Author: Jason Deveau

  • Wind and Flight Speed Shape Drone Spray Coverage: Lessons from 3D Deposition Mapping in Wheat

    Wind and Flight Speed Shape Drone Spray Coverage: Lessons from 3D Deposition Mapping in Wheat

    In this study, 3D sampling of drone spray applications in wheat demonstrates that coverage is strongly influenced by the interaction between drone downwash, flight speed, and wind conditions. These factors collectively determine where droplets land, how evenly they are distributed, and how reliable coverage is from pass to pass. In 2025 we characterized wheat head coverage from a DJI Agras T50. In this study, we explore the larger, faster DJI Agras T100, and relate the observations to what we’ve seen in previous studies.

    Materials and Methods

    Site and crop

    The experiment was conducted at 45939 John Wise Line, St. Thomas, Ontario (42°43’57.0″N, 81°05’49.8″W) on June 3, 2025. Wheat was seeded at 1.8 million seeds/ac on 19 cm spacing and was at the T3 stage (~0.7 m height) at application.

    Design

    Twenty-one poles spaced at 1 m intervals held 3D-printed mounts with 1×3″ water-sensitive papers oriented in four directions relative to the drone flight path: advance, retreat, left, and right. A tramline behind the array preserved canopy structure while allowing access to the samplers (Figure 1).

    Figure 1 – Volunteers retrieving and replacing samplers between passes.

    Drone Operational Settings

    The primary objective of the study was to explore the effect of flight speed on coverage. Speed was increased from 6, to 10, to 14 m/s with the following operational settings:

    • 4 LX07550SX (sprinkler) nozzles
    • 50 L/ha application volume
    • 350 µm droplet size
    • 4 m flight altitude
    • 7 m programmed swath width
    • Tank volume maintained at ~50 L

    The drone began spraying 50 m before and continued 20 m after the samplers, flown on full auto over pole 10 and 11 (the middle of the 21 poles). The spray liquid was municipal water with 0.5% v/v of MasterLock (Winfield United).

    The secondary objective was to compare coverage from the drone spraying 5 gpa (6 m/s) to a 10 gpa (7 m/s) condition.

    Weather

    Weather data was collected using a Kestrel 3550AG weather meter (Kestrel Instruments) in a vane mount positioned roughly 2 m below drone altitude. Data was logged as the drone passed the samplers (Table 1).

    Table 1 – Weather conditions for each spray pass.

    Flights were conducted under a prevailing tailwind (rather than the preferred headwind) due to field constraints. Wind conditions during application varied by treatment. The 6 m/s treatment experienced higher and more variable wind speeds (avg. 6.6 km/h, SD 3.7 km/h, 177°), predominantly from the north (tailwind). The 10 m/s treatment occurred under moderate and stable winds (avg. 4.8 km/h, SD 1.1 km/h, 136°) with a slight right-to-left crosswind component. The 14 m/s treatment experienced low and variable wind speeds (avg. 1.6 km/h, SD 2.0 km/h, 198°) including periods of calm .

    Results

    Deposition Magnitude and Orientation

    Papers were analyzed using a DropScope™ (SprayX, São Carlos, Brazil). Deposition differed strongly by collector orientation (Table 2). Some repetitions were removed if wind pushed spray beyond the collectors. This left a minimum 2 repetitions per condition.

    SpeedDirectionMean (deposits/cm2)Std DevMinMax
    6 m/sAdvance45.7067.031.3210.7
     Left40.0666.190.0211.8
     Retreat20.1020.250.071.0
     Right45.4773.550.0208.9
    10 m/sAdvance52.0763.050.1189.3
     Left31.9543.650.0136.0
     Retreat5.129.190.037.6
     Right37.2270.180.0202.1
    14 m/sAdvance39.0237.510.5115.3
     Left26.8943.910.0130.9
     Retreat0.431.270.05.7
     Right11.0322.530.071.2
    Table 2 – Average deposition by sampler orientation for each speed.

    Forward-facing collectors (advance) consistently recorded the highest deposition across all speeds, followed by lateral orientations. Reverse-facing collectors (retreat) recorded substantially lower deposition. Variability was high for advance and lateral orientations, whereas retreat collectors showed consistently low variability (Table 3).

    DirectionMean (deposits/cm2)Std DevMinMax
    Advance39.0237.510.5115.3
    Left26.8943.910.0130.9
    Retreat0.431.270.05.7
    Right11.0322.530.071.2
    Table 3 – Average deposition by sampler orientation for all passes.

    Directional Bias (Anisotropy)

    Anisotropy refers to the property of having different values when measured in different directions. We can quantify this by dividing the average coverage on one plane by the opposite plane; The resulting indices show the relative direction of deposition.

    For the lateral plane (left-to-right), we divide the average coverage on the left-facing orientation by the right. On the sagittal plane (advance-to-retreat), we divide the average coverage on the advance-facing orientation by the retreat (Table 4).

    SpeedLateral (L÷R)Sagittal (A÷R)
    6 m/s0.88 (slight right-dominant)2.27 (moderate advance-dominant)
    10 m/s0.86 (slight right-dominant)10.17 (strong advance-dominant)
    14 m/s2.44 (strong left-dominant)90.05 (almost entirely advance-dominant)
    Table 4 – Relative coverage indices for lateral and sagittal planes.

    Bias in the lateral index was relatively weak, with a subtle shift with the wind (wind-facing is left) at higher speeds. The sagittal index (advance-to-retreat) increased from a 2x between 6 m/s and 10 m/s to 5x between 10 m/s and 14 m/s, demonstrating strong forward bias with flight and wind direction despite the down-and-back vector created by the downwash.

    Spatial Distribution

    Peak deposition consistently occurred 1 to 5 m downwind of the flight line, rather than directly beneath it. A cross-tail wind shifted deposition laterally, while forward motion (inertia) and wind reinforced deposition in the advance direction. This can be illustrated by isolating the average coverage for each orientation, for all three speeds (Figures 2 to 5).

    Figure 2- Advance Orientation (Facing tailwind). Bars = SE
    Figure 3 – Retreat Orientation (facing away from tailwind). Bars = SE
    Figure 4 – Right Orientation (Facing cross wind). Bars = SE
    Figure 5 – Left Orientation (facing away from cross wind). Bars = SE

    By combining and plotting average coverage on all orientations in a top-down heatmap, we can clearly see the lateral shift to the left of the flight pass (with the light crosswind), the higher relative coverage on the advance face, and indications of bi-modal coverage that likely corresponds to the position of the rotary atomizers(Figure 6).

