Category: Spray Basics

  • Assessing Water Sensitive Paper – Part 2

    Assessing Water Sensitive Paper – Part 2

    This is part two of our three-part monster-article on methods for digitizing and processing water sensitive paper. You can read part one here.

    Image analysis software

    There are many choices of software designed to analyze digitized WSP images (E.g. Optomax, Stainalysis Freeware, DropVision, ImagePro Plus, DropletScan, AgroScan, DepositScan, UTHSCA ImageTool). Some were developed for aerial applicators to evaluate entire swaths and others to focus on single collectors. Some are more user-friendly than others, some cost money, and some are no longer supported. All of them employ algorithms (a set of rules a computer follows when making calculations) that often make image processing decisions. Sometimes these algorithms are pre-set, which can be convenient but may also restrict our analysis.

    ImageJ is a free, open-source application developed at the National Institute of Health by Wayne Rasband to adjust and analyze high-resolution images of small structures. There’s a variation called “Fiji” (Fiji Is Just ImageJ) which bundles ImageJ with tools specifically intended for biologists. Happily, they are equally valuable for analyzing WSP. The interface can be intimidating, but only because there are so many functions that we won’t be using. The learning curve is worthwhile because the user has complete control over the analysis.

    The ImageJ menu. Version 1.53e.

    Three steps to image analysis

    No matter the software, the operations used to analyze a digital image tend follow a three-step progression:

    1. Pre-processing: We select an ROI (a Region Of Interest) in the image and perform a few preliminary operations to improve image quality and contrast. In selecting a specific region, we can avoid unwanted flaws like drips or fingerprints as well as crop the image to some standard size for scaling purposes.
    2. Processing / Detection: Point and Morphological Operations are used to refine the image and establish a threshold so we can differentiate between deposits and the unstained background. The ideal outcome sees the original colour image converted to a binary (typically black and white) image.
    3. Measurement: We use ready-made computational routines to quantify some value. Typically, the percent area covered by deposits, but possibly the count and density of those deposits and perhaps even an estimate of the original droplet size.

    Let’s explore each of these steps.

    1. Pre-processing step

    Pre-processing establishes the scale of the image and allows us to isolate the specific region we want to analyze. Perhaps the water sensitive paper was folded during sampling and we want to analyze each half separately. Perhaps we want to avoid obvious imperfections that would interfere with our results. In some cases, pre-processing might also include adjusting the image brightness to improve the contrast between stains and the yellow background.

    Flaws and imperfections become obvious when water sensitive paper is examined under magnification. Part of pre-processing is to select a region of interest that represents typical coverage and does not include artifacts that might interfere with the analysis.

    2. Processing / Detection step

    Processing and detection can take time because of the degree of computation involved. The higher the resolution and the larger the ROI, the longer it will take. Depending on what you want to measure, it might be acceptable to sacrifice some accuracy for speed.

    We begin by determining which pixels represent part of a deposit stain and which represent part of the unstained background. We can accomplish this through global point operations called Thresholding and Filtering. If you haven’t already noticed, image analysis includes has a lot of jargon: “global” means the entire image and “point” refers to our focus on individual pixels. Ultimately each pixel is assigned one of two values, reducing the image to a binary (or 1-bit) format.

    Once we have a binary image, we explore the shapes of the deposits (which are sometimes referred to as objects) to determine the limitations of what we’re confidently able to measure. Morphological operations are used to refine or modify these shapes in order to smooth jagged edges and identify whether an object is the result of a single deposition or multiple overlapping deposits.

    3. Measurement step

    Depending on the image analysis software, the user may be limited in what they can measure. The spectrum ranges from a single value (usually percent area covered) to in depth data relating to each object in the image. The latter might appear in a pre-formatted report, or as a CSV (Comma Separated Value) file for further exploration in spreadsheet format.

    Thresholding

    A thresholding operation sorts all the pixels in an image by some characteristic, and then allows us to set a threshold dividing them into two camps. In our case, we want to divide them into “stained” and “unstained”. The process is almost like taring a scale, where anything above the weight of the container is identified as the weight of the contents.

    Thresholding is like taring a scale. Just as the weight of the container is isolated from the total weight of the container and contents, the stain colour is isolated from the background colour.

    The HIS thresholding operation

    ImageJ’s Colour Threshold operation is only one way to threshold an image, but it serves as a good example. This method uses HIS (Hue, Intensity and Saturation) to separate the deposit stain colours (blue-green) from the background colour (Yellow). As discussed, each pixel is represented by one or more 8-bit values. In this case pixels represent 0 – 255 hues, 0 – 255 intensities and 0 – 255 saturations. That may seem intimidating, but we mostly focus on hue.

    When the Colour Threshold operation is selected, ImageJ sorts all the pixel values in the image into a binned histogram (where the Y-axis is the pixel count and the X-Axis is the range of pixel values).

    a. Hue

    We begin by considering the hue, which is simply another word for colour. An image with no stains would produce a histogram with pixel colours producing a distinct peak in the yellow range. An image with stains would also display peaks in the blue-green range. The user then segments the background from the foreground by manually setting the threshold between these peaks.

    (Top) When the threshold overlaps the background yellow hue (set to 30 here), some portion of the background is falsely identified as a stain. (Bottom) When the threshold is adjusted to fall in between background hue and stain hues (set to 36 here), a sharper distinction is made.

