Tag: RPAS

  • RPAS Swathing in Broad Acre Crop Canopies

    RPAS Swathing in Broad Acre Crop Canopies

    This work was performed with contributions from Adrian Rivard (Drone Spray Canada) and Adam Pfeffer (Bayer Crop Science – funding partner). Dr. Tom Wolf is gratefully acknowledged for his editorial support and assistance interpreting the results.

    Introduction

    This research is part of a continuing effort to identify best practices for broad acre crop protection using remote piloted aerial systems (RPAS). Previous work in wheat, corn and soybean has provided insight into how RPAS operational settings and environmental factors affect drift potential, effective swath width and spray coverage. This information, paired with advancements in RPAS design, has helped operators to improve spray deposit accuracy.

    However, RPAS still produce what has traditionally been considered poor (or at least sporadic) broad acre coverage. Many studies have illustrated these shortcomings using herbicides or fluorescent tracers. Contributing factors include inappropriate operational settings, low application volumes (20-50 L/ha) paired with coarser spray qualities, and inaccurate swath widths. In light of these issues, we struggle to reconcile claims of acceptable disease control, which is arguably the greatest challenge in a spray-based crop protection paradigm.

    Tar Spot

    One real-world example of intermittent disease control from aerial applications (not just RPAS) is the case of tar spot in corn. Tar spot is a fungal disease caused by Phyllachora maydis and it is becoming a significant economic concern in Ontario. Left unchecked the disease causes rapid, premature leaf senescence. This reduces photosynthetic capacity, and ultimately, yield. Depending on spray timing, crop variety, environmental stressors, and the product applied, protection should last for up to three weeks.

    In the last few years there have been several reports (both in Ontario and in corn-producing US states) of tar spot “striping” following aerial sprays. Crops seem well protected directly beneath the flight path (green and healthy), but efficacy tapers to failure towards the edges of the swath (brown and desiccated). Fundamentally, this is likely due to inadequate spray coverage caused by an overestimation of the effective swath width.

    Figure 1 Tar spot striping in Western Illinois following two applications from a fixed wing sprayer (2023).
    Figure 2 Tar spot striping from RPAS volume trials. A brown strip can be seen between two passes in each RPAS treatment of 30 and 50 L/ha. The top is an application by a 100 foot horizontal boom. Each treatment is separated by an unsprayed check. (2023).
    Figure 3 – Tar spot striping in Ontario corn following fungicide application by helicopter (2024).

    Effective Swath Width (ESW)

    The measured swath width presents the lowest variability (as indicated by the coefficient of variability, CV) while minimizing the degree of over- and under-dosing. As a matter of operational productivity, wider swaths mean wider route spacing, which is attractive because it means fewer passes and faster applications. Once the agronomics are considered, the effective swath width is that portion of the swath that gives the desired biological result. It may equal, or only be a fraction of, the measured swath width. It is plausible that inappropriate effective swath widths from aerial applications are common, but have not always been detected, because:

    • Generally, fungicides are weakly systemic and give modest yield increases from disease suppression and their “stay green” properties. Until tar spot, a sub lethal dose of fungicide did not lead to rapid and acute crop failure.
    • Most growers do not intentionally leave unsprayed checks, or the check locations do not coincide with disease presence.
    • The applied product rate is sufficiently high to cover regions of under-application.
    • Taken together, deficiencies are often too subtle for passive detection.

    This is not to suggest that pilots intentionally inflate swath widths. Swaths are evaluated during fly-in calibration sessions using established protocols (e.g., Operation S.A.F.E.), and RPAS swath evaluation has emulated these practices. Calibrations take place on bare ground or stubble/grass using two-dimensional samplers (i.e., continuous samplers like string or bond paper, or discreet samplers like water sensitive paper). However, this protocol does not account for any physical interference from the crop canopy itself. This may have negative implications, particularly given the unique nature of the RPAS swath.

    RPAS tend to produce swaths with a very narrow span and a steep profile. To a certain extent, their swath widths share a direct relationship with altitude and headwind speed, and coarser sprays result in narrower swaths (with Dr. Michael Reinke, MSU). The outer edges of the RPAS swath represent the least amount of spray volume along the width, and this coincides with the turbulent dispersion zone of the downwash. Therefore, those extremes should contain a higher proportion of low-energy droplets moving in multiple directions relative the centre of the swath.

    While crop morphology and planting architecture are contributing factors (i.e. part of the agronomic use case), it is generally accepted that the degree of spray penetration falls off exponentially with canopy depth. It follows that this should also be the case for any lateral movement, resulting in a significantly shorter swath in-canopy versus on bare ground.

    Materials and Methods

    Spray Sampling

    Spray deposition was sampled using a 15.8 m (52 ft) Speed Track (Application Insight LLC) loaded with 3-inch bond paper (Staples Canada). The spray mix was 0.3% v/v FD&C Blue #1 Liquid. Bond papers were analyzed using a Swath Gobbler (2nd gen software – Application Insight LLC) at 100 mm sampling rate (i.e., ~150 discreet images per sample). Hue: 32-180. Saturation 17-60. Value: 156-255.

    The Swath Gobbler produces a complete, correlated and ordered record of the cross-section of a swath. For each discreet image, it reports the number of individual droplet stains on the sampler per area. It also reports percent area covered by measuring the total number of pixels with dye divided the total number of pixels in the image.

    The device deliberately does not calculate a Droplet Size Distribution (DSD) of the stains. This is because any DSD calculated from paper collectors relies on assumptions that cannot be validated, such as the fact that all droplets are captured and detected, spread factors are known for that application condition and similar for all stain sizes, there are no multiple hits, etc.

    RPAS

    The sprayer was a DJI T40, calibrated according to the pilot’s standard operating procedure (Drone Spray Canada). Certain operational settings varied with treatment and will be detailed later in this section.

    The flight path was perpendicular to the sampler, aligned with the centre using pin flags as references for the pilot. Spraying began approximately 20 m prior to the sampler to ensure the RPAS was at target speed and continued some 20 m past the sampler.

    Figure 4. DJI T40 approaching sampler on bare ground. Sampler was later moved into the adjacent wheat field (left).

    Defining Coverage

    Swath width will be calculated from two different coverage metrics.

    Percent Area Covered describes the amount of surface area covered by deposit. Given the variable degree of stain diameter (a function of sampler material, spray mix, and droplet velocity) this value can only be used as a relative index (i.e., can only be compared to itself). No conclusions can be drawn about how spray interacts with plant tissue, but generally more coverage correlates to improved crop protection.

    Deposit Density describes the number of individual droplet stains on the sampler per area. Higher densities can imply more uniform distribution over the plant surface, which is particularly important for contact materials.

    Previous studies (with Dr. Tom Wolf, Agrimetrix Research and Training, data not shown) indicate a higher correlation between deposit density and swath width at lower versus higher spray volumes. Lower volumes are typically comprised of finer droplets, which are more accurately resolved using deposit counts. Swath widths determined by deposit density also tend to be longer than those determined using percent coverage, better aligning with real-world observations of efficacy.

    Wheat

    R40 wheat was planted on October 9th, 2023, at 808,000 seeds/ha (2 million seeds/ac). Wheat height at the time of the trial was 60 cm (25 in). The location was 45180 Fruit Ridge Line, St. Thomas, Ontario. Deposition trials took place on May 23rd. Wheat stubble swath testing also took place at this location on May 15th.

