Category: General Concepts

For basics category

  • John Deere See & Spray Select

    John Deere See & Spray Select

    On March 2, 2021, John Deere entered the optical spot spray (OSS) market with its first product, See & Spray Select™. This “Green on Brown” system identifies green material on a non-green background and is thus suited for pre-seed burnoff, chem fallow, or post harvest. It is competing for the same market space as Cropland’s WEEDit and Trimble’s WeedSeeker, but uses a slightly different approach.

    At the heart of the See & Spray system is a relatively simple RGB camera that is mounted directly to the boom and looks about 1.5 m ahead.  When this camera detects a spot of green colour, it assumes that this is a plant and activates a nozzle in line with that plant. John Deere says the weed size threshold is about ¼” (6 mm), and is evaluating its experimental data to identify exceptions to that rule of thumb.

    See & Spray Select uses an RGB camera to detect weeds (Image courtesy John Deere)

    In 2017, John Deere conducted a highly publicized acquisition of Blue River Technologies, a start up that pioneered artificial intelligence (AI) plant identification and coined the term “See & Spray”. However, the technology John Deere announced this time originated with the University of Southern Queensland near Toowoomba, Australia. The university’s Centre for Agricultural Engineering had received some initial seed financing from Sugar Research Australia, Cotton Research and Development Corporation, and Hort Innovation, and eventually partnered with John Deere. This is yet another example of the value of farmer investments in research.

    Blue River contributed to this project but remains committed to its path of developing Green on Green OSS through machine learning. John Deere says this first product is part of an evolution of spraying with ever-increasing precision that will culminate in spot spraying weeds within a canopy.

    The pixels in the See & Spray camera chip are mapped during its initial calibration, allowing the processor to know which nozzle to turn on. There are two user-selected modes. In “Single Nozzle” Mode, the system turns on as few nozzles as possible. If the weed is directly under a nozzle, just that nozzle is turned on. Should the weed be in between two nozzles, both will be turned on. In “Overlapping” Mode, a detection will turn on at least three, and up to six adjacent nozzles. This mode is intended for herbicides that contain specific nozzle recommendations on the label, such as dicamba. By fitting these tips on the spot spray location, the required overlap and subsequent coverage can be guaranteed to be compliant with that label, a unique feature of See & Spray.

    The number of nozzles activated by a weed detection depends on the location of the weed relative to the nozzles, and the mode selected by the user (Image courtesy John Deere)

    In all modes, the user can specify the distance before and after the detected plant that the nozzle will spray. This feature is useful when boom height varies or when travelling faster to provide extra assurance that the target will be covered by the spray. The boom height range for See & Spray is 26 to 47” (66 to 120 cm), and the maximum travel speed with nozzles pointed down is 12 mph.

    Installation of a 40 degree angled adaptor allows sprays tom be emitted backwards, and increases the spray speeds to 16 mph due to the extra distance and time afforded the sensors andoin processors to make a decision.

    See & Spray has a built in contingency for suboptimal conditions, for example when the boom falls outside its height range, or the nozzle speed (not tractor) exceeds the 12 or 16 mph maximum in a turn, or a light or sensor or processor fault occurs. Called “Fallback Mode”, the boom can be configured to shut off, or to go into broadcast mode (using the spot spray nozzles) at that time. These types of insurance are a necessary part of an OSS on the market today.

    To prevent fallback mode from occuring unecessarily, operators often choose to reduce their tractor speed one or two mph to allow for yaw without triggering all the nozzles.

    No OSS system is perfect. Tiny weeds, or those obscured from camera view by crop residue, may be missed. The contingency for WEEDit is “Combined Mode”, where the entire boom emits a broadcast spray at a user-determined fraction of the full dose, while still maintaining spot spray capability at the full dose when a detection occurs. The reduced dose is sufficient to control the smallest weeds, whereas the spot spray is emitted at the full label rate for the larger ones. This capability is made possible through Pulse-Width Modulation (PWM) control of each nozzle.