    The hue thresholding process is less reliable (or can outright fail) when WSP has coverage in excess of 50%. This is because the color of the intermittent unstained areas changes as the distance between stains decreases. Think of it as blue bleeding into the yellow. The result is that the level of contrast between stained and unstained regions is inconsistent, making it difficult to confidently differentiate between “stained pixels” and “unstained pixels”. Similar issues arise when humidity causes the background colour to change but this tends to be more uniform and easier to threshold.

    If the hue threshold it is set too low, stains will appear smaller and lose their shapes. If it is set too high, stains will appear larger and the gaps between adjacent, separate stains can disappear. This can have a significant impact on deposit count and distribution assessments, resulting in the loss of thousands of tiny, distinct deposits. Threshold accuracy has less impact on the determination of percent area covered. Research has shown that the use of a single threshold for multiple papers gives an absolute error of +/- 3.5% area covered. This is considered well within the intrinsic variability of spray coverage data.

    b. Intensity

    Sometimes referred to as “Value”, intensity can be thought of as pixel brightness. No threshold is required here because capturing the entire 256 pixel value range improves the contrast between colours.

    c. Saturation

    Finally, saturation (a measure of the difference between red, green, and blue levels) is a useful thresholding adjustment when WSP has been exposed to humidity. Humidity does not affect WSP’s ability to resolve stains, but as we mentioned it can cause the background to take on an overall greenish hue. An increase in low end saturation limit can increase the contrast between the stain colour and a less-distinct background colour.

    When the HIS thresholds are set, ImageJ converts the pixels values closer to the foreground (stains) to black and those closer to background (yellow) to white. Users can invert this if they wish, or even make the stains red. The important part is that we now have a binary image that makes a clear distinction between the stains and the unstained background. Ideally, thresholding should be performed for each consistent set of samples.

    Learn how HIS thresholding is being used to perform weed recognition functions on sprayers in this article.

    This 2 cm x 2 cm ROI was HUI thresholded. You can double check your accuracy by having ImageJ “show outlines” which outlines and numbers each distinct object. Zoom in to see if any artifacts remain (or were inadvertently created) and go back to make minor thresholding adjustments if needed.

    Pixel filtering

    We won’t belabor filtering because it isn’t often required when analyzing WSP. Filtering operations compare pixel values to those of their neighbours and then replace those values with some form of weighted average. This reduces the relative differences between pixel values, smoothing the image and reducing noise (at the cost of lost detail).

    Last article –Part three: Morphological Operations and Interpretation.

  • Assessing Water Sensitive Paper – Part 1

    Assessing Water Sensitive Paper – Part 1

    This is not a typical article for www.sprayers101.com. We like to develop actionable, data-driven content written in an easily-read format. Some articles discuss the results of research, some describe best practices and techniques, and occasionally there’s a song parody. But this article is different.

    We recently wrote about the three commercially-available brands of water sensitive paper (WSP). The article was an impartial comparative evaluation of how these papers resolved spray coverage. But in order to be fair in our evaluation, we had to decide the best method for assessing them. This led us down a rabbit hole far deeper than we could have imagined. The science of image processing is complicated and there are many ways that WSP can be interpreted using a myriad of home-grown and commercial methods. We decided to share what we learned.

    Which method is best for you? How far should we take a tool that was originally developed for quick, subjective comparisons? What follows is a three-part primer in digitizing and analyzing water sensitive paper. If you’re a grower that has never used WSP, you need only read to the end of the next sentence. Buy it and try it. If you’re a consultant, a researcher, or just interested in wringing all you can from this excellent agronomic tool, then get comfortable.

    Here, in part 1, we’ll explore a brief history of WSP, describe a few limitations in what it is capable of resolving, and start down the road of how to capture a digital picture for later analysis. Welcome to the world of blob-analysis.

    Introduction

    Spray coverage describes the degree of contact between spray droplets and the target surface area. This metric can be used to predict the success of an application. One of the easiest methods for visualizing coverage is to use water sensitive paper, which is a passive, artificial collector that turns from yellow to blue when contacted by water.

    WSP is often used to evaluate iterative changes to a spray program. Placed strategically throughout a target canopy, or directly on the ground, achieving uniform, threshold coverage translates into improved efficacy, reduced waste, reduced off-target contamination and reduced risk of pesticide resistance development. WSP tends to underestimate the spreading effect that can occur on plant surfaces (especially when surfactants are used), but they are effective as a relative index.

    The simplest use of WSP, and the primary reason it was commercially developed, is to perform qualitative assessments. For example, when observers judge one paper to be visibly “bluer” than another, subsequent measurements have shown it can represent an increase of 20% in foliar deposit. In other words, if you can perceive a difference with the naked eye, it likely represents a biological impact. This fast and rudimentary use of WSP provides immediate and actionable feedback and is therefore valuable information for any sprayer operator. It has been suggested that manual counts become impractical at ~200 stains/cm2, but higher counts are possible using a loupe or linen tester-style lens.

    A loupe or folding linen tester (originally designed to check the quality of woven fabrics) provides 5-10x magnification to resolve smaller stains. Every deposit counts.

    WSP can also be used for in-depth, quantitative assessments. This requires a camera or scanner to produce a digital image of the WSP and specialized analytical software to extract the relevant data. Considerable research has been performed to establish the limits of what can be learned from WSP. The four pieces of information commonly sought are listed here from easiest and most reliable to hardest and least reliable (or arguably, impossible) to determine.

    • The percent surface area covered.
    • The density of deposits on the target area.
    • The size of the droplets that left the stains.
    • The dose applied to the target surface.