    Figure 6 – Coverage heatmap created by smoothing the average deposition data for each speed (σ ≈ 1.1 – 1.2 m) over a 300 x 300 grid to illustrate deposition gradients. The colour scale supports a direct comparison of deposition intensity. The 21 samplers are indicated by black dots spaced at 1 m intervals, and the drone flight path appears as a black arrow between posts 10 and 11. Average wind speed and direction appears as an inset white arrow (vector).

    Effect of flight speed on swath width

    Swath width was determined by averaging all deposition on each post for each speed and using our online swath width calculator. The range of flight speeds used in this study did not significantly affect swath width.

    • 6 m/s: 8 m swath width (16.3 % C.V.).
    • 10 m/s: 7.5 m swath width (22.5 % C.V.)
    • 14 m/s: 7.5 m swath width (22.3 % C.V.)

    These widths are 15-20% wider than the widths calculated in the same manner during the 2025 study with the T50.

    Averaging swath widths can mask variability

    This method of calculating and comparing average swath widths is convenient, but it hides any variability in the amount of spray deposited within the swath. Consider that an 8 m swath with 10 deposits/cm2 every meter would have the same C.V. as an 8 m swath with 100 deposits/cm2. Deposit variability can be illustrated by plotting the average coverage along the swath with standard error (figure 7). We see that flight speed significantly influenced the degree of deposition, where higher speeds reduced the average droplet density (counts) as well as the variability (standard deviation).

    Figure 7 – Average coverage, all orientations, for each speed. Bars = SE

    Think of each repetition as a randomly-selected cross section of the swath from somewhere along a spray pass. Calculating swath width from averaged coverage data can hide shifts in the relative position along the flight path, making the composite value greater than that of any single replicate. This variability and the potential for inadvertent smoothing can be exposed by plotting each repetition. (Figure 8).

    Figure 7 – Average coverage (all orientations) from each pole plotted by speed and repetition.

    Therefore, the order of operations matters. When swath width is calculated for each repetition, and then averaged, we would expect the widths to be somewhat smaller. They are presented here in table form next to the previous values for comparison (Table 5).

    Speed (m/s)(A) Deposition averaged, then swath calculated (m)(B) Swaths calculated, then average (m)Difference (A-B) (m)
    68 (16.3% C.V.)6 (29.6% C.V.)-2
    107.5 (22.5% C.V.)6 (33.5% C.V.)-1.5
    157.5 (22.3% C.V.)7.5 (30.5% C.V.)0
    Table 5. Average swath widths generated by two methods.

    Statistical Analysis

    No matter the method, we can draw conclusions from the swath widths calculated here.

    • 6 m/s: highest deposition but greatest variability.
    • 10 m/s: best balance of deposition, uniformity, and swath width.
    • 14 m/s: lowest deposition and most directional bias.

    A two-way analysis of variance (ANOVA) was conducted to evaluate the effects of flight speed and collector orientation on spray deposition. Deposition differed significantly between Advance, Left, Right, and Retreat collectors (F = 6.1, p = 0.0005). Flight speed had a statistically significant effect on deposition, where deposition was reduced with speed (F = 3.03, p = 0.05). The effect of orientation did not significantly depend on speed (F = 0.46, p = 0.83), suggesting that the pattern of deposition was consistent across speeds.

    Effect of volume on deposition

    In a secondary investigation, the drone was flown at 7 m/s, applying 10 gpa to compare coverage to the 6 m/s, 5 gpa condition (Figure 8).

    Figure 8 – Average coverage as deposit counts, all orientations, for 5 gpa and 10 gpa. n=2 for each condition . Bars = SE

    Surprisingly, there was no significant increase in total deposition within the swath when volumes were increased. In fact, the 5 gpa condition is ~8% higher when all deposits are summed or when area under the curve is calculated. The relative shape of the curve was notably different with 5 gpa producing a sharper, higher-intensity central peak, while 10 gpa produced a broader and more uniform deposition profile.

    It was expected that higher volumes would result in higher counts. One theory for the absence of this result was that overlapping depositions in the high volume treatment might have underestimated counts when the papers were digitized. Therefore, the percent surface area was also analyzed (Figure 9). Once again, there was no significant difference in the total percent area or a comparison of area under the curves.

    Figure 9 – Average coverage as area covered, all orientations, for 5 gpa and 10 gpa. n=2 for each condition . Bars = SE

    When swath widths were calculated for each repetition, then averaged for each speed, we arrived at (5 m + 7 m) ÷ 2 = 6 m for the 5 gpa condition, and (5.5 m + 6.5 m) ÷ 2 = 6 m for the 10 gpa condition. We have no explanation for why there was no volume-related difference.

    Discussion

    Wind direction strongly influenced deposition, overriding the down-and-back pattern seen in previous studies. A tail-cross wind likely drove deposition (likely occurring after the drone passed the sampling location), explaining why retreat-facing collectors captured minimal deposition, and peak deposition was accordingly displaced from the flight line.

    Overall, results confirm that wind conditions fundamentally reshape spray distribution. The implication is that wind direction must be accounted for alongside swath width when developing flight path spacing to minimize the potential for overlaps and gaps between passes.

    Further, previous studies have demonstrated a direct and positive relationship between flight speed and swath width up to 8-10 m/s with no further response after ~8 m/s. This study supports the hypothesis that rotary-wing drone speed and swath width share an asymptotic relationship that inflects at ~8-10 m/s (variability makes it difficult to determine an exact value). Flight speed also has a direct and inverse impact on the degree of spray deposition and deposit variability within the swath.

    Finally, caution is advised when interpreting average swath widths. There may be no indication of the degree of coverage within the swath (affecting efficacy), or the lateral variability along the flight path (affecting fieldwide uniformity).

    Related video

    Thanks to Adam Pfeffer and Bayer Canada for in kind and financial support, and thanks to volunteers Erin Jewson (OMAFA Engineer), Halle Barton and Nikki Intranuovo (Bayer Summer Students) for their help with the field work.