    The RPAS was programmed to apply 50 L/ha using a 260 µm droplet diameter according to the DJI software. Air speed was 5 m/s and the flow rate was 11-12 L/min as it passed over the sampler. Swath was programmed at 8 m.

    Coverage was evaluated for water (control) and for a spray mix containing 0.15% v/v Interlock (a drift mitigating adjuvant – Winfield United) and 0.15% v/v Interlock + 0.125% v/v Activate Plus (a spreader adjuvant – Winfield United). For bare ground, each treatment had three passes (n=3) except for water, which had four (n=4).

    The wheat canopy was only sprayed with water three times (n=3). Limited passes were made because it served as a proof of principle. Any indication of relevant differences in the swath width would justify later trials in corn and soybean. These first passes revealed issues with the experimental design that were later corrected:

    • The RPAS spray tank level was not held constant. The RPAS weight affects the intensity of the downwash. The volume dropped from 30 L to ~20 L over the course of the experiment. In future trials, a tank volume of 20 L was maintained from a premixed source.
    • The wind direction occasionally shifted from a direct headwind to a partial cross wind from the RPAS’s right. In future experiments, we waited for an optimal wind direction before starting each pass.
    • The RPAS altitude was set to 3 m above bare ground. We assumed it would climb to account for the height of the wheat, but the canopy did not register with the RPAS sensors. As a result, spray was released ~60 cm closer to the wheat heads than to the ground in bare ground swathing. In future experiments, we confirmed that the RPAS was 3 m from the top of the crop canopy.
    • Despite best efforts, moving the sampler into the wheat parted and distorted the canopy. As a result, the sampler was not as obscured as it should have been. We developed strategies to minimize canopy distortion in corn and soybean that will be described later.
    Figure 5. Top-down view of sampler in wheat canopy. Note that the canopy did not close over the sampler as intended.

    Corn

    Corn was planted on May 15th, 2024, at 13,300 seeds/ha (33,000 seeds/ac). The sampler was erected in the field on July 3 to allow the canopy to grow up and around it. Deposition trials took place on July 26 and every effort was made to leave the canopy undisturbed around the sampler. Corn measured 2.4 m (9 ft) at the tassel and 1.2 m (4 ft) at the silks. The sampler height corresponded to the ears. The location was 42°40’52.1″N 81°04’45.9″W near 5277 Quaker Road, Sparta, Ontario.

    Figure 6 Sampler erected to 4 ft. Crop grew around the sampler to minimize any canopy disturbance.
    Figure 7 Sampler position relative to ears during sampling.
    Figure 8 Installing Speed Track for swath testing in wheat stubble.

    Soybean

    Soybean was planted on June 30th, 2024, at 80,800 seeds/ha (200,000 seeds/ac) on 38 cm (15 in) centres. Deposition trials took place the morning of August 14. While the densest area was selected for the trials, the field was patchy with crop height spanning 20-25 cm (8-14 in). Each section of the Speed Track was inserted under the canopy separately to avoid disturbing or damaging the plants. The track was elevated ~10 cm off the ground. The location was at 42°46’50.4″N 81°08’20.8″W near 43900 Talbot Line, Central Elgin, Ontario.

    Figure 9 Sampler in soybean.

    Corn and Soybean Treatments

    The following treatments were repeated three times in-canopy (n=3) (Table 1). The actual flow rate (recorded as the RPAS passed over the sampler) was always ~1.5 L/min less than programmed.

    Treatment #Droplet Diameter (µm)Programmed Swath (m)Volume (L/ha)Rate (L/min)Flight Speed (m/s)Spray Mix
    1320102010.510water
    232083010.58.3water
    332085010.55water
    43208305.75water
    550085010.55water
    63208505.750.5% Masterlock
    732083010.58.30.5% Masterlock
    Table 1 RPAS operational settings for corn and soybean treatments

    The following treatments were repeated three times on wheat stubble (n=3) (Table 2). Once again, the actual flow rate (recorded as the RPAS passed over the sampler) was always ~1.5 L/min less than programmed.

    Treatment #Droplet Diameter (µm)Programmed Swath (m)Volume (L/ha)Rate (L/min)Flight Speed (m/s)Spray Mix
    1320102010.510water
    232083010.58.3water
    332085010.55water
    43208305.75water
    Table 2 RPAS operational settings for wheat stubble treatments

    Weather Data

    The RPAS flight path was into the prevailing wind, but minor variations occurred throughout sampling. Weather was recorded as the RPAS passed over the sampler using a Kestrel 3550AG weather meter in a vane mount positioned on a tripod 2 m above ground (Table 3).

    TerrainWind Speed (km/h)Direction Relative to Flight PathTemperature (°C)Cloud Cover (%)RH (%)
    Bare Ground3-5Headwind +/- 25° from starboard20-21060
    Wheat Canopy5-7Headwind +/- 25° from starboard21-22060
    Corn Canopy2-4Headwind +/- 15° from starboard23-26<1075
    Wheat Stubble4-7Headwind +/- 15° from starboard26-28<1065
    Soybean3-4Headwind +/- 15° from starboard22055
    Table 3 Average weather conditions during trials.

    Results

    Raw Coverage Expressed as Percent Coverage or Deposit Density

    Coverage can be presented as raw data plotted by swath position. This is a qualitative means for assessing the swath. The bare ground data has been presented (using both coverage metrics) as an example (Figures 10 and 11).

    Figure 10 Swath coverage data for water on bare ground expressed as percent area covered. All four passes are plotted.
    Figure 11 Swath coverage data for water on bare ground expressed as deposit density. All four passes are plotted.

    Repetitions were similar enough to imply that environmental conditions were consistent during sampling. By averaging the repetitions, coverage in-canopy can be more easily compared to that on bare ground Figures 12 and 13).

    Figure 12 Average swath coverage data expressed as percent area covered. Bare ground (n=10). Wheat canopy (n=3).
    Figure 13 Average swath coverage data expressed as deposit density. Bare ground (n=10). Wheat canopy (n=3).

    The magnitude of coverage on bare ground exceeded that in-canopy, tapering to similitude and near-zero at the edges of the pattern. It can therefore be concluded that the entire swath was captured, and that spray was filtered by the canopy before reaching the sampler within.

    The difference between bare ground and the wheat canopy was greater when the data were presented as percent area versus deposit density. Differences in the number of deposits from finer sprays were more accurately resolved using deposit density than percent coverage. Since it can be expected that smaller droplets penetrate a canopy better than coarser droplets, it may be more appropriate to use deposit density to document their presence. We also saw indications of wider swaths when data were presented as deposit density, as well as a bimodal distribution that reflected the positions of the two rotary atomizers.

    While informative, this raw coverage format did not allow empirical comparisons. Each pass must be converted to a swath width.

    Converting to Swath Width

    Each pass was transformed by averaging Swath Gobbler data to a single value every 0.5 m. Data were then entered into the www.sprayers101.com swath width calculator and the SW was manually determined for each pass. Criteria was the lowest overdose, lowest underdose and lowest CV for an idealized threshold coverage of 90% that of the highest value in the swath. In the following histogram, the SW from all treatments have been averaged for ground and canopy terrains (Figure 14).

    There was a significant reduction in swath width in a wheat canopy compared to stubble or bare ground. There was a 41.2% reduction in swath width in a canopy when measured as percent area covered and a 26.6% reduction when expressed as deposit density. As previously stated, deposit density better reflects the contribution of finer deposits, which tend to penetrate deepest into crop canopies.