    This document will describe the fundamentals of image analysis and provide examples of commercial tools and protocols used to extract coverage data from WSP. It will also describe the assumptions and the limits intrinsic to these methods so the user can decide the degree of time and effort to invest versus the reliability of the results.

    A brief history of WSP

    In 1970, a journal article described a new method for sampling and assessing spray droplets. Photographic paper treated with bromoethyl blue created a yellow surface that changed colour when it encountered moisture. The pH-based reaction was fast and irreversible, leaving a distinct blue stain to mark the deposition.

    Ciba-Geigy Ltd. made water sensitive paper commercially available in 1985 (later as Novartis in 1996 and as Syngenta since 2000). It is produced in several formats, but aluminum foil packages of 50, 76 x 22 mm (1 x 3 in.) papers are the most popular. Odds are if you’ve ever used water sensitive paper, it originated from Syngenta in Switzerland. Recently, two new options have been made commercially available: Innoquest’s SpotOn Paper (United States) and WSPaper (Brazil).

    Once dry, the blue stains on WSP are irreversible and papers can be stored for a few years. However, unstained portions will continue to react to moisture from humidity, dew, or fingerprints, so care must be taken in their handling and storage. According to Syngenta, stains can be permanently fixed using isopropanol (or a similar solvent) to remove the yellow layer, leaving black stains on white paper.

    Limitations of water sensitive paper

    a. Minimum drop detection diameter

    In hot and dry conditions, not all droplets that contact WSP will leave stains. En route from nozzle to target, droplets can concentrate through evaporation, leaving insufficient water to stain the paper. Syngenta states that droplets <100 µm in diameter will not be reliably resolved in “tropical conditions”. For most conditions, their minimum droplet diameter is closer to 50 µm. Innoquest states that their minimum drop detection diameter, under most environmental conditions, is between 60 and 90 µm.

    This isn’t to say that smaller droplets can’t be detected. In absolutely ideal conditions, the smallest detectible droplet diameter for any brand of WSP is closer to 30 µm (Syngenta, Innoquest, SprayX – Personal Communication). Microscopic analysis of Syngenta’s papers reveals that droplets finer than this can leave physical “craters” on the surface, but have insufficient water to cause the colour change. The stain diameter created by a droplet is always larger than the droplet diameter, to a degree that is dependent on the spread factor.

    b. Spread factor

    The size of a stain is sometimes used to extrapolate the size of the droplet that produced it. The stain diameter is divided by a spread factor, which is determined under specific conditions. For example, Syngenta’s spread factors were established using the magnesium oxide and silicon‐oil‐method at 20°C and a 40% relative humidity for droplets at sedimentation velocity. “Sedimentation velocity” can be thought of as terminal velocity, which accounts for the fact that droplets moving at higher speeds will leave larger diameter stains. Consider the splash produced by a water balloon hitting a surface fast or hitting it slow.

    Spread factors are not constant for all droplet sizes. For Syngenta’s WSP, a 59 µm droplet is expected to leave a 100 µm diameter stain (a spread factor of 1.7) and a 285 µm droplet is expected to leave a 600 µm diameter stain (a spread factor of 2.1). This relationship is sometimes captured using calculus. One research article employed this formula: feDm = 0.74057 + 0.0001010399 × Dm + 0.02024884 × ln(Dm) (where fe is spread factor and Dm is stain diameter in microns). Volume was then calculated per: Vg = (π × Dg3) / 6 (where Vg is droplet volume in µm3 and Dg is droplet diameter in µm). Innoquest determined their spread factors to be [0.4508 × Ln(Observed Stain Diameter)] – 0.6221 (Personal Communication).

    Given that droplet sizing excludes the finest droplets, relies on situation-specific spread factors, assumes the droplet has reached terminal velocity and can be stymied by overlapping and elliptical stains (discussed in the next installment of this document) it is questionable whether there is any practical value in the exercise except perhaps for a relative comparison under highly controlled conditions.

    Digitizing WSP

    Digital images are produced using cameras or scanners. Cameras employ a grid of light-sensitive sensors, each of which reflect and record their portion of an image. Cameras capture images quickly but are prone to focus and distortion issues because the lens must be held very close to the WSP. SprayX’s DropScope accounts for this by individually calibrating each unit to compensate for variation during assembly and by employing software to account for lens distortion. When high resolution is required, cameras are the more expensive and complicated option. However, when resolution is not an issue, even a smartphone camera can be used (as with the SnapCard app; which as of 2026 may no longer be available).

    Flatbed scanners press multiple papers against a glass platen fixed above the light-sensitive sensors. This minimizes potential focus issues. Compared to cameras, scanners experience less distortion because they do not use a fixed grid of sensors. Instead, they rely on the speed and consistency of a carriage motor that draws an array of sensors along the image, capturing discrete slices. Scanners are less expensive than cameras, but they are much slower and low-end varieties can sometimes skip tiny slices of the image. This is not an issue when scanning office documents, but it can cause problems when analyzing a high resolution image.

    a. Pixels

    A pixel (a contraction of Picture Element) is the numerical information recorded by a light-sensitive sensor. The word “digitizing” means “converting to numbers”. The most rudimentary pixel value is an eight digit (or 8-bit) number. Each bit is either a 1 or a 0, so each pixel value is one of 28 (that’s 256) possible unique combinations. A picture displayed as a grid of 8-bit numbers wouldn’t make sense, so the computer substitutes shades or colours according to a look-up table.

    Today’s sensors report higher pixel values to give more depth to the digitized image. An RGB (Red, Green, Blue) image records separate 8-bit values for red, green and blue colours. That’s 2563, or 16,777,216 possible unique colour combinations for a single 24-bit pixel.