    Author’s Note: These results were adjusted in July to exclude outliers and include the results of the spray volume comparison.

  • Airblast Nozzles – Distributing Flow

    Airblast Nozzles – Distributing Flow

    There’s a certain deer-in-headlights expression that creeps onto a sprayer operator’s face when we discuss nozzle selection. We sympathize with our field sprayer clients given the variety of brands, styles, flow rates and spray qualities they must choose from. And PWM has made the process even more complex. However, airblast operators face an additional challenge; Unlike horizontal booms, vertical booms often distribute the flow unevenly to reflect relative differences in the distance-to-target and the density of the corresponding portion of target canopy. We discuss the broader, iterative process of nozzling an airblast boom here, but in this article we focus on the topic of flow distribution.

    An overwhelmed operator trying to nozzle a boom.

    The question of “which rate goes where” is still debated. It’s led to diagnostic devices called Vertical Patternators which show the profile of the spray. Operators can use these to visualize their distribution… but they are few and far between. For the rest of us, deciding on the best distribution begins with understanding how the practice evolved.

    The AAMS vertical patternator. The mast moves back and forth across the swath of a parked sprayer. Each black collector intercepts the spray at different heights. The fractions collect in the tubes at the bottom to show relative volume.
    A blurry shot of an OMAFA-built vertical patternator. The sprayer parks in front of the screens, which intercept spray. It’s collected in troughs and runs into columns that show relative volume.

    1950s

    In the 1950s, the mantra was to blow as much as you could, as hard as you could, and hope something stuck. At the time, John Bean promoted a method called “The 70% Rule” whereby operators used full-cone, high volume disc-core nozzles to emit the vast majority of the spray from the top boom positions. John Bean provided a slide-rule calculator to help operators configure booms to align the top nozzles with the deepest, densest portion of the 20-25 foot standard trees they were trying to protect. Back then, most airblast sprayers were engine-driven low-profile radial monsters capable of blowing to the tops of those trees. The practice persisted into the 60s and was encouraged by Cornell University (Brann, J.L. Jr. 1965. Factors affecting the thoroughness of spray application. N.Y. State. Arg. Exp. Sta. J. paper no. 1429).

    The profile of the spray would have looked something like the following graph:

    1970s

    In the 70s, extension specialists began advising operators to tailor the distribution to match the orchard spacing, tree architecture, canopy density and weather conditions. we reached deep into our archives for the Ontario Ministry of Agriculture and Food’s 1976 publication entitled “Orchard Sprayers” to see what we used to tell airblast operators.

    The 1976 update of Ontario’s 1971 “Orchard Sprayers” guide. A trove of hard-earned knowledge that still has relevance today. I love that the Minister’s name, and not the author’s, is on the cover. Let’s set the record straight: R. W. Fisher, D. R. Menzies and A. Hikichi, based in Vineland and Simcoe, Ontario.

    Here’s a synopsis of what was advised:

    1. Choose a tree size and shape that is typical of your orchard and park the sprayer at the normal spraying distance from it.
    2. Find one or two middle nozzle position(s) and air deflector or vane settings that direct the spray up through the top-inside of the tree. This is called the “middle volume zone”.
    3. Find rates that will give a large output in this middle volume zone, and smaller outputs for positions above and below.
    4. The total output must still add up to the target volume.

    It seemed operators were getting away from high rates in the top positions and instead shifting the distribution to match the canopy shape and density. If we were to follow these recommendations, the spray profile would look something like this:

    Later, Agriculture Canada’s 1977 publication entitled “Air-Blast Orchard Sprayers – A Operation and Maintenance Manual” had similar advice. Here we find the “2/3 boom rule” as the authors state: “To ensure good distribution through the trees, about two-thirds of the spray should be emitted from the upper half of the manifold.”

    1980s

    Operators followed this approach well into the 80s, as they endeavored to aim the majority of the spray into the densest part of the canopy. Many can relate to the following illustration that divides the boom, which I modified from an array of period factsheets. The fractions represent the portion of the available boom. The percentages indicate the relative volume. Of course, it matters how large and how far away the target is for either the 2/3-boom or 70% rule to make sense (the middle volume zone is shown receiving 65-70% in the silhouette).

    1990s-2000s

    The 2/3 or 70% rules still work for standard nut and citrus trees, and perhaps for large cherry trees, but pome and tender fruit orchard architecture is densifying at this point in time. In the 90s and 00s we started transitioning from semi-dwarf into trellised, high density orchards.

    Leaping to 2005, Ohio’s Dr. Heping Zhu et al., found that a high density orchard is effectively sprayed by the same rate in each nozzle position. They wrote: “[Historical] recommendations are to use a larger nozzle at the top of each side, with the capacity of the top nozzle at least three times greater than other individual nozzles. However, results in this study with three different spray techniques showed that spray deposit was uniform across the tree canopy from top to bottom with the equal capacity nozzles on the air blast sprayer.”

    What a pleasant surprise to simplify our lives! If we can use an even distribution for dense, nearby trees, it follows that any vertical crop with the same width and density located close to the sprayer (e.g. cane fruit, trellised vines, etc.) would benefit from even distribution:

    Today (2020s)

    So, how do we do it today? There is still no simple answer; Conditions change, not all sprayers are the same, and not all applications have the same target. Let’s build on what history has taught us and establish a process to achieve better coverage uniformity and reduce waste.

    No matter the crop, the operator must first adjust air settings. Air volume and direction play the most critical role in transporting a droplet to (and into) a target canopy. Too high an air speed will cause spray to blow through the target, rather than allowing it to deposit within. Aim the air just over, and just under, the average canopy. Ensure there’s enough air to overcome ambient wind and to push the spray just past the middle of the target canopy.

    It should be noted that we assume the operator is spraying every row. With certain exceptions, alternate row middle spraying is not generally recommended. Not only can it compromise coverage on the far side of the target, it makes it far harder to match the nozzling on a single-row sprayer and is a sure-fire way to increase drift.

    Next, determine which nozzles are not needed (e.g. spraying the ground or excessively higher than the top of the canopy). Remember: hollow cones overlap very close to the boom and spread as much as 80°. Airblast sprayers rarely if ever need the lowest positions and unless spraying overhead trellises they may not need the highest either. Turning off the highest, and most drift-prone, nozzle positions in high density orchards is illustrated in the logo I was asked to create for Washington’s short-lived Pound the Plume awareness campaign in 2017.