    Figure 14 Average effective swath width for all treatments on all terrains. Swaths expressed from both percent coverage and deposit density metrics. Standard error bars presented. Canopy (n=45). Ground (n=22).

    When the data is considered by terrain and by crop, we see that swathing on bare ground or in wheat stubble doesn’t have a significant impact. This justifies combining those data as “Ground” in subsequent analyses.

    Another observation that supports the use of deposit densities is the difference between the intended (i.e., programmed) swath width and the detected swath width on ground (Figure 15). The SW on ground was closer to the intended 8 or 10 m swath width when expressed as deposit density. It was approximately half the desired width when expressed as percent coverage, which is considerably less than common practice.

    Figure 15 Average effective swath width for each crop and terrain. Swaths expressed from both percent coverage and deposit density metrics. Standard error bars presented. Ground 8 m swath (n=19). Ground 10 m swath (n=3). Canopy 8 m swath (n=39). Canopy 10 m swath (n=6).

    Canopy Effect

    By percent area, corn had the biggest reduction in swath width compared to bare ground, then soybean, then wheat (Table 4 and Figure 16). This suggests the SW shares an inverse relationship with the canopy depth. However, the relationship reversed when SW was expressed as deposit density. The relationship between droplet size, crop physiology, planting architecture and canopy penetration is complicated, and no conclusions can be drawn beyond a reduction in SW in-canopy.

    Crop% Reduction in SW (% area)% Reduction in SW (deposits/cm2)
    Corn44.020.6
    Soybean32.228.3
    Wheat21.731.5
    Table 4 Reduction in average effective swath width in-canopy by crop compared to on ground. Swaths expressed from both percent coverage and deposit density metrics.
    Figure 16 Average effective swath width for each terrain. Swaths expressed from both percent coverage and deposit density metrics. Standard error bars presented. Bare ground (n=10). Wheat Stubble (n=12). Corn Canopy (n=21). Soybean Canopy (n=21). Wheat Canopy (n=3).

    Effect of Volume on SW

    The effect of spray volume on swath width is not immediately clear. When the data were expressed as deposit density, volume shared an inverse relationship with SW in canopy (Figure 17). There appeared to be no effect when expressed as percent coverage. The inverse relationship is weakly expressed, if at all, for both metrics on bare ground.

    Figure 17 Average effective swath width by volume and terrain. Swaths expressed from both percent coverage and deposit density metrics. Standard error bars presented. Canopy 20 L/ha (n=6). Canopy 30 L/ha (n=18). Canopy 50 L/ha (n=21). Ground 20 L/ha (n=6). Ground 30 L/ha (n=3). Ground 50 L/ha (n=13).

    Effect of Speed on SW

    For most RPAS designs, lower volumes are applied at higher flight speed (Table 5). Previous work demonstrated that higher flight speeds tended to result in wider swaths and an increase in drift. Do higher speeds cause wider swaths in-canopy, despite lower volumes?

    Volume Applied (L/ha)5 m/s Flight Speed8.3 m/s Flight Speed10 m/s Flight Speed
    20–3 treatments9 treatments
    309 treatments12 treatments
    5034 treatments––
    Table 5 – Number of treatments for each flight speed by volume applied.

    Flight speed had a clearer impact on swath width than spray volume did (Figure 18). There was a positive relationship between flight speed and swath width as measured by deposit density in canopy and on bare ground.

    Figure 18 Average effective swath width by speed and terrain. Swaths expressed from both percent coverage and deposit density metrics. Standard error bars presented. Canopy 5 m/s (n=27). Canopy 8.3 m/s (n=12). Canopy 10 m/s (n=6). Ground 5 m/s (n=16). Ground 8.3 m/s (n=3). Ground 10 m/s (n=3).

    Just as with volume, there appeared to be no significant effect on swath width in either canopy when expressed using percent coverage. This was likely because finer sprays were better able to penetrate a canopy and deposit density is better able to resolve their presence.

    Conclusions

    There was no difference in SW between stubble and bare ground. The SW on-ground was far closer to the programmed 8 or 10 m swath width when expressed as deposit density.

    There appears to be a significant reduction of SW in-canopy versus on-ground. A crop canopy created a 26.6% reduction when expressed as deposit density. Specifically, corn was -20.6%, soybean was -28.3%, and wheat was -31.5%. Previous work has demonstrated diminishing coverage with canopy depth in corn, but it is difficult to make comparisons between agronomic use cases (e.g. different planting architectures and plant physiologies).

    When the data were expressed as deposit density, spray volume shared an inverse relationship with SW in-canopy, but the effect on SW on-ground was less clear. However, RPAS speed had a clear inverse relationship with SW in-canopy and strong trend on-ground.

    It is understood that finer spray is better able to penetrate canopies. One reason is because finer droplets are able to become entrained the downwash. Another is simply mathematical advantage, given that finer sprays are comprised of exponentially higher numbers of droplets than coarser sprays, increasing the odds of deposition. Conversely, coarser droplets (which have the greatest influence on percent area covered), are more likely to impinge on the canopy structure before reaching the sampler. Deposit density appears to be the more accurate metric for calculating SW both on-ground and in-canopy.

    The reduced SW in-canopy versus on-ground explains, in part, why striping is occurring in aerial corn fungicide applications. The route spacing reflects on-ground swath width, where it should reflect the shorter, ESW.

  • Exploring Spray Drones in Soybean

    Exploring Spray Drones in Soybean

    White mould is caused by the fungus Sclerotinia sclerotiorum and it’s an annual threat to soybean when cool, wet conditions correspond with flowering. Variety selection (e.g. high tolerance) and cultural control (e.g. crop rotation and wider row width) are important management tools, but ultimately the application of a crop protection product between R1 and R2 is required for high-risk fields. (Learn more here).

    This article describes the results of an experiment exploring soybean canopy coverage and fungicide efficacy from a rotary spray drone. All work was performed under PMRA research authorization. There are currently no labels to apply crop protection products in Canada.

    Experimental design

    For the spray coverage trials, two locations were selected in southern Ontario (one south of Sparta and one west of Talbotville). This was a full field-scale trial with a single application made at R1.5 on July 18 (Sparta area) and July 22 (Talbotville area), 2023. There were two replications in each field and treatments were laid out parallel with the planting direction in a randomized design. Four other locations in Ontario and Quebec were also used in the larger efficacy/yield study. All locations had some level of white mould infection.

    1. Untreated Check
    2. *DJI Agras T30 – 20 L/ha (6.8 m/s, 2.5 m above canopy, TJ TT110015)
    3. DJI Agras T30 – 30 L/ha (5.7 m/s, 2.5 m above canopy, TJ TT110015)
    4. DJI Agras T30 – 50 L/ha (3.3 m/s, 2.5 m above canopy, TJ TT110015)
    5. New Holland 345 – 150 L/ha (TeeJet XR11006 nozzles on 50 cm spacing)
      *Not included in spray coverage trial

    We established an effective swath width of approximately 4 m (13.1 ft). The drone made three passes to cover the 12 m (40’)-wide treatment area, corresponding to the widths of the 9 m (30’) or 12 m (40’) headers later used to harvest in each field. Buffers were left on either side the treatment area. Fungicide was applied at label rate plus 0.125% Activate.