    With such a nuanced spectrum, two colours might look the same to the naked eye but represent different pixel values. Therefore, image analysis is more precise when we can work with the pixel values (the actual numerical data) and not shades or colours (an interpretation of the data).

    (Left) A digitized scan of water sensitive paper. (Middle) Zoomed in on a single deposit. (Right) Extreme zoom to show the actual pixels, both as look-up table colours and as 8-bit pixel values. The colours may appear similar, but the actual pixel values are different.

    b. Resolution (scale)

    Before we can analyze an image, we must first know the scale. Each pixel is the smallest element in a grid that makes up the digital image. The scale of the image determines the real-life size that each pixel represents, making it possible to calibrate size measurements. We often refer to image resolution in Dots Per Inch (DPI). In this case, “dots” refers to pixels. The higher the DPI, the higher the resolution as the diameter of each pixel represents a smaller real-life length.

    Camera resolution is described in terms of megapixels (MP) where 1 MP represents a grid of light-sensitive sensors capable of producing a 1 million-pixel area. When planning to print an image, the convention is to use a minimum resolution of 300 pixels per inch. For example, a standard 8 x 10 in. print would need 2,400 x 3,000 pixels for a total area of 7.2 million pixels. This would require a 7.2 megapixel camera.

    The area we are dealing with is typically less than the entire 1 x 3 in. paper. Even if we captured the entire paper, a 1 MP camera would provide 600 pixels per inch, or approximately twice the resolution required for a typical 8 x 10 in. print. That may seem sufficient, but remember we are examining the image very closely, which would be similar to blowing the print up to the size of a billboard. For reference, the SprayX DropScope uses an 8 MP camera. The iPhone 7 camera used to capture SnapCard images for this document is 12 MP.

    So, what is the ideal resolution and what are the downsides of getting it wrong? A low resolution image has a low pixel density, which might cause us to see multiple deposits as a single deposit or to miss the smallest deposits entirely. The minimum diameter of a detectable deposit must be about the same as the imaging resolution. For example, if a pixel represents a 30 μm diameter, the smallest deposit we could reliably resolve would be about the same size. Software that registers deposits sizes less than the limit of resolution are likely due to an algorithm error and should be ignored.

    (Left) The scale of the original image is known. (Middle) A close-up of a low resolution image with pixels calibrated to scale. Is this a single stain or a cluster of multiple stains? (Right) A close-up of a high resolution image with pixels calibrated to scale. It is easier to see this is likely a single deposit.

    This is further complicated by stains that lie at an intersection overlapping multiple pixels. In this case, more than one pixel might represent a colour that is blue enough to register as stained, reporting a larger deposit than was actually there.

    Resolution errors. (Left) Low resolution can cause pixels to misrepresent small, discrete deposits as a single, large stain. (Middle) Pixels may not reflect deposits smaller than their diameter. (Right) Multiple adjacent pixels may falsely represent a single, smaller, intersecting stain.

    It is tempting to go to the highest resolution possible, but this can also cause problems, such as detecting and misidentifying inconsistencies in the surface texture of the paper as stains. Additionally, high-res images create logistic issues; They take longer to scan and to process as well as create large files that take up a lot of storage space. Image formats (e.g. JPEG) can compress the image file to make it smaller, but data is lost. Other formats (e.g. TIFF, PNG, BMP) are not as efficient at saving space, but they preserve the original data and are therefore preferred.

    We suggest that 10 µm : pixel provides enough resolution, a reasonable processing time and a manageable file size. Further, given that a deposit could overlap multiple pixels, we propose employing a filter that removes any deposits less than a three pixel, or 30 µm, span. This lower limit eliminates artifacts and is still smaller than the smallest stain WSP can possibly produce. Some software allows the user to set this limit, and some make the choice on our behalf.

    Next article: Image analysis software and thresholding.

  • Optical Spot Spraying and AI Scouting

    Optical Spot Spraying and AI Scouting

    Listen to the Audio article here

    Site-specific treatments have long been a goal in agriculture. It makes sense to provide inputs or treatments at rates that reflect the local situation. And to a large degree, those capabilities have been available for fertility and seed inputs for some time, with input zones reflecting soil types or topography.

    Typical prescription map for nutrients (Source: Field Crop News)

    But the sprayer world has not seen as much site-specific treatment. One reason is that pest maps are time-consuming to generate and their usefulness may be short-lived. Or perhaps weeds are fairly ubiquitous, and it usually makes sense to treat an entire field. Another reason could be that sprays are relatively inexpensive compared to fertilizer or seed.

    For spraying, we need to re-define site-specific.

    While traditional zone maps (corresponding to, say soil type and/or elevation or slope position) allow unique treatments on a scale of acres, new sensors have allowed sprayers to basically leapfrog this approach and treat each square foot uniquely. These sensors identify plants directly and create an immediate treatment response.

    Optical Spot Spray(OSS) principle (adapted from WEEDit)

    The idea, and technology, has and been around agriculture since the early 1990s, with the Concord DetectSpray and later the Trimble WeedSeeker. For various reasons, these two never became widespread in North America, although a significant market formed in Australia and New Zealand.  New cutting edge technologies are about to change this.

    Green on Brown

    Two main manufacturers have occupied the traditional Green on Brown Optical Spot Spraying (OSS) space, the Trimble WeedSeeker and WEEDit. Both have been available for over 10 years and are well established and proven reliable. WeedSeeker uses the Normalized Difference Vegetation Index (NDVI) principle to detect green on a non-green background. It employs one sensor per nozzle and the nozzle is either on-or off based on what the sensor detects. The WEEDit system is manufactured in the Netherlands by Rometron (https://www.weed-it.com/), and is widely adopted for use in Australia and South America. It is now making inroads into North America. The most recent version is named Quadro.