    John Deere has developed a mode of its ExactApply system to create the same outcome. Called “A & B Mode”, the rear nozzle (B location on the ExactApply nozzle body) is being activated by See & Spray. The front nozzle (A location) can be asked to spray simultaneously over the entire boom width. By choosing a smaller nozzle, a fraction of the label rate can be applied as a broadcast while maintaining spot spray capability. The broadcast boom is pulse-width modulated and retains the swath control and turn compensation of ExactApply. This mode also makes it easier to ensure coverage of these smaller weeds by selecting a finer, wider (110 degree) angles spray on the broadcast boom, and retaining a coarser, narrower fan angle banding nozzle for the spot application. The spot spray does not use PWM, relying on conventional speed and pressure to ensure the correct rate.

    If planning to use A & B Mode, a user would first need to decide if they will calculate the spot spray dosing on a single or a multiple activated nozzle system. If priorizing the single nozzle actiation, one would first determine the band width of that nozzle, and size the nozzle accordingly. The band width should be ar close to the nozzle spacing as possible to maximize savings. Say the sprayer has 15″ spacing, and the nozzle’s band width is 20″. Now, whenever multiple nozzles are activated, they would operate as a 15″ spacing and would over-apply 20/15 = 1.33, or 33%. Say you want to apply 15 gpa (you may need to boost the spot spray volume to allow you to cut that with the broadcast feature). You can do it with the band (and overdose when using multiple nozzles, or apply 15 gpa with the multiple nozzles, underdosing by 28% when a single nozzle is activated. Or split the difference.

    The next step is to select the application rate of the broadcast. If you want to apply 30% of the spot spray rate using the broadcast nozzles, size these accrodingly to apply 5 gpa.

    For band- and spot-sprays, the width of the spray pattern at the target height determines the dose, therefore careful selection is advised. A worksheet that shows boom heights at various fan angles and nozzle spacings is downloadable here. TeeJet and Hypro offer a selection of narrower flat fan tips, but none yet in a low-drift design. Other nozzles are in development. Agrotop has already developed a low-drift “Spot Fan”, and MagnoJet, a Brazilian ceramic nozzle supplier, has 30 and 40 degree low drift tips for sale. Wilger has develped the DX series ComboJet tips in 20, 40, and 60 degree fan angles, in a low drift (pre-orifice design that works with PWM.

    The camera sensing threshold can be adjusted to optimize a specific target. For example, to specify a certain weed size, that weed can be held in view of the sensor and the user can adjust the sensitivity until the weed is properly detected. As with any higher sensitivity, this runs the risk of more false detections, resulting in over-application. But it gives the user some knowledge that an important weed stage is being targeted properly.

    The See & Spray camera relies on ambient light conditions, and John Deere recommends it not be used within 30 minutes of dawn or dusk. Both WEEDit and WeedSeeker, in contrast, can operate under any light conditions.

    One of the challenges of running a OSS boom is the unpredictable fluctuation in flow requirement, which can theoretically range from just a few nozzles spraying to the whole boom activated in less than one second. While this extreme example is rare, a sophisticated and fast-responding pressure-based flow capability is nonetheless required. WEEDit uses a Ramsay Valve into their units to handle this challenge, whereas John Deere is relying on its existing plumbing design.

    As a factory install, the See & Spray is fully integrated into the Series 4 display and is tied into JD Link. As a result, it can generate a high resolution map that shows each spot spray activation, by nozzle. The agronomic utility of this capability is significant, as it provides a very high resolution plant density map. This capability is also inherent in WEEDit and most green on Green systems available..

    See & Spray Select is a factory option and comes integrated into the 4600 series monitor (Image courtesy John Deere)

    It’s no secret that I believe optical spot sprays represent the future of pesticide application (see here). And it’s great news to see John Deere enter the OSS area with a factory installed option. As an influential force in ag, it lends credence to the concept and will benefit all other companies vying for this space. As they say, a rising tide raises all ships.