    Then, finally, we decide on distribution. If the crop is nearby and relatively narrow, you can try even distribution. If you elect to distribute the spray unevenly to better match the variable-width target, or compensate for distance, aim half the overall output at the densest part of the canopy (the middle volume zone). Consider how the following factors might influence your choices:

    1. High humidity means more spray will reach the target, and vice versa. This is because all droplets are prone to evaporation. We have heard it said in hot, dry conditions (described by Delta T), a droplet can lose ½ its diameter every 10 feet. As they evaporate they get lighter, meaning they are less subject to their original vector and the pull of gravity, and more subject to deflection by wind. The use or coarser droplets, and/or humectants, can help, but higher volumes can help too – they increase the odds of some droplets hitting the target and actually humidify the air to slow evaporation.
    2. Windspeed increases with elevation, so spray is most likely to deflect at the top of canopies where they have already lost size (and momentum and direction). Early in the season when there is little if any foliage, wind speeds are higher overall. This is why we advise adjusting air settings using a ribbon test before considering boom distribution – you need enough air volume, aimed correctly, to get the spray to the top.
    3. The denser and deeper a canopy, the more spray is filtered and unavailable for coverage. This is why you will always achieve more coverage on the adjacent, outer portion of a canopy versus the interior. In semi dwarf apple orchards we have seen the coverage drop by half for every meter of canopy. Finer spray can penetrate more deeply because there are more droplets and they move erratically, whereas coarser droplets move in straight lines and impact on the first thing they encounter. Higher volumes will improve penetration and overall coverage, but there is a diminishing return and runoff will occur more quickly leading to more waste.
    4. Further to the last point, remember that it’s the air that propels the spray, not the pressure. Higher liquid pressure can propel coarser droplets further, but has little effect on finer droplets. imagine throwing a golf ball and a ping pong ball into a light headwind and envision how they fly. Plus, the higher the pressure, the finer the mean droplet diameter.

    Confirm Your Work

    To know how all these factors play out, you must use water sensitive paper (or some other form of coverage indicator) to diagnose the results. Remember, the goal is uniform coverage and for most foliar products, we want to achieve a minimum coverage threshold of 10-15% and a droplet density of 85 deposits per cm2 on at least 80% of the targets.

    Taking the time to match your output to the target has the potential to greatly improve coverage and reduce waste. Nozzle body flips and quick-change nozzle caps make the process of switching nozzles between blocks fast and easy. It’s worth it.

    Grateful thanks to Mark Ledebuhr, Gail Amos and Heping Zhu who edited, corrected and contributed to this article.

  • How Higher Speeds Affect Drone Swath Width

    How Higher Speeds Affect Drone Swath Width

    Speed Study

    Swath width is a fundamental parameter in spray drone mission planning. It facilitates the uniform application of broadacre pesticides at the target rate. Pilots adjust the swath width via operational settings such as droplet size, flight speed and altitude to produce the most effective and efficient application.

    Rapid advances in drone design, however, may warrant a re-evaluation of how operational settings affect swath width. For example, the most recent generation of drones are now capable of speeds up to 20 m/s (72 km/h), which is twice that of the previous generation.

    In late 2025 we conducted a series of comparative herbicide applications using the DJI Agras T50 and T100. For both drones, swath width increased with speed up to ~10 m/s, as expected. However, between ~10 m/s and 18.5 m/s, swath width from the T100 did not seem to increase further. Similar observations have been reported by researchers at AgroEfetiva (São Paulo, Brazil; personal communication).

    These results suggest that the relationship between speed and swath width is positive and direct at lower speeds, but reaches a saturation point beyond which any further increase in speed no longer affects swath width. This is an asymptotic relationship. To test this hypothesis, we conducted a deposition study where swath width was measured at flight speeds that increased incrementally from 8 m/s to 20 m/s.

    Configuration Study

    The standard T100 configuration uses two rotary atomizers (“sprinkler” nozzles; LX07550SX) with a reported maximum combined flow rate of 30 L/min. The alternate orchard configuration incorporates a boom that supports two additional “mister” nozzles (LX09550SX), increasing the reported maximum flow rate to 40 L/min.

    To improve productivity in broadacre applications, some operators have adopted a hybrid configuration. In this setup, the orchard boom is retained, but the reputedly drift-prone mister nozzles are replaced with a second set of sprinklers. This approach is intended to achieve a higher flow rate than the standard two nozzle configuration while maintaining a larger mean droplet size.

    A secondary objective of this study was to compare the Hybrid configuration with the Orchard configuration (Figure 1).

    Figure 1 – Left: DJI Sprinkler Nozzle (LX07550SX). Four such nozzles comprised the “Hybrid” configuration. Right: DJI Mister nozzle (LX09550SX). Four such nozzles comprised the “Orchard” configuration.

    Materials and Methods

    Location and Layout

    The study was conducted at Ontario’s Simcoe Research Station on May 12, 2026. The site (42.857414, -80.271759) was a flat, recently tilled sand/loam field with no vegetation present. A DJI Agris T100 drone was used to perform the spray applications, supported by the D-RTK 3 relay station and flown on full auto.

    The spray mix was 0.2% v/v Super Signal Blue (Precision Laboratories) and 0.125% v/v Activate Plus NIS (Winfield United) in municipal water, pre-mixed to ensure consistency. A volume of 40 L – 60 L was maintained throughout the trial to minimize the effect of a changing payload.

    The sampler was a flat, horizontal, continuous bond paper strip measuring 7.5 cm wide and 30 m long (secured in Speed Tracks™, Application Insight LLC). The sampler was oriented perpendicular to the prevailing wind, with the intention of flying the drone with a headwind across the 15 m mark (the centre) (Figure 2). Test passes determined that the T100 required 210 m to reach 20 m/s while half-full.

    Figure 2 – T100 spraying indicator dye across the 30 m continuous sampler.

    Before the swathing runs began, the prevailing wind shifted direction slightly.  It was decided to fly the drone 5 m upwind (at the 20 m mark along the 30 m sampler) to ensure any downwind displacement was captured on the sampler (Figure 3).