    WEEDit spray booms contain sensors placed at 1 m intervals. These scan the ground ahead of the boom, identify the presence of plants, and trigger the nozzle in line with the plant. The newest Quadro sensor contains four channels so that its resolution is actually 25 cm (10″) wide. The boom therefore contains a nozzle every 25 cm, and this nozzle has a correspondingly narrow fan angle that treats just this space.

    Hypro even spray (banding) nozzle with 30 degree fan angle. 30 and 40 degree nozzles are currently installed on WEEDit on 10″ spacing.
    30 degree fan achieves approximately 8″ to 10″ band at target height. Boom stability is important

    The detection principle is based on the quality of light that is reflected from living plant tissue compared to everything else. A red (older generation) or blue (newest generation, Quadro) light is emitted, and chlorophyll-containing plants reflect a unique wavelength that differentiates them from ground or dead plant material.

    Older generation WEEDit sensors were placed at 1 m intervals and had five channels, each covering a 20 cm band. There were 180 nozzles on a 36 m (120′) boom.

    The response time of the system is very fast. Triggered by small solenoids, a sprayer travel speed of up to 15 mph is possible when the sensor looks 1 m ahead. Furthermore, the software allows the user two important controls: first, the sprayed distance before and after a detected plant can be buffered between 5 and 20 cm, resulting in a sprayed patch between 10 and 40 cm long. This could be useful when boom heights fluctuate and placement of the sprayed patch shifts accordingly. Second, the user can select from among four sensitivity settings. Higher sensitivity can detect smaller weeds but will also result in more false results.

    WEEDit Quadro sensor

    One reason the system has been successful in the southern hemisphere is the long growing season that may require multiple spray passes outside of the crop each year, and in which the weeds are relatively large at treatment time and therefore easier to detect.

    Water sensitive paper can be used to show whether a target has been detected (and therefore sprayed).

    In North America, the pre-seed spray window is relatively narrow and weeds may be very small or just be emerging. The risk of a miss due to non-detection is therefore greater. Fortunately, the WEEDit system has a feature that addresses this risk.

    PWM valve for WEEDit, capable of instantaneous response at 10 to 50 Hz

    The solenoids that trigger an individual nozzle are pulse-width modulated (PWM). This means that the application rate is adjusted according to travel speed via a duty cycle. And it offers an innovative capability: The entire boom can be programmed to spray a defined fraction of the full dose, to a maximum of 50%, as a background broadcast rate (called “Dual Mode” or “Bias”). The smallest weeds that escape detection are likely to be susceptible to this lower dose. Larger weeds are then detected and sprayed with an individual spot spray at the full dose. Dual Mode is typically set to about 25%; overall savings are less, but control is improved for those very early season situations.

    A WEEDit Quadro boom can also be operated in “Cover Mode” for broadcast spraying where it functions as a full PWM system with turn compensation.

    Currently, several hundred WEEDit sprayers are operating in Australia, and they’ve been available in Canada and the US since 2017. in 2019, Croplands, an Australian sprayer manufacturer owned by Nufarm, started representing WEEDit in Canada. It is available as a retrofit on existing booms, and can be ordered with a WEEDit Millennium aluminum boom that contains mounting brackets and wiring harness channels. Savings compared to broadcast spraying range from 65 to 85%.

    In early 2021, John Deere announced its entry into the Green on Brown space with See & Spray Select™. This system is built around the ExactApply nozzle body and uses RGB cameras to differentiate green plants from non-green background colours. It will be in fields in 2022 according to John Deere. Similar RGB-based systems are in development by other manufacturers. Although their performance has not been compared side by side with WEEDit or WeedSeeker, initial specs suggest that the RGB systems are slower and are less able to detect small plants. Nonetheless, the future looks very promising.

    In 2021, Hardi Australia announced a new product, called GeoSelect. This system does not have boom-mounted sensors, and instead sprays according to a prescription map developed by a drone. The advantage of this system is that the amount of herbicide needed is known in advance of spraying, and the knowledge of weed distribution in the field can allow for a more efficient coverage plan to be used. This system allows for spraying under any light condition, and adjusts for boom sway to ensure accurate placement. Drone map development is the responsibility of the applicator.

    Green on Green

    Green on Green spraying, which detects weeds within a crop and differentiates them from that crop, is advancing and the earliest commercial releases are now available in Australia, offered by a partnership between Bilberry and Agrifac (WeedSmart podcast here), as well as Bilberry and Goldacres with Swarmfarm. Others, notably the SmartSprayer from Amazone in partnership with Xarvio and Bosch and Greeneye Technology are entering field testing with commercial sized units in 2021 and 2022, respectively.

    Opportunities for Optical Spot Spraying

    Taken as a whole, optical spot spraying offers a number of opportunities for weed management.

    Cost Savings: OSS has an appealing rate of return on investment. On a 5000 acre farm, a pre-seed treatment of glyphosate plus tank mix for resistance management may cost $10/acre, or $50,000 per year. At an average savings of 75%, that represents $37,500 per year. Add other non-crop uses, such as post-harvest, and savings increase. With eventual weed recognition in-crop, virtually all herbicide treatments are candidates for such savings.