    Target placement and retrieval

    Soybeans were planted on 38 cm (15”) row spacing. The coverage sampling area was positioned in the middle of the treatment area. A length of rebar was positioned in-row and sheathed in PVC tubing. Two SpotOn brand water sensitive papers (WSP) from the same production run were secured face-up approximately 1/3 and 2/3 deep in the canopy. A block of six such samplers were positioned in a 3 x 2 grid (every third row and approximately 2 m apart in row). This block was then repeated 10 meters (33’) further into the block for a total 24 water sensitive papers per replicated treatment (see below).

    The papers were retrieved and temporarily placed on clipboards to dry before they were placed in paper bags for short term storage. They were digitized using a SprayX DropScope within 48 hours of retrieval on the “ground sprayer” setting, measured as percent surface covered (% area), and deposit density (# deposits/cm2).

    Weather during coverage trials

    Weather data was monitored using a Kestrel 3550AG weather meter (Kestrel Instruments) in a vane mount positioned 1.5 m (5 ft) above the ground. Wind speed fluctuated during the treatments, but wind direction remained relatively stable at 90 degrees to the flight path. The Sparta location averaged 6.4 km/h (4 mph) while the Talbotville location was considerably higher at 14.4 km/h (9 mph). Nevertheless, targets remained within the swath, despite any offset, as indicated by visual confirmation as well as the consistent coverage observed on the windward WSP compared to other, downwind samplers in each pass. Cloud cover was high at both locations.

    Results

    Coverage

    The coverage recorded from each WSP was averaged by canopy position (bottom 1/3 or top 1/3 of canopy) and presented in the following histograms with standard error. There were some spoiled collectors, primarily in the lowest canopy position, ruined by high humidity and physical contact with the plant. However, the lowest n for any treatment was 31 collectors and the highest was the full 48. Coverage is presented both as % area covered and as deposit density in counts per cm2.

    Efficacy and yield

    Three phytotoxicity ratings were performed 7, 14 and 21 days after treatment. White mould was rated at harvest and final crop yield reported in bu/ac.

    Observations and Considerations

    As expected, both water volume and canopy depth share direct relationships with percent-area covered (i.e. lower water and lower canopy depths mean lower coverage). Water volume also shares a direct relationship with deposit density for a given droplet size, but canopy depth is more complicated as smaller droplets tend to penetrate more deeply into canopies and low water volumes tend to produce smaller droplets. However, as a general observation, less water translates to less coverage no matter the metric for coverage, and this has been shown to reduce product efficacy.

    How, then, can we reconcile the claims of efficacy from low-volume drone applications? It’s typical that the % area covered from a 50 L/ha drone application is ¼ or less than that of “conventional” field drop systems which in North America tend to employ 150-200 L/ha. In speaking with Mark Ledebuhr (Application Insight LLC) about how low volumes could possibly be efficacious, he explained that in sugarcane production in Guatemala, the condensing humidity is likely the reason why their 1 gallon/acre applications are working. The droplet survivability, and the re-hydration and secondary movement of the deposits were a good thing.

    In the case of contact fungicides in North America, it may be humidity as well, but also the deposit density, combined with higher concentrations of active ingredient, that explain the similar efficacy and yields as seen here between the 50 L/ha (drone) treatment and the 150 L/ha (field sprayer) treatment. This would concentrate both the active ingredient (possibly increasing uptake rate, or residue persistence, depending on the product mode of action and the target’s physiology) as well as the adjuvant load (possibly improving sticking/spreading of deposits).

    Another consideration surrounds how deposit spread is analyzed. Water sensitive paper underestimates the spreading effect that can occur on plant surfaces (especially where surfactants are used). This is why WSP tends to be used as a relative index, meaning that papers should only be compared to other papers. Perhaps deposits are spreading more on the plant surfaces in the low-volume drone application (again, given the higher concentration of formulated adjuvants) than the water sensitive paper is indicating, and that is improving efficacy.

    This concept of how low-volume applications might affect coverage and subsequent efficacy, and the potentially positive impact of re-formulating products to include higher adjuvant loads, is well-described in this precis by Dr. Andrew Chapple and Malcolm Faers. Currently, accepting that the amount of control provided by the drone application falls short of that provided by a field sprayer, this study indicates that drones have the potential to produce acceptable results in fungicide applications if conditions are suitable, timing is optimal and water volumes are sufficiently high.

    This study was a collaborative effort with Bayer Canada and Drone Spray Canada.

  • Exploding Sprayer Myths (ep.14): Maverick’s Mojo

    Exploding Sprayer Myths (ep.14): Maverick’s Mojo

    Have you lost that loving feeling?

    Well here’s your chance to get it back while learning how to operate spray drones safely, effectively, and legally.

    Join Maverick and Goose as Viper barks out best practices when using this new application technology.

    Remember: You’re not just a pilot – you’re an applicator!

    Thanks to RealAgriculture, the Simcoe Research Station, Drone Spray Canada and special guest, Dr. Michael Reinke.

  • Insecticidal Management of Rangeland Grasshoppers using RPAS

    Insecticidal Management of Rangeland Grasshoppers using RPAS

    This article summarizes a 2022 paper that can be downloaded here.

    Grasshoppers are an integral part of rangeland ecosystems, but population outbreaks can cause significant damage to adjacent crops and forage. For example, cattle consume about 1.5-2.5% of their body weight in forage per day, so pound for pound, a grasshopper will eat 12-20 times as much plant material as a steer. This represents serious economic damage to the cattle industry, especially during times of drought when forage is already scarce.

    Insecticides are the primary means for suppressing grasshopper populations, and they are typically applied on rangelands using conventional fixed-wing aircraft, per the United States Department of Agriculture (USDA) Animal and Plant Health Inspection Service (APHIS) Rangeland Grasshopper and Mormon Cricket Suppression Program. This is particularly effective when population hotspots can be targeted before those populations grow to critical numbers, but depending on the location, this short turn-around time can be challenging.

    Recently, the use of remote piloted aerial systems (RPAS or drones) for spraying in small farm operations and for site-specific management of crop pests in terrains not easily accessible to fixed-wing aircraft has received increased attention around the globe. Drones have the potential to occupy this niche because they can fly (or even hover) closer to plant canopies with more precision and safety than conventional aerial systems.

    Given these strengths, and given that swath uniformity is less of a critical issue when pests (such as grasshoppers) are particularly mobile, we investigated the efficacy of treating grasshopper population hotspots using a drone.

    Plots

    A randomized plot design with two treatments (untreated and control) and eight plots was established on rangeland near Estancia, New Mexico, with each rectangular plot measuring approximately 4 ha (10 ac).

    The RPAS was a six-rotor Precision Vision 35 (Leading Edge Aerial Technologies, New Smyrna Beach, FL, USA) equipped with four Turbo TeeJet XR110-01 nozzles (two on each side of the aircraft) mounted to spray booms.

    Swathing

    Twenty-six water sensitive papers were placed in a line perpendicular to the flight path, approximately 0.91 m (3 ft) apart, to capture the spray deposition. The RPAAS was flown at a speed of 8.94 m/s (20 mph).

    Layout of water sensitive cards and flight line during spray deposition measurements.

    The papers were scanned with an Epson Perfection 1240U laser bed scanner and swath analysis was performed using DropletScan (WRK of Arkansas, Lonoke, AR, USA; WRK of Oklahoma, Stillwater, OK, USA; and Devore Systems, Inc. Manhattan, KS, USA). Papers were scanned at 600 dpi and the software was used to determine the coefficient of variation (CV) for both a simulated racetrack and a back-and-forth spray pattern.