    Figure 3 – Trial layout and prevailing wind conditions.

    Drone Settings and Swathing Order

    The primary objective of the study was to explore the effect of flight speed on swath width. Speed was increased from 8 m/s to the maximum 20 m/s by 2 m/s increments. Trial and error with the controller indicated that we could achieve these speeds by balancing an application volume of 30 L/ha and a programmed swath width of 7 m.

    Altitude was set to 4 m which is lower than the 5 m minimum recommended by DJI for high-speed flight. This was a compromise above the preferred 3.5 m altitude we have historically used with the T50. It was felt that higher altitudes would create unacceptable potential for swath displacement.

    Rotary atomizer design is not standardized, and as a result, the droplet size selected on the controller did not necessarily produce the desired results. The Hybrid configuration was programmed to emit 350 µm droplets, selected as a compromise between drift mitigation and coverage potential. To offset the Mister nozzles’ reputation for producing a finer spray, the Orchard configuration was set to the maximum 500 µm. Operations settings are noted in Table 1.

    ConfigurationNozzleDroplet size (µm)Speed (m/s)Altitude (m)Programmed Swath Width (m)Application Volume (L/ha)
    Hybrid4 Sprinklers3508, 10, 12, 14, 16, 18, 204730
    Orchard2 Misters, 2 Sprinklers5008, 10, 12, 14, 16, 18, 204730
    Table 1 – Operational settings for trials.

    Three repetitions of seven speeds were flown for each configuration. Anticipating an increase in temperature and wind speed throughout the day, it was decided move through all seven speeds (a single repetition) before resetting and doing so two more times. The intent was to preclude confounding weather effects. Ideally, we should have alternated between configurations as well, but this proved impractical. As a result, we flew the Hybrid configuration first and the Orchard configuration last.

    Weather

    Weather data was collected using a Kestrel 3550AG weather meter (Kestrel Instruments) in a vane mount positioned 2.5 m above ground. Temperature and relative humidity were comparable throughout the ~3 hours of data collection, but as anticipated, wind speed was higher for the later Orchard configuration passes.

    As previously indicated, wind direction shifted from an ideal headwind situation just before trials began, and was somewhat changeable, but the average wind direction for the two configurations was comparable (Figure 4 and Table 2).

    Figure 4 – Weather conditions recorded at roughly 10-minute intervals, corresponding to the drone passing over the sampler.
    TimeConfiguration FlownAverage Temperature (°C)Average Relative Humidity (%)Average Wind Speed (km/h ± SD)Average Direction (° ± SD)
    11:55 am – 1:16 pmHybrid11.759.98.4 ± 3.5323 ± 46
    1:43 pm – 2:52 pmOrchard12.951.611.9 ± 2.5316 ± 48
    Table 2 – Average weather conditions for the Hybrid configuration passes and for the Orchard configuration passes.

    Collector analysis

    Bond paper digitization

    Bond papers were scanned using a Swath Gobbler™ (Application Insight LLC). The software measured deposition as both percent area covered (% area) and deposit density (deposits/cm2) every 100 mm, with a thresholded Hue of 23-280, a Saturation of 5-120 and a Value of 156-255.

    Effective swath width calculation

    The large data set produced by each pass was reduced in size by averaging the deposition for every 50 cm. This data was entered into our Excel-based swath width calculator, which assumes a racetrack pattern and sums deposits from adjacent swaths. The resulting swath width for each pass was the maximum width that minimized over- and under-dosing as well as the coefficient of variation (CV).

    Analysis

    The average swath width derived from deposit density data was wider than that derived from percent area covered (Table 3).

    Table 3 – Group means and standard deviation for average swath widths derived from deposit density data and percent coverage data. The average CV was between 29 and 32%.

    A two-way ANOVA (Analysis of Variance; α = 0.05) was performed to determine any significant effect of speed or configuration on swath width (Table 4). Flight speed had no significant effect on swath width, no matter how it was derived (% area covered or deposit density), for either configuration (Hybrid or Orchard). However, the average swath width derived from deposit density was significantly wider for the Orchard configuration compared to the Hybrid and presented higher variability.

    Table 4 – Results of two-way ANOVA, exploring interactions between speed, swath width and configuration (95% confidence interval).

    Orchard configuration was prone to displacement in a side wind. Shifting the flight path 5 m upwind improved the downwind capture, but for some flights it did trim a small portion of the upwind deposition. Deposit density gives greater resolution and exposes more variability than percent area covered. Figure 5 shows the average deposition by speed based on deposit density. Figure 6 shows the average deposition by speed based on percent area covered.

    Figure 5 – Average deposition by speed based on deposit density. Arrow indicates flight path.
    Figure 6 – Average deposition by speed based on percent area covered. Arrow indicates flight path.

    Figure 7 shows the average deposition by configuration based on deposit density. Figure 8 shows the average deposition by configuration, based on percent area covered. Based on deposit density, there were 55% more deposits on the downwind side of the sampler for Orchard configuration set to set to 500 microns compared to the Modified configuration set to 350 microns.

    Figure 6 – Average deposition by configuration based on deposit density. Arrow indicates flight path.
    Figure 8 – Average deposition by configuration based on percent area covered. Arrow indicates flight path.

    The average swath widths calculated from deposit density (Figure 9) and percent area covered (Figure 10) are shown with standard deviation. While it appears the swath width is less around 14 m/s, it is statistically insignificant and the response to speed is essentially flat. As with prior studies, swath widths calculated from deposit density are larger than those calculated from percent area covered.

    Figure 9 – Average swath widths for each speed, derived from deposit density data. SD shown. n=3 for each speed, while n=2 for Hybrid configuration at 12 m/s and 14 m/s.
    Figure 10 – Average swath widths for each speed, derived from percent coverage data. SD shown. n=3 for each speed, while n=2 for Hybrid configuration at 12 m/s and 14 m/s.

    Observations

    Previous studies demonstrated a direct and positive relationship between drone speed and swath width up to 8-10 m/s. Here, we see no further response after ~8 m/s. This supports the hypothesis that rotary-wing drone speed and swath width share an asymptotic relationship that inflects at ~8-10 m/s. Variability makes it difficult to determine an exact value.