    Herbicide Resistance Management: Delaying the onset of herbicide resistance requires the use of multiple effective modes of action in a tank mix. Cost is a deterrent to this practice. With OSS, these tank mixes become affordable.

    Efficiency: With 75% product savings, a tank of product will last longer. The time lost to hauling water and product, as well as filling the sprayer, will decrease. For example, WEEDit users are spraying a full day on a single load. Or they may choose to use a much smaller load, decreasing equipment weight.

    Pre- and Post-Harvest: Whether for desiccation or weed control, site-specificity of late season sprays can also be based on living tissue. Only regions in the field requiring the desiccant are treated. Perennial or late-season weeds are selectively controlled pre-harvest. Since herbicide rates in these applications are typically higher, savings are significant.

    High value crops: Row crops requiring multiple fungicide applications per season, such as potatoes, can benefit from OSS. Sprays applied prior to canopy closure can thus avoid gaps between plants, saving product.

    Producer Innovation: One user of the WEEDit system in Saskatchewan developed an innovative use. Having missed a pre-seed spray, the applicator was faced with large weeds in a 1-leaf RoundupReady canola crop. By turning down the sensitivity of the system so the canola crop did not trigger the sensors and turning on Dual Mode, he was able to broadcast spray the field at a low glyphosate dose (sufficient to control the small weeds) and then apply a full dose to the larger weeds, triggered by the sensor.

    Equipment Innovation: Since individual zones or weeds require unique doses or products, technologies like direct injection, remote nozzle switching, multiple smaller tanks and booms, and PWM will make more sense and grow. But the whole concept of detection and treatment can be moved away from pesticides to mechanical control or other techniques such as lasers, as does Carbon Robotics.

    License to Farm: OSS makes intuitive sense not only to applicators, but also to the public at large. Showing and using these technologies demonstrates stewardship practices that are easy to communicate and understand.

    Artificial Intelligence Scouting

    Another approach is pioneered by several companies, for example Dronewerkers in the Netherlands (https://www.dronewerkers.nl/english/) Taranis (http://www.taranis.ag/), and Xarvio (https://www.xarvio.com). These companies have developed plant recognition algorithms that are currently able to identify over 100 different species. Each species can be divided into several growth stages. Taranis has launched a business in North America that scouts fields by high-resolution drone imagery, and then provides customers with maps that highlight potential agronomic issues such as weeds, disease, or insect damage.

    Example of information available from artificial intelligence scouting. In this case, plant and foreign material information by species, relative abundance, and growth stage.

    Resolution of the output can be species-specific (lambsquarters vs redroot pigweed), or by coarser resolution (broadleaf vs grass). The resulting output then shows the plant density at each location.

    Weeds in a soybean crop (courtesy of Taranis)

    Xarvio Scouting is a product in their Field Manager line (https://www.xarvio.com/en-CA/Scouting). App-based, the agronomist or producer takes pictures of their crops and the app is able to recognize weeds, diseases, insect feeding damage, as well as nitrogen status. The app is aware of other users in the area and basically crowd-sources emerging agronomic issues as they arise, communicating them back to the user.

    The Xarvio Scouting app can identify certain weeds, diseases, and insect feeding damage from pictures taken while scouting (Screenshot from Xarvio.com).

    The agronomic value of this information is clearly very high. Imagine knowing the distribution of weeds by species before and after treatment. Although we can already assess this when we walk fields, by conducting the task via drone we are measuring on a wide scale, permitting an accurate quantification of the treatment effect so its value can be assessed. This level of measurement intensity was not possible before. Yield loss models for time of removal of certain weeds at certain growth stages can be applied across the entire field, and economic analyses allows follow-up treatments to be tailored to specific portions of the field.

    Green-Eye Technology artificial intelligence can differentiate these ragweed plants from the pea crop. (Courtesy Green Eye Technology).

    Or imagine following specific patches of weeds over time, to monitor the effectiveness of a certain cultural practice, or be alerted to the establishment of a resistant population while it’s still feasible to contain it.

    Heat maps can be generated to document weed patches, and perhaps monitor their size over time. (Courtesy Green Eye Technology).

    When this information is converted to a prescription map, rate and tank mix composition (or cultural controls) could be varied as necessary by zone, or weeds could, in the future, be sprayed individually. Perhaps future autonomous robots could be deployed more efficiently.

    Identification of plant symptoms in canola (Courtesy of Taranis)

    Development and improvement of these technologies is ongoing rapidly. Finally, we may have all the pieces that can bring site specific weed, disease, and insect management to market.

  • Ten Tips for Spraying in the Wind

    Ten Tips for Spraying in the Wind

    Choosing the right time to spray can be tricky. Our gut tells us that spraying when it’s windy is wrong.  The experts tell us that spraying when it’s calm is wrong. So when can you actually spray?

    I’ve always advised my clients to spray in some wind, because it has a few advantages. The main one is that wind helps disperse the spray upward and downward, diluting the spray cloud fairly rapidly. Another advantage is that winds tend to be reasonably steady in their direction and velocity (or at least that can be forecast), so downwind areas can be identified and potential impacts are known or predictable. It helps if it’s sunny, because that improves the dispersion of the cloud even more.

    First, let’s define “windy”. The classic wind scale is the Beaufort Scale, intended for the sea, but also used on land. The upper limit for spraying is probably Force 3 or Force 4, with upper limits of 20 – 25 km/h or so.  The Beaufort Scale calls these “Gentle or Moderate Breezes” (they had to save the alarming words for hurricanes), and the scale provides good visual clues such as what wind does to flags, leaves, or dust.