    The CV for multiple effective swaths was calculated and the swath width with the lowest CV was chosen for the study. This effective swath width was 12.2 m (40 ft) with an average application rate of 2.75 L/ha (37.6 fl. oz./acre). Following a 50% RAAT IPM strategy, a 24.4 m (80 ft) swath was used for treatments, with an average treated swath application rate of 1.4 L/ha (18.8 fl. oz./acre).

    Application rate of tank mixture S across swath. Dashed blue lines indicate effective swath width.

    Treatment and Efficacy

    Two types of grasshopper population density estimation methods (visual estimation and sweep netting) were performed on the day before treatment (pre-count) and 3, 7, 10, and 14 days after treatment.

    The RPAS was flown at 9.8 m/s (22 mph) at an altitude of 3.05 m (10 ft). Treatments with the liquid insecticide Sevin XLR PLUS required a single flight to cover each replicated plot, without completely depleting the battery, and each of the four treatments was completed within approximately five minutes of the total flight time.

    Our results demonstrated that Sevin XLR PLUS significantly suppressed grasshopper populations over a 14-day period (normalized population reduction was 79.1 ± 8.4% SEM) and quite rapidly (mostly by day 3) compared to untreated controls. These results are comparable to those achieved with fixed-wing aircraft.

    Effects of Sevin XLR PLUS treatment on grasshopper density and mean ± SEM across the trial period.

    Observations

    Because RPAS are relatively portable, the potential exists to shorten the average length of time between the identification of a hotspot and a treatment. The RPAS covered the whole test area in a single flight in approximately 5 minutes, making these population hotspot treatment applications relatively rapid, potentially more cost-effective, and more targeted in comparison to fixed-wing aircraft.

    We would describe the observed efficacy as successful in the sense that grasshopper populations, when accounting for reduction in untreated control plots, were reduced by 79.1 ± 8.35% SEM. This is very close to what is typically expected for APHIS program treatments, which is 80 to 95% population reduction. Our lower-end results can probably be attributed to the arid conditions, lower levels of rangeland forage observed during the study in that region of New Mexico, and a mobile, rapidly aging population.

    Before adoption as an application method option, further research is recommended on using an RPAS to cover larger areas in combination with using diflubenzuron-based insecticides, which are often preferred.

    This research was funded by an interagency agreement with USDA-APHIS-Plant Protection
    and Quarantine (PPQ): 20-8130-0893. It may not necessarily express APHIS’s views.

  • Rotary-Wing Drone Spray Coverage and Drift in Field Corn

    Rotary-Wing Drone Spray Coverage and Drift in Field Corn

    This work was performed with Mark Ledebuhr (Application Insight LLC.), Adrian Rivard (Drone Spray Canada) and Adam Pfeffer (Bayer Crop Science – funding partner). Amy Shi is gratefully acknowledged for her assistance with statistical analysis.

    Introduction

    In June 2017, Transport Canada cleared the general use of drones. In 2018, Health Canada clarified that the use of Remote Piloted Aircraft Systems (RPAS) for pesticide application is not permitted under the Pest Control Products Act without sufficient data to characterize any associated risk. Currently, there are no liquid pest control products registered for application by drone in Canada.

    Stakeholders want to use drones to apply pest control products in Canada. To that end, several research trials have been approved by Health Canada. However, multi-rotor drones represent a unique application technology more akin to air-assisted ground sprayers than manned aircraft. As such, conventional models for drift, exposure and efficacy may not apply. Fundamental questions surrounding the utility of drones must be addressed before efficacy and residue can be considered in any relevant context.

    Research and user experience has identified, and is beginning to understand the relative influence of, external factors such as crop morphology, planting architecture, topography, and environmental conditions. Considered with the product mode of action, these factors inform operational settings such as altitude, travel speed, nozzle choice, and application volume to optimize applications. This collective “Use Case” depends on drone design, which is highly variable and rapidly evolving.

    Having performed preliminary work characterizing effective swath width, and recognizing its popularity in North America, we used DJI’s Agras T10 in this study. Our objective was to evaluate fungicide efficacy on Northern Corn Leaf Blight, Tar Spot, Grey Spot and Common Rust in field corn, as applied using the T10. Drift and coverage would be characterized to provide context for the efficacy analysis, but also to develop data to inform best practices and possibly regulatory decisions surrounding risk. Aspects of the study would be repeated using conventional ground sprayer technologies to form a basis for comparison.

    Objectives

    1. Quantify spray coverage in field corn at three canopy depths, on adaxial and abaxial surfaces, as recovered tracer dye (indexed to % of applied rate ac-1), area covered (%) and deposit density (deposits cm-2).
    2. Quantify drift as recovered tracer dye (indexed to % applied rate ac-1) collected using the horizontal flux method up to eight meters high on the immediate downwind edge of the application.
    3. Evaluate the fungicide efficacy, applied using the T10, at 2 and 5 gpa as compared to a conventional overhead broadcast treatment at 16.7 gpa.

    Material and Methods

    Design

    Trials were conducted between July and August of 2022 in three Ontario corn fields. The locations, the application methods and data collected are detailed in Table 1.

    FieldLocationCorn VarietyApplication MethodRate (gpa)Data Collected
    1Jaffa (42°45’56.6″N 81°02’06.5″W)DKC45-65RIBAgras T102 and 5Drift, Coverage, Efficacy
    Overhead Broadcast16.7Coverage, Efficacy
    2Fingal (42°42’17.9″N 81°15’15.3″W)DKC49-09RIBAgras T102 and 5Drift, Coverage, Efficacy
    Overhead Broadcast16.7Efficacy
    3Port Rowan
    (42°35’53.6″N 80°30’43.2″W)    		P0720AMDirected (Drop hoses)20Coverage
    Table 1 – Trial sites by application method and data collected

    Treatments were arranged in a randomized complete block design (Figure 1). Corn was planted on 30″ centres, with about 6” in-row spacing between stalks. We targeted spray for the R1 stage of development (approx. 8’ high). Fields 1 and 2 each hosted two replicated treatments of 2 gpa, 5 gpa, and 16.7 gpa, as well as two unsprayed checks. In field 1, blocks were 60’ (24 rows) wide by 1,150’ long for the T10, and 120’ (48 rows wide) by 1,150’ long for the broadcast field sprayers. A single, 120’ swath was applied using the field sprayers, and four 10’ (4 row) swaths were required to spray the centre 40’ (16 rows) of corn using the T10. This was based on a 10’ effective swath width determined in previous research. Field 2 had a similar layout but was 1,820’ long.

    Figure 1- Sample experimental layout for Field 2. In this example, horizontal flux collectors are positioned 3’ downwind to intercept any off-target drift from the edge of the adjacent 2 gpa treated area.

    Coverage Analysis

    To account for variability, each treatment block was subdivided into two regions, each containing an array of nine spray collectors. Each spray collector (Figure 2) consisted of a vertical, 8’ pole in-row between corn plants. Samplers were attached at three depths to span the silking region: Top: 1.5’-2’ below the tassel. Bottom: 1.5’-2’ from the ground. Middle: halfway between them. Samplers were parallel with the ground to ensure the highest degree of spray interception. On one side, two 1”x3” water sensitive papers (WSP; Innoquest Inc.) were clipped back-to-back with a sensitive side positioned up (adaxial) and facing down (abaxial). The other clip held two 4” square sheets of Mylar in the same orientation. Sampler type was alternated vertically (e.g. Mylar – WSP – Mylar or WSP – Mylar – WSP).