    Despite increasing the programmed droplet size to the maximum 500 microns for the Orchard condition, there was 55% more downwind deposition compared to the Hybrid condition, which was set to 350 microns. This supports the claim that the Mister nozzle produces a span of droplet sizes that include far more fines than the Sprinkler nozzle, and underpins the need for a better understanding of the spray quality produced by rotary atomizers.

    Spraying at high speeds is not an advisable practice. While swath width is no longer affected after ~8 m/s, there are other considerations. Note that it required 200 m for the drone to reach the highest speed, and in a related study we have seen swath width taper during initial acceleration and final deceleration, leaving gaps in coverage.

    Further, the minimum 5 m altitude advised by DJI ensures a safe margin for the drone to respond to obstacles and topography during high speed flight, but is not conducive to spraying. The author is aware of a situation where flying the T100 at 4 m altitude and 18 m/s over a canola field with rolling hills caused it to perform an emergency landing.

    An ideal speed is one that maintains the most consistent swath width at a reasonable altitude.

    Thanks to Drone Spray Canada and Bayer Canada for in kind and financial support, and thanks to Cesar Cappa, OMAFA horticulture weed specialist for his participation in the study.

  • A Drone’s Swath Width Tapers During Acceleration and Deceleration

    A Drone’s Swath Width Tapers During Acceleration and Deceleration

    It can be challenging to illustrate how a drone deposits spray. We can take a cross-section of spray deposit (that is, the swath) using a continuous sampler, or a series of discrete samplers. This typically reveals a sharply-peaked curve that is symmetrical in a head- or tail-wind, or skewed if there is a cross-wind.

    Spray coverage from a DJI T100 (4 m altitude, 30 L/ha, 16 m/s, 350 µm droplets) measured as percent area covered over a 30 m continuous sampler (Speed Track and Swath Gobbler – Application Insight LLC). Note: the prevailing wind was 8 km/h and generally from the right, skewing deposition to the left.

    But this analysis only captures a slice of the swath at a single moment in time. Swath width and relative position along the flight path is inconsistent and highly variable. Some have shown this using herbicides and aerial photographs analyzed through a Normalized Difference Vegetation Index (NDVI). NDVI reveals the health and density of vegetation based on how they reflect different wavelengths.

    A 50 m continuous swath visualization. This stitched graphic from work by K. Falk shows annotations for upwind/downwind edges and width measurements.

    Design

    We wanted to try something similar, but using dye instead of herbicide. Rhodamine WT is a synthetic fluorescent dye that absorbs green light and emits red light. It’s highly water-soluble and is the industry standard for environmental monitoring, including mapping groundwater flow, tracing pollution plumes and studying flow rates in rivers. We’ve used it in the past on Ontario’s snowy fields to show how boom height affects coverage uniformity with boom sprayers. Why not from a drone?

    Spray coverage from a boom at an optimal height (left), too low (middle) and too high (right). From episode seven of our Exploding Sprayer Myths series.

    The idea was to lay down swaths from a DJI T100 at different altitudes and speeds, and then take aerial photographs of the stained snow using a DJI Mavic 3 Multispectral drone. It would be visually impactful, and perhaps we could even quantify the deposition using photogrammetry and drone mapping software.

    We flew a snowy field on Stefina Line, Blenheim, Ontario. It was 2°C and wind was gusting from 8-14 km/h. We mixed a 0.5% v/v solution of 0.5 L rhodamine WT in 900 L and flew a trial run (4 m altitude, 50 L/ha, 10 m/s, 500 µm droplets). But, as Robbie Burns said, “The best laid plans of mice and men [often go awry]. (If you’re not into late 18th century poetry, you’re forgiven.)

    Rhodamine WT applied by drone did not result in high-contrast deposition. Hoping it was a function of concentration, we emptied the remaining rhodamine into the nurse tank, essentially doubling the concentration (we didn’t measure it), but to no avail.

    Rhodamine WT applied by drone did not result in high-contrast deposition.

    New Plan

    We decided to pivot. We laid out a ~2.0 ha (5 ac) area and programmed the drone to fly a 10 m route spacing. Our hope was that the swaths would overlap for complete coverage, and perhaps we’d see a completely-stained rectangle. But, it seemed striated rather than solid, and the faint colouration made it difficult to see. This led us to shift to another location and spray another block, but this time we flew it four times in the hope that “multiple coats” would result in a brighter stain. We were already there, so why not?

    Feeling a bit underwhelmed, we slowly flew the Mavic M3M to capture aerial footage. In the following image you see the less-than-spectacular result in the visual spectrum (left). Then our operator had the bright idea to filter the spectrum to enhance the rhodamine, and suddenly we had something (right). By reducing the green light, the rhodamine glowed gold, and we could see the faint block (sprayed once) and the much more intense block (sprayed four times).

    (Left half) Image from DJI Mavic M3M stitched together in Pix4D. Faint pink colouring can be discerned in the visual spectrum. (Right half) when green is reduced, the rhodamine glows gold on a red background. We begin to see the faint ~2.0 ha (5 ac) block sprayed with a single pass on the right, and the much brighter ~2.0 ha (5 ac) block sprayed 4x on the left.

    Note that the block is not a uniform, gold rectangle. The deposition is concentrated along the flight paths and does not seem to overlap (or even reach) the spray from adjacent passes. While this might be a resolution issue, it implies the swath width was less than the route spacing.

    Then we zoomed-in on the downwind and upwind edges of the block that received multiple passes and noticed something. Given the variability of spray deposition, and the fact that we made four passes, we expected coverage would be blurred. Instead, there were clear, intense tapers where the drone was either accelerating or decelerating at the boundary (depending on the direction it was flying).

    Close-up of the downwind edge of the ~2.0 ha (5 ac) area sprayed four times. Note the intensity and tapered profile of the swaths as they approach the boundary.
    Close-up of the upwind edge of the ~2.0 ha (5 ac) area sprayed four times. Note the intensity and tapered profile of the swaths as they approach the boundary.

    We know from prior work with the T10, T30 and T50 that swath width and travel speed are directly related up to a certain speed (perhaps around 10 m/s, but this is yet to be determined). The practical implication was made clear in these images: there were significant gaps in coverage on two field margins. How long were the gaps? Did relative wind direction matter? Did this change with acceleration or deceleration? We went to the DJI flight record to find out.

