altitude, droplet size, spray volume, and canopy geometry and density. Studies should also extend to major Polish fruit crops such as black currants, raspberries, and blueberries. Ultimately, defining spray performance (deposition and drift) under both optimal and limiting legal conditions will support the development of science-based regulatory guidelines for drone spraying in orchards.