    Figure 2- Spray collectors temporarily loaded with WSP and Mylar samplers. These were held above the tassels as they were carried to the collection sites in each block. Three clips were positioned per pole, alternating Mylar and WSP samplers on each side, on two arrays of nine poles, as previously described.

    This study used 864 WSP and 864 Mylar samplers for the RPAS treatments, and 162 WSP for the overhead broadcast and directed applications. Following the application, samplers were retrieved as soon as they were dry enough to handle (about 30 minutes) and individually placed into pre-labeled sealable plastic bags, each uniquely coded to the exact position and orientation of the collector.

    Operational Use Cases

    • 5 gpa: DJI Agras T10 was operated at 3.3 m/s, 2 m above tassels. TeeJet 11002 AIXR nozzles equipped with 50 mesh filters were operated at 70 psi.
    • 2 gpa: DJI Agras T10 was operated at 7.0 m/s, 2 m above tassels. TeeJet 11002 AIXR nozzles equipped with 50 mesh filters were operated at 45 psi.
    • 16.7 gpa: Overhead broadcast condition. Field 1 ran a John Deere 4038R operated at approx. 10 mph with TeeJet XR11006 nozzles on 20” spacing. Pulse width modulation (ExactApply) was engaged. Field 2 ran a New Holland 345 front-mounted boom sprayer with TeeJet XR11006 nozzles on 20” spacing.
    • 20 gpa: Directed condition. John Deere R4038 operated at approx. 4.5 mph with Beluga drop hoses suspended on 30” centres to correspond with alley spacing. Two nozzle bodies were positioned 15″ apart equipped with Greenleaf Spray Max 110015 nozzles to span the silking area.

    Drift Analysis

    Three free-standing 26’ (8 m) horizontal flux collectors were positioned in the corn field approximately 3’, or 1.5 rows from the downwind edge of the spray plot downwind of the area treated by drone (Figure 3). The sampling poles were positioned about 30’ apart parallel to the treatment block. Sterilized, 1.8 mm braided polyethylene collector line was run up the poles on pulleys just prior to application. Following applications, the line was collected in 1 m lengths into sealed bags.

    The assumption was that by placing the horizontal flux samplers as close to the “zero” downwind edge position as possible, nearly the entire off-swath movement of drift would be captured. A compromise of placing the samplers in the middle of the first row past the downwind swath edge was made due to the scale of the sample and the relative low swath precision of the drone. Placing the samplers closer to the zero downwind line was deemed to be too high a risk of inadvertently sampling in-swath.

    Figure 3- Moving horizontal flux poles into the field prior to positioning them for trials. String collectors were run up the poles just before spray application and retrieved immediately afterwards.

    Spray Solution (Formulated Product plus Tracer)

    Fungicide was applied at field rates (8 oz/ac or 586 mL/ha). The field sprayer applied this at 16.7 gpa. The drone applied it at 2 or 5 gpa but also included tracer solution at 0.2% (20 ml/10L solution) vol./vol. of a 20% mass/mass solution of PTSA in dH2O. PTSA residue data assumes 100% recovery and 0% degradation of the tracer. Tests of PTSA with fungicide prior to the study showed no physical antagonism and >98% tracer recovery. Prior testing of PTSA showed an acceptable 1-2% solar degradation in the timeframe required to collect samplers. Tank samples were drawn from the drone at the beginning and end of each trial and used to confirm tank concentration and to establish fluorescence curves.

    Weather Conditions

    Weather data was collected using a Kestrel 3550AG weather meter (Kestrel Instruments) in a vane mount positioned 1 m above the tassel (approximately 1 m below drone altitude). Data was logged every 5 seconds. Issues with data loss required us to supplement local data with Field Level Weather Summary data (Table 2).

    Date (2022)FieldVol. (gpa)Avg. Temp. (°C)Avg. Windspeed (km/h)Start TimeDuration (min.)
    Jul 2515*22.36.213:0035
    Jul 251521.47.518:4535
    Jul 26116.718.85.410:0045
    Jul 261223.97.715:3025
    Jul 2925**n/a16.411:0035
    Jul 2922***23.621.014:0025
    Aug 12320****25.46.313:3015
    Table 2- Date, location, and weather conditions for each treatment
    *Trial pass over spray collectors only – no horizontal flux collectors employed.
    **All bottom-level water sensitive paper samplers spoiled by high humidity. Wind changeable and horizontal flux poles moved 2x before application to orient downwind.
    ***Noted flocculation in tank samples likely from rainfastness adjuvant. Did not affect analysis.
    ****Coverage data from a single array of nine spray collectors with water sensitive paper samplers.

    Results

    Statistics

    The % applied rate ac-1, % area covered, and deposits cm-2 were subjected to analysis of variance using SAS® OnDemand for Academics PROC GLM. When a significant treatment effect was found, means were compared using Tukey’s honest significant difference test (HSD) at p=0.05.

    Data Collation

    Each spray collector was a vertical structure that supported Mylar samplers at three depths. Each depth held two samplers oriented abaxially or adaxially, in parallel with the ground. When discussing the amount of PTSA recovered by sampler depth or by sampler orientation, the % applied rate ac-1 of each of the nine related samplers were averaged within each array (n=2 arrays per block times two replicates equal n=4 per treatment).

    When considered from above, the six Mylar samplers are vertical cross-sections of the same area of ground. Therefore, the % applied rate ac-1 from each sampler was added to represent the total mass of tracer intercepted per collector. When these nine sub-samples are averaged, we arrive at the average % applied rate ac-1 per array.

    Similarly, the % applied rate ac-1 from each 1 m length of string on a horizontal flux collector could be averaged across collectors by relative position to explore drift by height (n=3 poles per block times two replicates equal n=6 per treatment). Alternately, the total PTSA recovered per pole could be calculated (n=3 poles per block times two replicates equal n=6 per treatment). This interpretation allowed us to perform a mass balance accounting of residue in-canopy and as drift compared to the known applied rate ac-1.

    It was not possible to collate the data in this fashion for the WSP because it was not possible to index % area or deposits cm-2 on a 1”x3” area to a theoretical maximum. Therefore, we averaged the nine samplers within an array relative to their position and orientation (n=2 arrays per block times two replicates equal n=4 per treatment) or averaged the six samplers per collector prior to averaging all collectors in an array (n=2 arrays per block times two replicates equal n=4 per treatment).

    RPAS Coverage – Mylar Samplers

    There is a negative linear relationship (r2=0.997) between the depth of the sampler and the average % applied rate ac-1 (Table 3). The deeper the sampler, the less tracer recovered. The sum of the average % applied rate ac-1 at each depth was 17.7% of known rate applied rate ac-1.

    Sampler DepthAvg. % Applied Rate ac-1Significance
    Top9.6A
    Middle5.7B
    Bottom2.4C
    Total:17.7–
    Table 3- The depth of the sampler had a significant effect on the overall average amount of PTSA recovered.

    The orientation of the sampler significantly affected the overall average amount of tracer recovered (Table 4). The abaxial surfaces intercepted an average 11.1 % applied rate ac-1 less (a 97% difference) than adaxial surfaces. Note: When Mylar was retrieved a few had physically shifted, potentially exposing the back side of abaxial collectors to primary deposition from above. Therefore, it is assumed that the actual deposit is lower than reported here.

    Sampler OrientationAvg. % Applied Rate ac-1Significance
    Adaxial11.4A
    Abaxial0.3B
    Table 4- The orientation of the sampler had a significant effect on the overall average amount of PTSA recovered.