    Using the playback feature, we could see the flow rate relative to the drone speed and geographical position (that is, latitude and longitude). Using a single pass, we established the point where the drone started or stopped spraying, and the point where the the target flow rate and speed were constant. Then, using the latitude and longitude, we calculated the intervening distance between coordinates.

    Screen captures from the DJI flight recorder. (Left) The ~2.0 ha (5 ac) sprayed four times. You can see the drone icon displaying its position on the flight path, as well as the flow rate, altitude, speed and longitude (latitude was cropped out of the image). (Right) A close-up of one of the turns at the lower side of the sprayed block. All four passes are traced.

    Analysis

    We should note that we were able to spray just under 5 ac with a single tank, so the area that received four passes required multiple refills. This means the weight of the drone at any given location (and therefore the strength of the downwash) changed throughout the trial. We feel weight would affect acceleration and deceleration distances, but it wasn’t controlled in this study.

    Most runs showed the acceleration distance was significantly greater than deceleration distance (Paired t-test: t = 2.54, p = 0.044).

    PassRelative VelocityHeadwind Distance (m)Tailwind Distance (m)
    1Acceleration–54.4
    1Deceleration–44.1
    2Acceleration57.1–
    2Deceleration43.4–
    3Acceleration–44.3
    3Deceleration–38.9
    4Acceleration54.8–
    4Deceleration64.6–
    5Acceleration–52.2
    5Deceleration–37.7
    6Acceleration56.7–
    6Deceleration46.7–
    7Acceleration–55.9
    7Deceleration–43.5

    Headwind increased both acceleration and deceleration distances. There was a 7.5 m (16%) increase in a headwind, which was significant (r2 = 0.23). There was a stronger effect on deceleration, although not significantly.

    Wind DirectionMean (m)Std Dev (m)Min (m)Max (m)Count
    Headwind53.97.743.464.66
    Tailwind46.47.037.755.98
    Relative VelocityHeadwind Mean Distance (m)Tailwind Mean Distance (m)
    Acceleration56.251.7
    Deceleration51.641.0

    Conclusion

    In a previous trial with the DJI T100 we noted the drone was not able to exceed 18.3 m/s over a 250 m treatment distance. It took roughly 200 m to get up to 18.3 m/s before the drone began to slow in anticipation of the end of the treatment block. At the time we assumed the drone would not be under- or over-applying because the pump flow rate compensated for a changing travel speed.

    However, we had not considered the effect on swath width. Here, we see the T100 requires about 50 m to accelerate to / decelerate from 10 m/s. This distance is a function of wind direction, and to a lesser extent, whether the drone is accelerating or decelerating. During this time, the swath width is less than when at target flight speed, leaving (larger) gaps between passes.

    This is a concern for broadacre applications where consistent, uniform coverage is required. Consider a fungicide that may no longer be effective at the extremes of the swath, or a systemic post-emerge herbicide where crops directly beneath the drone (especially horticultural crops or GMO’s with stressed metabolisms) could experience phytotoxic damage.

    We have no practical work-around to suggest, but operators should be aware of the effect.

    Thanks to Drone Spray Canada for in kind support and access to flight records, and thanks for Cesar Cappa, OMAFA horticulture weed specialist for his participation in the study and assistance with data analysis.

  • Establishing an Optimal Airblast Carrier Volume

    Establishing an Optimal Airblast Carrier Volume

    North American product labels may or may not include carrier volume recommendations. When they do, it could be based on a two-dimensional value like the planted area, or perhaps on row length which is more appropriate for trellised crops that form contiguous hedge-like canopy walls. Volume may also be tied to product concentration, which sets minimum and maximum volumes based on product rates. Or, more commonly, volume recommendations take the form of vague guidelines such as “Spray to drip” or “Use enough volume to achieve good coverage”.

    In all cases, spray efficacy and efficiency can be greatly improved by dialing-in the carrier volume to optimize coverage uniformity and reduce off-target spraying. This is easier said than done because the optimal spray volume is case-specific. It depends on a complicated relationship between:

    • Weather conditions (E.g. temperature, humidity, wind speed and direction)
    • Sprayer design (E.g. air handling, droplet size and flow distribution over the boom)
    • Traffic pattern (E.g. every row or alternate row)
    • Product chemistry (E.g. mode of action and formulation).
    • Target (E.g. Crop morphology, planting architecture)

    It is the final variable, the nature of the target, which is the focus of this article. To learn more about the other variables, grab a copy of Airblast101.

    The plant canopy and planting architecture dictate volume

    Quite often, the target in airblast applications is the plant canopy. The plant canopy is the collective structure containing all plant surfaces. This could be the foliar portion of a single pecan tree, a panel of grapes, or a bay of container crops. The planting architecture describes how those canopies are arranged on the planted area. If we consider the canopy and architecture geometrically, we can make relative statements about the volume required when all other variables are equal.

    Six geometric characteristics of the plant canopy and planting architecture.
    Geometric CharacteristicRelationship to Carrier Volume (per unit planted area)
    Row SpacingThe greater the row spacing, the less volume needed.
    Plant SpacingThe greater the plant spacing, the less volume needed. This assumes gaps between the canopies (I.e. not a contiguous hedgerow).
    *Canopy DepthThe greater the canopy depth, the more volume needed.
    *Canopy WidthThe greater the canopy width, the more volume needed.
    *Canopy HeightThe higher the canopy, the more volume needed.
    Canopy DensityThe denser the canopy, the more volume needed.
    *The product of average canopy depth, width and height is the canopy volume. This value forms the basis for many dose expression models and historic carrier volume calculators such as Tree Row Volume.

    Canopy density

    Let’s focus on a single plant canopy. Research has demonstrated that with the possible exception of canopy height, canopy density has the greatest influence on optimal sprayer settings.

    Density describes the amount of matter inside a canopy relative to the volume of space it occupies. The denser the canopy, the more surface area there is to cover and the more difficult it is for spray to penetrate.

    While air handling plays a significant role in improving coverage, a denser canopy will almost always require a greater carrier volume.

    When two physiologically diverse blocks share an alley, use the sprayer settings suitable for the larger of the two. It’s more important to ensure good coverage on the big block than to save on the smaller.
    When two morphologically-diverse blocks share an alley, a two-sided, every-row sprayer should employ settings suitable for the larger of the two. It’s more important to ensure good coverage on the big block than to save on the smaller. Once the hybrid row is sprayed, settings should be modified to suit the block.