    Beaufort Scale-1

    Spraying under breezy conditions can be done fairly safely if you follow specific steps. The idea is to understand what the risks are and to manage them.

    The cornerstone is to use a low-drift spray and match it to a pesticide that will work well with larger droplets. But there are other important aspects to consider. Below are the top ten to think about:

    • Choose a herbicide that can handle large droplets. Glyphosate products are well suited to coarse droplets. But glyphosate commonly has contact actives in the mix, members of Group 6, 14, and 15, and these are less likely to perform well with big droplets than those that contain Group 2 and 4 mixes. Actives with soil activity also have more tolerance for larger droplets.
    • Use a low-drift nozzle and operate it so it produces a Coarse (C) to Very Coarse (VC) spray quality, as described by the manufacturer. Dicamba labels call for Extremely Coarse (XC) to Ultra-Coarse (UC) sprays, and Enlist requires at least Coarse. To achieve these you may need to purchase new nozzles. Low-pressure air-induced nozzles operated at about 50 – 60 psi will generally be very low-drift, but lower drift models are available. If you need a finer spray, produce it either by increasing the pressure or moving to a finer tip. Do this when the weather improves, for contact modes of action.
    The name, symbol and range of droplet sizes used to describe the median droplet diameter produced by nozzles according to ASABE S572.3
    • Keep your boom low. Lowering the boom ranks as the second-most effective way to reduce drift, after coarser sprays. But there’s a limit. For low-drift sprays, you need at least 100% overlap (more for PWM), which is for the edge of one nozzle pattern to spray into the centre of the adjacent pattern. In other words, the spray pattern should be twice as wide as your nozzle spacing at target height.  For most nozzles, a boom height of close to 20 inches is enough to achieve this overlap. That’s pretty low by current standards from suspended booms on self-propelled sprayers, so being too low for a good pattern will only happen due to boom sway.
    • Maintain reasonably slow travel speeds. These reduce the amount of fine droplets that hang behind the spray boom, reduce turbulence from sprayer wheels, and they also make low booms more practical. An added bonus is less dust generation.
    • Know what’s downwind and what harms it. Survey the fields on all sides of the parcel you’re treating. When you have a choice, avoid spraying fields that have sensitive areas downwind such as water, shelterbelts, pastures, people, etc. If you can’t avoid being upwind of these areas, make sure you check and obey the buffer zone restrictions on the label. These will also give you an idea if the product can cause harm in water or on land, or both.
    • Consider a dicamba tip for special situations, even if you don’t use dicamba. If you’re in a situation where quitting and waiting is a poor option, these tips allow you to finish the job with minimal drift risk and with only slight reductions in product performance due to poor coverage.
    • Use a low-drift adjuvant. Specific products such as Interlock or Valid have been shown to reduce driftable fines (<150 microns) by between 40 – 60%, without adding significant volume in coarser droplets. The response will depend on the nozzle and the tank mix, but can be very noticeable.
    • Study drift and how it forms and moves. It’s about more than wind speed and droplet size. Knowledge in this area can help you work out the best strategies.
    • Invest in productivity. You may not need it every day, but on occasions when you have a small window to avoid bad weather, it pays dividends.
    • If you feel that drift is unavoidable and someone might be impacted by it, talk to those people first. It’s one of the most important things you can do.

    Keeping pesticide sprays on target continues to be one of our top responsibilities.

  • Fundamentals of Spray Drift

    Fundamentals of Spray Drift

    The year 1989 marked my first spray drift trial under the watchful eye of Dr. Raj Grover and John Maybank. We evaluated the performance of several spray shrouds, Flexi-Coil, AgShield, Brandt, and Rogers, and wanted to measure just how effective they were. But in my heart I wasn’t interested in drift. I wanted to study herbicide efficacy. Anyway, I thought, we’ll do this trial and I’m pretty sure we won’t have to revisit the topic.

    It’s now thirty-two years later and spray drift has interwoven itself into all my projects, remains one of the most powerful drivers of regulatory activity, is likely the most visible consequence of poor stewardship, and will stay as one of the dominant creators of public opinion around modern agricultural practice.

    Drift has not gone away. And yet our understanding of it is far from complete.

    Spray drift is defined as the wind-induced movement of the spray cloud away from the treated swath. Droplet drift can occur for all sprays, and it happens within minutes of the spray pass. Its cousin, vapour drift, is limited to active ingredients that are volatile, that is, they can evaporate from dry deposits after application. Vapour drift happens after the spray application is complete and can last several days.

    Droplet Drift

    Droplet drift can be divided into two phases that are separated by about 1 second and that are measured differently. “Initial drift” happens first and refers to the product that leaves the treated area immediately after atomization. It is airborne and can be measured by placing air-samplers (any device that can capture droplets in air) close to the downwind edge of the spray swath.

    Figure 1: Initial vs Secondary drift. Once the drift cloud leaves the treated swath, the relative strengths of turbulence and sedimentation determine the amount that remains airborne and the amount that lands downwind.

    Secondary drift describes the airborne spray cloud that continues to move downwind from the swath edge, where it either remains aloft or deposits on the surface below it. It is typically measured using samplers placed on the ground that capture sedimenting spray droplets. The difference in method is important because it goes to the heart of the problem of understanding spray drift.

    Figure 2: Droplet drift occurs when displacement energy exceeds droplet energy. The droplet’s combination of mass and velocity cannot withstand the energy presented by moving air.