  • Simple Statistics for Growers

    Simple Statistics for Growers

    In North America, winter is agriculture conference and lecture season. Growers are inundated with graphs and charts and left to make sense of what they’re seeing. It might be an agrichemical rep promoting a new pesticide or a seed dealer comparing yields. Maybe an equipment dealer is illustrating return on investment. Even university researchers and government extension specialists have been known to flash the occasional graph from time to time.

    Without a basic understanding of how data can be abused, we are at the mercy of those presenting the data. So, in 2012, I was asked to develop a talk that would give growers a basic grounding in descriptive statistics. More to the point, it would empower a grower to raise their hand and ask questions if they felt a presenter was talking out of their… *ahem*.

    • Did the researcher do their stats correctly?
    • Is the data clear and easy to understand?
    • Is the presenter skewing something to make us see what they want?

    Since writing it, I’ve been asked to deliver this talk to several grower groups, which is surprising because very few people love statistics. Now that we’re all webinar-savvy, I took the opportunity to update the presentation and record it.

    The video is only 15 minutes. When you’re done I hope you’ll appreciate that it’s OK to be skeptical of data. Ask questions and dig deeper.

  • Putting a Number on Pesticide Waste

    Putting a Number on Pesticide Waste

    Waste (noun): an act or instance of using or expending something to no purpose.

    In agriculture, environment and economy are intertwined. Producers strive to obtain the maximum return on their inputs. They study incremental returns and avoid applying more inputs than necessary, especially if conditions don’t warrant it.  The financial incentive is powerful, and waste is a four-letter word. This applies to seed, fertilizer, and pesticide. Pesticide labels identify the rate needed to obtain the desired result, and there is no incentive to over-apply. In fact, it’s illegal.

    But there are plenty of other places where applications incur waste. As with time efficiency, it’s a good idea to identify where this waste occurs, and the only tool needed is a sharp pencil.

    When might we incur waste in the spray application process?

    • Mixing more than we need because we don’t trust the flow meter or the tank gauge entirely, or don’t know the exact field size.
    • Priming the boom before the first swath.
    • Overlapping due to curvy terrain and coarse sectional control.
    • Spray drift away from the intended swath.

    How big are the losses?

    Let’s say we have a clean sprayer and need to spray 160 acres before moving to a new crop and product. We plan to apply 10 gallons per acre and have a 1,200 gallon tank with a 120 foot boom. That means we need 1,600 gallons of spray mix in total.

    Once we’re at the field we prime the boom. Each sprayer is different, but depending on operator experience, 30 to 50 gallons are usually needed to push product from the tank to the last nozzle. Only part of that is lost to the ground, as boom sections can be shut off as soon as product has reached every nozzle of that section. We’re assuming 0.2 gallons per foot of boom is lost.

    Spraying itself is relatively straightforward. Swath and sectional control handle the overlaps, but in less ideal terrain, double application is known to account for 4 to 5% of the area to reach non-square parts of the field. This is even more likely when the outer section is 10’ or more. Early turn-on of the boom prior to leaving the headland, to allow boom to reach operating pressure, adds to this.

    Air-activated shutoff for individual nozzles reduces section size at a reasonable cost.

    With an average nozzle, we can expect about 2% of the product to airborne drift. Most airborne won’t return to the ground within the field borders, so it’s a complete loss.

    Most of the spray that travels more than 5 m after leaving boom stays airborne and should be considered a total loss from the field.

    As we finish, the pump will draw air before the tank is empty due to sloshing or foaming, and a 50 to 60 gallon remainder may not be unusual. This simulation has assumed 5% of tank volume remains.

    We also need to purge spray from the boom at cleanout, consuming approximately 0.4 gallons per foot of boom. This occurs after the field is completely sprayed and is therefore considered waste.

    So how does this add up? The following table shows the approximate losses associated with five setups.

    Table 1: Spray mix losses during a sprayer operation. Setup 1 = baseline, Setup 2 = low application volume, Setup 3 = baseline with recirculating boom, tank level monitor, and low-drift nozzles, Setup 4 = large area between cleaning, Setup 5 = large area with recirculating boom, tank level monitor, and low-drift nozzles.