    For most perennial crops, canopy density changes over the growing season. The influence of age and staging on canopy size and density will depend on the crop variety, plant health and canopy management practices. The practical implication is that as the canopy grows and fills it typically warrants an increase in spray volume.

    As illustrated in the figure below, the volume used should reflect the current stage of canopy development. If a volume suitable for the densest and largest stage of development is used all season, it will create a great deal of waste early in the season. However, if volume is increased incrementally to reflect canopy growth, a better fit between coverage and volume will minimize waste.

    Note that volume is increased around petal fall, but the fit could be improved with more increments. Caution is advised to ensure the volume is raised (if required) prior to immediate need, particularly during key developmental stages like bud break or bloom where fungicide coverage is critical.

    The curved line represents the leaf area in a canopy (Y-axis, right) increasing over the growing season (X-axis). The volume of the spray (Y-axis, left) providing effective coverage is indicated in green. Spraying the same volume throughout the season means a lot of over-spray (red) early in the season. The target simply isn’t there yet. Using less volume early season and changing about midway through the season, or as required by canopy development, has the potential to save a lot of spray (blue) without compromising spray coverage. Note that the first volume should give sufficient coverage to reach mid-season, and the second volume should be sufficient to reach the end of the spraying season. Always err on the side of excessive coverage to buffer against the impact of unanticipated variables.

    There are exceptions to this rule. Many nursery crops and mature evergreens often do not require changes to volume. High density apple orchards may or may not require an increase in volume. Early in the season, sparse canopies have low profiles that result in very low catch efficiencies. In other words, a great deal of spray misses the target.

    The amount of waste is a function of the application equipment design and the weather conditions. Most low-profile axial airblast systems envelop the target in spray with limited means of reducing air energy sufficiently, or to turn off the spray between trees.

    Further, sparse canopies do not restrict wind, which means ambient wind speed tends to be higher early in the season compared to when the trees become wind breaks. This creates a drift-prone situation and higher volumes are often used to compensate for the loss. The collective result is that excess spray volume is inevitable early season.

    As the canopies fill, the wind is reduced and catch efficiency increases, so trees intercept more spray without having to raise volumes. This balance eventually tips, however, and an increase in volume may be advisable.

    Watch the following video to see the impact of using excessive spray volume (and poor air adjustment settings) in a young cherry orchard. The waste becomes particularly apparent at ~43 seconds when the sprayer passes in front of the woods and the plume can be seen with higher contrast.

    While some loss is inevitable in such a sparse canopy wall, this situation could be improved by using less carrier volume, larger droplets, the correct air settings, canopy-sensing optics and/or a tower or wrap-around sprayer design.

    Adjusting spray volume sprayer settings to reflect the canopy can save money and reduce environmental impact during early-season applications and in young plantings. Mix the tank as you normally would to maintain the pesticide concentration on the label, but adjust the sprayer output to match the plant size.

    Performed correctly, you will be able to go further on a tank without compromising efficacy. This crop-adapted spraying method and the relationship between spray volume, concentration and dose are described further in this article and this article.

    Estimating volume from canopy geometry

    It is challenging to decide on an appropriate spray volume. Many operators resort to historical or regional practices and do not make adjustments to reflect their specific situation. Others refer to models such as Tree Row Volume (a.k.a. Canopy Row Volume) which relates canopy volume per planted area to spray volume.

    In this case, catch efficacy is expressed as a coverage factor, which is determined through experimentation specific to the crop, environment and sprayer.

    Tree Row Volume = (Avg. Canopy Height × Avg. Canopy Spread × Planted Area) ÷ Row Spacing

    Spray Volume = Tree Row Volume × Coverage Factor

    In New Zealand, coverage factors for dilute applications to deciduous canopies range from 0.07 to 0.1 L/m3 (0.00052 to 0.00075 US gal/ft3) or 71 to 100 ml/m3 (0.067 to 0.096 US oz/ft3). The range captures variation in canopy density and any product-specific coverage requirements. Oil sprays, for example, require more surface coverage than most products. While closer to “the truth”, the Tree Row Volume method is still only an estimate.

    If the operator has no prior experience with the crop or the sprayer and wants a sanity-check on their estimated spray volume, we propose the following guidelines for full canopy dilute application to mature crops using every-row traffic patterns. The volumes may seem high, but recognize we have selected a very challenging scenario.

    • Small canopies (E.g. bush, vine, cane, high-density fruiting wall): 500 L/ha (55 US gal./ac.) to 1,000 L/ha (110 US gal./ac.).
    • Medium canopies (E.g. tender fruit, pome): 750 L/ha (80 US gal./ac.) to 1,250 L/ha (135 US gal./ac.).
    • Large canopies (E.g. tree nut, citrus): >2,000 L/ha (214 gal./ac.) and up tp 7,000 L/ha (748 US gal./ac.).
    • For sprayer operators that think in 100 m row lengths, consider 20 L volume per 100 m row length per 1 m canopy height.

    Further Resources

    No matter the approach to determining spray volume, it is imperative that coverage is assessed. It is amazing what we ask of airblast sprayers. Read this short article for some perspective on the coverage we hope to achieve from a given spray volume.

    We propose the use of water-sensitive paper to assess spray coverage. We describe its use and evaluation in detail in this article, this article and in this article.

    Dialing-in an optimal spray volume is an iterative process that requires careful observation and keeping records on what works and what doesn’t for your specific operation.

    Jon Clements (University of Massachusetts) has noted special considerations when it comes to establishing effective volumes for plant growth regulators that go beyond this article. You can explore the concepts in this 2021 factsheet (Spray Mixing Instructions – Considering Tree Row Volume) by Terence Robinson and Poliana Francescatto (Cornell University) and Win Cowgill (Professor Emeritus, Rutgers University).

    Finally, if you really want to get lost the weeds, check out this video recorded in 2021. I had an opportunity to learn from pros like Dr. Terence Bradshaw (University of Vermont) and participants from the Great Lakes region. They’ll tell you all you ever wanted to know about Tree Row Volume. Settle in!

    Thanks to Mark Ledebuhr of Application Insight LLC for his contributions to this article.