    When we separate the data to focus on the volume applied, we see volume had a significant impact on the amount of tracer recovered (Table 5). The average % applied rate ac-1 was 2.1% less (a 58% difference) in the 2 gpa condition compared to the 5 gpa condition.

    FieldAvg. % Applied Rate ac-1Significance
    17.1A
    24.6B
    Table 5- The field location had a significant impact on the average amount of PTSA recovered.

    When we isolate the volume applied by field, the 2 gpa treatment resulted in less coverage in field 2 (average 1.4 % applied rate ac-1 or 28% less) and significantly for the 5 gpa treatment (average 3.6 % applied rate ac-1 or 41% less: Table 6).

    DateFieldVolume (gpa)Avg. % Applied Rate ac-1Significance
    Jul 25159.2A
    Jul 26125.0B
    Jul 29255.6C
    Jul 29223.6B
    Table 6- The average amount of PTSA recovered by date and location show lower overall recovery in Field 2.

    When sampler depth is included in the field analysis (Table 7), we see similar deposition patterns; a negative linear relationship between coverage and canopy depth in all treatments save the 5 gpa treatment in field 2. Closer inspection confirms a reduction in coverage for the 2 gpa condition in field 2 versus field 1, and a significant reduction for the 5 gpa condition in field 2 versus field 1.

    Sampler DepthField 1 – Jul 25: 5 gpa.
    Avg. % Applied Rate ac-1 (Sig.)    		Field 1 – Jul 26: 2 gpa.
    Avg. % Applied Rate ac-1 (Sig.)    		Field 2 – Jul 25: 5 gpa.
    Avg. % Applied Rate ac-1 (Sig.)    		Field 2 – Jul 29: 2 gpa.
    Avg. % Applied Rate ac-1 (Sig.)
    Top15.2 (A)8.3 (A)8.0 (A)6.7 (A)
    Middle9.1 (B)4.7 (B)6.1 (AB)2.9 (B)
    Bottom3.5 (B)1.9 (C)2.6 (B)1.5 (B)
    Total:27.814.916.711.1
    Table 7- The average residue recovered by date, location and sampler depth is significantly less in the 5 gpa condition in field 2 and does not distribute linearly by sampler depth.

    RPAS Drift – Horizontal Flux

    Overall, the volume applied had a significant impact on drift, where the 2 gpa treatment resulted in an average increase of 1.6 % applied rate ac-1 (44% difference: Table 8) versus the 5 gpa treatment.

    Volume Applied (gpa)Avg. % Applied Rate ac-1Significance
    23.6A
    52.0B
    Table 8- The volume applied had a significant impact on the amount of the PTSA recovered.

    As with the Mylar samplers, there was a “field effect” where the field had a statistically significant impact on the amount of tracer recovered (Table 9). However, unlike the Mylar samplers in the crop, more tracer was recovered in field 2 (average increase of 3.2 applied rate ac-1 or a 67% difference) than in field 1.

    FieldAvg. % Applied Rate ac-1Significance
    11.4A
    24.2B
    Table 9- The field location had a significant impact on the amount of PTSA recovered.

    The pattern of deposition by height was similar across all treatments. For context, note that the first 2.5-3 m of string were within the corn canopy and drone altitude was approximately 5 m off the ground (2 m over the tassels) per Figure 4 and 5. The differences were only statistically significant in field 2 (Table 10) where an average 33% applied rate ac-1 was intercepted compared to 11% in field 1.

    Figure 4- Average PTSA recovered (% applied rate ac-1) by height and field.
    Figure 5- Average PTSA recovered (% applied rate ac-1) by height and volume applied.
    Height
    (1m segment in m from ground)    		Field 1:
    Avg. % Applied Rate ac-1    		Sig.Field 2:
    Avg. % Applied Rate ac-1    		Sig.
    80.7A1.5C
    73.2A3.6BC
    63.3A8.8A
    52.7A9.9A
    40.7A4.9AB
    30.4A2.2BC
    20.1A1.8C
    10.1A0.7C
    Total:11.2–33.3–
    Table 10- The average amount of PTSA recovered by height for field 1 and field 2.

    The volume applied had a significant effect on the total PTSA tracer detected in both fields, with an average 4.4% applied rate ac-1 more (a 59% difference) recovered in the 2 gpa treatment (Table 11 and Figure 5). Separated by fields, the 5 gpa treatment had an average 1.4% % applied rate ac-1 more (a 77% difference) in field 2 and the 2 gpa treatment had an average 2.8% % applied rate ac-1 more (a 76% difference) in field 2.

    Volume Applied (gpa)Field 1:
    Avg. % Applied Rate ac-1    		Sig.Field 2:
    Avg. % Applied Rate ac-1    		Sig.
    22.2A5.0A
    50.7B2.1B
    Table 11- The volume applied had a significant impact on the amount of PTSA recovered.

    Mass Balance Accounting

    It is never possible to entirely “close mass” in spray studies because there are other surfaces (e.g. leaves) within the vertical profile that intercept spray, as well as off-swath deposition and the ground itself (not measured in this study). Nevertheless, the exercise does allow us to estimate and compare how much spray was captured and how much remains unaccounted for (Table 12). We see that the 2 gpa treatment in field 1 had the highest unaccounted-for fraction, and on average we were able to account for an average 53% of the applied rate ac-1 in this study.

    Field
    (Volume in gpa)    		Coverage:
    Avg % Applied Rate ac-1
    (A)    		Drift:
    Avg % Applied Rate ac-1
    (B)    		Total % Detected
    (A+B)    		Unaccounted Fraction
    [100-(A+B)]
    1 (5)5155644
    1 (2)26.51743.556.5
    2 (5)30245446
    2 (5)19.54059.540.5
    Table 12- Closing mass using % PTSA detected on in-canopy samplers and on drift collectors.

    RPAS and ground rig coverage – Water Sensitive Paper

    The depth of the sampler had a significant effect on the overall average % area covered at all depths (Table 13). However, there was no significant difference at the two lower depths for deposit density (Table 14). In both cases, the negative linear relationship between coverage and sampler depth corresponds closely to the PTSA recovered on the Mylar samplers (see Table 3).

    Sampler DepthAvg. Coverage (% Area)Significance
    Top2.80A
    Middle1.28B
    Bottom0.62C
    Table 13- Overall average % coverage by sampler depth.
    Sampler DepthAvg. Coverage (Deposits cm-2)Significance
    Top44.5A
    Middle17.9B
    Bottom7.2C
    Table 14- Overall average deposit density by sampler depth.

    The sampler orientation had a significant effect on both overall average % area covered (Table 15) and deposits cm‑2 (Table 16).

    Sampler OrientationAvg. Coverage (% Area)Significance
    Adaxial3.03A
    Abaxial0.12B
    Table 15- The orientation of the sampler had a significant effect on the average % area covered.
    Sampler OrientationAvg. Coverage (Deposits cm-2)Significance
    Adaxial43.5A
    Abaxial3.1B
    Table 16- The orientation of the sampler had a significant effect on the average deposit density.

    The treatment had a significant effect on the overall % coverage (Table 17) with the overhead broadcast condition covering an average 3.31% more sampler surface (a 60% difference) compared to the next highest treatment value. The directed application delivered a significantly higher 67 deposits cm-2 (a 72% difference) compared to the next highest treatment value (Table 18).