    Initial drift is actually quite easy to understand because its creation is intuitive. The displacement of droplets from the spray plume is a function of balancing two types of energy. The first, droplet energy, is the product of droplet diameter and velocity. The more energy in the droplets, the more difficult they are to displace, and that’s why larger, heavier droplets or fast-moving air assist are useful drift reducing tools.  The second, displacement energy, comes from relative air movement, either from forward travel speed or wind and the associated turbulence. More wind or turbulence means more power to displace.

     Figure 3: Initial drift follows an expected response to greater wind speeds and coarser sprays. Data from a pull-type sprayer travelling 13 km/h with 60 cm boom height.

    Because initial drift is easier to understand, our most common advice for reducing drift is based on maximizing droplet energy and minimizing displacement energy. Lower booms, larger droplets, slower travel speeds, shrouds, or properly implemented air assist all help reduce initial drift. It makes sense that creating less initial drift will also reduce downwind deposition arising from secondary drift.

    Figure 4: Management of initial drift is intuitive. We reduce drift by adding energy to the droplet and by protecting the droplet from exposure to moving air.

    Downwind Deposition

    After leaving the spray swath, the moving secondary drift cloud has two main options. It can deposit or it can remain airborne. Basic physics suggest that all objects eventually fall to the ground, and since smaller objects need more time, they drift further. But when atmospheric turbulence and topography are considered, it’s not quite that simple. These two complicating factors control what proportion of the drift cloud remains airborne, and what proportion deposits.

    Drift trials show that about 20% of the initial drift amount returns to the surface within the first 100 m or so of the sprayer. The rest remains and rises in the atmosphere where it evaporates and gets mixed further.

    Figure 5: The majority of secondary drift remains airborne. Data are for Medium spray quality from a pull-type sprayer with 60 cm boom height and 13 km/h travel speed

    It happens quickly. Just 5 m downwind of the spray swath, the cloud is already 4 m tall. At 100 m downwind, we’ve measured its height to be 30 m.

    The proportion of the spray that remains airborne depends on the spray quality and the nature of the atmosphere. If it’s windy and sunny, or if the spray is finer, turbulence sends more into the air. If it’s cloudy and the wind is low, we have little atmospheric mixing. As a result, a smaller proportion will remain airborne and more will sediment, and overall, we may actually have more potential to damage downwind areas.

    When we graph spray drift deposit data from a windy day, the deposit amount decreases exponentially with downwind distance. Usually, drift damage follows the same pattern. The larger droplets that contain the majority of the dose deposit first. The smaller droplets go further and are more likely to mix in the atmosphere and rise with thermals.

    Figure 6: Deposited drift decreases logarithmically with distance. Top, linear axes. Bottom, log axes.

    Under temperature inversion conditions that are common on calm summer evenings, overnight, and early mornings, the damage from the drift cloud does not decrease the same way. The cloud containing the buoyant mist lingers over a large area. Without atmospheric mixing and its resulting dilution with time and distance, large areas can be damaged.

    The Effect of Turbulence on Deposition

    We’ve established that the more atmospheric mixing we have, the less spray will deposit on the ground, at least in the short term. How does this affect our thinking on the role of wind?

    When we evaluated drift data from a number of trials, we always saw more initial drift with higher wind speeds, as expected. However, the downwind deposit did not usually increase significantly. We attributed this observation to turbulence generated by wind which lifted more of the initial drift higher into the atmosphere. To be clear, deposited drift did not go down with higher wind. It just didn’t rise as fast as initial drift.

    Figure 7: The effect of wind speed on airborne drift (top line) vs deposited drift (bottom line) from a high clearance sprayer travelling 23 km/h and emitting a Very Coarse spray.

    The effect of turbulence can be viewed as a good thing because it protects downwind objects. Rapid dilution reduces immediate drift damage. We can use turbulence to protect objects on the ground. It’s certainly better than the alternative, emitting sprays when the atmosphere can’t dilute them, such as in an inversion. In that case, downwind areas remain at risk for a long distance, and for a long time.

    But we have to also consider what happens to airborne spray droplets. Some pesticides degrade in sunlight and stop being a problem. But others are more stable and may persist in the atmosphere for days or longer. During that time, they may move significant distances, ultimately returning to the earth’s surface in precipitation or in dust. Even though the atmosphere has diluted them, these deposits are measurable, and will show up in environmental monitoring of air, soil, and water.  We may not be able to find out where they originate, but the public knows who to blame. Agriculture.

    Vapour Drift

    Vapour drift is another issue altogether. It occurs hours and days after application, as long as the volatile product remains on a surface and conditions that allow formation of vapours persist. Vapour pressure is related to surface temperature, and losses increase with warmer surfaces. Some products enter the vapour phase when in contact with water, and release vapour after a rainfall.

    In situations where vapour is released for several days after application, it becomes impossible to control its subsequent movement. For droplet drift, if we know the wind direction at the time of spraying, we know where the impact is likely to be. But vapour movement depends on conditions that may occur between now and three days from now, and these could include high temperatures, various wind directions, and even inversions in which vapours accumulate. Ultimately, the best way to avoid off-target vapour movement is to avoid using volatile products.

    The Public Good

    Spray drift is one of agriculture’s most important stewardship challenges, and our industry needs to continue to improve its track record. Sprayers have a difficult task of converting a relatively small volume of liquid into a spray that offers good target coverage yet doesn’t move off the treated area. Favourable weather combined with droplet size management are at the heart of making this system work, but there isn’t a lot of wiggle room. Once again, an emphasis on sprayer productivity is one of the most fruitful areas to invest in, as this makes the best of the sometimes rare conditions in which spraying conditions are optimal.