RPAS swathing in broad acre crop canopies

Posted on

About Jason Deveau (Spray Guy)

Dr. Jason Deveau has been the OMAFA Application Technology Specialist since 2008. He researches and teaches methods to improve the safe, effective and efficient application of agricultural sprays in specialty crops, field crops and controlled environments. He is the co-administrator of Sprayers101, co-author of the Airblast101 Textbook, a slow cyclist and an even slower runner.

See all posts by Jason Deveau (Spray Guy).

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)

A swath’s effective 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 are attractive because they mean fewer passes and faster applications. It is plausible that inappropriate 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, 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

ESW 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 ESW width at lower versus higher spray volumes. Lower volumes are typically comprised of finer droplets, which are more accurately resolved using deposit counts. Swaths 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.

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.

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).

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 single value: the effective swath width.

Converting to Effective 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 ESW 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 ESW 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 ESW 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 on ESW

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 ESW shares an inverse relationship with the canopy depth. However, the relationship reversed when ESW 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 ESW in-canopy.

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 ESW

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 ESW 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 ESW

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?

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 ESW between stubble and bare ground. The ESW on “ground” was far closer to the intended 8 or 10 m swath width when expressed as deposit density.

There appears to be a significant reduction of ESW in-canopy versus 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 such different planting architectures and plant physiologies.

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

It is understood that finer spray is better able to penetrate canopies, due to the droplet’s potential for entrainment in air currents and by mathematical odds given that finer spray carries more droplets than a coarser spray. 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 ESW both on bare ground and in-canopy.

Author

  • Dr. Jason Deveau has been the OMAFA Application Technology Specialist since 2008. He researches and teaches methods to improve the safe, effective and efficient application of agricultural sprays in specialty crops, field crops and controlled environments. He is the co-administrator of Sprayers101, co-author of the Airblast101 Textbook, a slow cyclist and an even slower runner.

    View all posts