    Treatment (gpa)Avg. Coverage (% Area)Significance
    Broadcast (16.7)5.91A
    Directed (20)2.32B
    Drone (5)1.34BC
    Drone (2)0.55C
    Table 17- Overall average % coverage by treatment.
    Treatment (gpa)Avg. Coverage (Deposits cm-2)Significance
    Broadcast (16.7)92.6A
    Directed (20)25.8B
    Drone (5)22.9B
    Drone (2)5.9B
    Table 18- Overall average deposit density by treatment.

    When we increase resolution to include sampler orientation, we see high standard errors typical of the variability inherent to spray coverage analysis (Figures 6 and 7). The broadcast treatment had the highest average adaxial % area coverage and the second highest average deposit density. The directed treatment had the second highest average adaxial % area coverage and the highest average deposit density but had the highest overall average coverage on the abaxial samplers. RPAS coverage on all samplers was lowest overall and was relative to the volumes applied.

    Figure 6- Coverage (% area) by treatment, sampler depth and orientation.
    Figure 7- Coverage (Deposits cm-2) by treatment, sampler depth and orientation.

    Focusing on RPAS treatments, the orientation of the sampler significantly affected coverage (Tables 19 and 20).

    Sampler OrientationAvg. Coverage (% Area)Sig.Avg. Coverage (Deposits cm-2)Sig.
    Adaxial1.1A11.6A
    Abaxial0.0B0.4B
    Table 19- RPAS (2 gpa) coverage by sampler orientation.
    Sampler OrientationAvg. Coverage (% Area)Sig.Avg. Coverage (Deposits cm-2)Sig.
    Adaxial2.5A47.3A
    Abaxial0.2B7.5B
    Table 20- RPAS (5 gpa) coverage by sampler orientation

    Continuing to focus on the RPAS treatments, the depth of the sampler had a significant effect on overall average coverage at both 2 gpa (Table 21) and 5 gpa (Table 22). Just as with the average % applied rate ac-1 (included here for comparison), the overall average coverage on the top adaxial sampler was significantly higher than the other two depths for % area covered and deposits cm-2.

    Sampler DepthAvg. Coverage
    (% Area)    		Sig.Avg. Coverage
    (Deposits cm-2)    		Sig.Avg.
    % Applied Rate ac-1    		Sig.
    Top1.2A12.8A7.5A
    Middle0.4B3.9B3.8B
    Bottom0.1B1.3B1.7B
    Table 21- Coverage on the top sampler was significantly different than other depths at 2 gpa.
    Sampler DepthAvg. Coverage
    (% Area)    		Sig.Avg. Coverage
    (Deposits cm-2)    		Sig.Avg.
    % Applied Rate ac-1    		Sig.
    Top2.1A48.3A9.2A
    Middle1.1B17.6B5.8B
    Bottom0.9B16.4B2.3B
    Table 22- Coverage on the top sampler was significantly different than other depths at 5 gpa.

    Comparing data from WSP to Mylar Samplers

    There was a correlation between the % area coverage detected using WSP and the tracer recovered from the Mylar samplers. Deposit density provides valuable information about the distribution of spray over the target surface but does not always correlate with % area covered, and it is therefore omitted from this comparison. When we plot the average % area covered from the adaxial WSP against the average % applied rate ac-1 from the Mylar samplers, we see the same near-linear pattern of decay with depth (Figure 8).

    Figure 8- Average coverage from adaxial samplers plotted by depth and volume applied show similar coverage patterns.

    If we assume each top, adaxial sampler (irrespective of sampler material) represents the highest degree of coverage, we can assign it a value of 100% and index the data to this value. This allows us to visualize and compare the two sampler types directly (Figure 9) and illustrates similar relative coverage, but perhaps a greater rate of decay for the WSP.

    Figure 9- Average coverage from adaxial samplers plotted by depth and volume applied show similar coverage patterns. Normalized to top sampler.

    Net Revenue and Disease Pressure

    Crops were harvested at the R4 stage of development. There was no disease pressure detected in any field and no clear impact of application method on net revenue (Figure 10). Results based on the following formula: (CAD $/ac) = (Seed Yield × Corn Sale Price) – Drying Cost. No conclusions regarding efficacy can be drawn from this data.

    Figure 10- Net Revenue (CAD $/ac) by field and treatment.

    Key Observations

    1. Water Sensitive Paper (WSP) measurements of percent area covered (% area) and deposit density (deposits cm-2), and Mylar samplers measuring mass deposit (% applied rate ac-1), revealed similar coverage patterns, making both samplers viable methods for RPAS coverage analysis. These are complimentary methods that reveal different aspects of coverage. When possible, they should be used simultaneously to produce a more complete analysis.
    2. RPAS and conventional overhead broadcast applications produced similar deposition patterns in the corn canopy: A negative linear relationship between coverage and adaxial sampler depth was observed for most treatments (r2=0.997) and abaxial coverage was very low or more often, nonexistent. Further, overall coverage shared a direct relationship with volume for RPAS and conventional overhead broadcast applications.
    3. Directed applications in this study employed a finer spray quality, released laterally from within the canopy. This produced a different coverage pattern than the RPAS and overhead broadcast applications. Per WSP, this treatment resulted in the highest overall deposit density and was the only treatment to produce significant deposition on abaxial surfaces.
    4. For RPAS, spray coverage was significantly reduced by -58% (based on avg. applied rate ac-1), by -59% (based on avg. % covered) and by -74% (based on avg. deposits cm-2) and drift was significantly increased by +73% for the 2 gpa treatments versus the 5 gpa. We attribute this primarily to drone travel speed, which increased from 3.3 m/s at 5 gpa to 7 m/s at 2 gpa. For context, and with certain exceptions, travel speed shares a negative relationship with spray coverage and a direct relationship with drift in airblast and field sprayer applications.
    5. There was a “field effect” where field 2 had lower overall RPAS coverage for both 2 and 5 gpa treatments. Compared to field 1, by -28% for the 2 gpa treatment, and by -41% for 5 gpa. Average drift increased by +76% for 2 gpa and by +77% for 5 gpa. We attribute this to the significantly higher wind conditions in field 2.
    6. Given the lack of disease pressure in the two fields, and the lack of any significant difference in revenue by treatment within each field, efficacy is inconclusive. This study represented only two of eight fields in a larger RPAS efficacy trial where five locations had disease pressure high enough to rate. Preliminary results suggest that Tar Spot control from a 5 gpa drone application may be comparable to that of a 16.7 gpa overhead broadcast application from a field sprayer (data not shown).

    Summary

    Drone and conventional overhead broadcast treatments deposited spray in a similar pattern (a negative linear relationship with canopy depth and very low or no abaxial coverage), irrespective of the method used to analyze coverage. RPAS produced significantly lower coverage than the conventional overhead broadcast treatment, which is attributed primarily to the low volumes employed, per the direct relationship between volume applied and overall coverage (up to some point of diminishing return). High ambient windspeed significantly increased drift in both the 2 and 5 gpa conditions and reduced spray coverage. High travel speeds (required to apply 2 gpa) likely contributed to the significantly increased drift and reduced coverage in that treatment versus 5 gpa. For the use cases explored in this study, low volumes and high travel speeds are not advisable for RPAS, particularly in high wind conditions. Future work separating the travel speed and ambient wind speed variables would clarify their relative influence on RPAS drift and coverage.

    This video presentation is covers the highlights of the study. And disregard the verbal slip-up: we didn’t travel 110 mph.