    In the first scenario, we spray just 160 acres at 10 gallons per acre. Priming the boom with 0.4 gallons per foot (allowing for all associated feed lines) consumes 48 gallons, but only wastes half of that, or 1.5% of the total volume needed for the field.

    Four percent overlap consumes another 64 gallons.

    If we have 5% of the tank volume left over, that’s 60 gallons. That amount is so small it doesn’t even register on the sight gauge but nonetheless it represents another 4% of the total sprayed amount.

    Upon cleaning the boom, we need to push the spray mix out of all the  plumbing after the pump, as it has nowhere else to go. At an assumed 0.4 gallons per foot, that’s another 48 gallons or 3%.

    If we add to that a conservative 2% drift loss, it sums to a surprising 14% of the total spray volume. For those that use lower water volumes (the second scenario), the volumetric losses are slightly less, but their proportion is higher, now accounting for 23% (!) of the total spray mix.

    In the third scenario, let’s assume we use a recirculating boom that returns the initial prime volume to the tank, eliminating any waste. We’ll also upgrade to individual nozzle sectional control, reducing overlap to 1%. And, since we want to know exactly what’s left in the tank, let’s invest in an AccuVolume system to precisely monitor tank volume. This allows us to make small rate adjustments up or down to be sure as much of the mixed product goes onto the sprayed swath as possible.

    Recirculating booms allows the spray mix to pass through entire length of boom without being sprayed, saving waste during priming and allowing waste-free boom rinses.

    When the sump begins to empty, we can introduce some water from the clean water tank to push the last of the mix to the boom (a continuous rinse system makes this easy).

    An AccuVolume sensor shows the exact volume left in the tank at any slope position and with 1 gallon resolution, allowing greater accuracy when filling and emptying.

    We’ll assume our sump waste is now reduced to 12 gallons. We still need to dispose of the content of the boom somehow, so the recirculating boom offers no saving there. But let’s also add better low-drift nozzles to reduce drift by 50% (now 1% total volume). Total loss is now just 6%.

    Low-drift nozzles such as this AirMix (Agrotop) SoftDrop reduce airborne drift by 50 to 90%.

    The last two rows in the table repeat the first and third scenarios for a larger sprayed area (1000 acres) before a tank cleaning is needed. This doesn’t change the magnitude of the volumetric loss, but reduces its proportion. Percent loss is down by a factor of two from the 160 acre interval, to 3 to 7%.

    Experienced operators might cheat the system a bit by mixing the required pesticide with some extra water to make up for the plumbing waste. Doing so prevents extra pesticide from being consumed, but it doesn’t reduce the inherent inefficiency.

    Lessons

    This exercise suggests that waste from spraying is probably higher than we assumed. If we average the scenarios, there is 10 to 15% waste. At, say, $200,000 spent on pesticide for a single spraying season, that’s $20 – $30,000 worth of product and water hauled that ends up where it doesn’t belong. Beyond the time and money, there can also be environmental consequences depending on how that waste is treated.

    There are some things that can be done.

    • Know the exact area of the field to be sprayed.
    • Study your sprayer plumbing and consider improvements such as recirculating booms and continuous cleanout.
    • Improve monitoring of tank content to allow lower remainders.
    • Consider individual nozzle shutoff to improve sectional control. These are part of Pulse Width Modulation (PWM) systems, but can also be achieved with less expensive valves.
    • Plan spray operations to minimize the amount of product changeovers.
    • Consider direct injection.

    The return on investment for plumbing improvements can be high and result in considerable future savings over the life of the sprayer. It’s worth thinking about.

    Spray Waste WorksheetDownload
  • The Most Important Developments in Spraying

    The Most Important Developments in Spraying

    Some things have improved a lot. Others have lost ground.

    Some years ago, a few of us weed scientists sat around a table and debated the most important developments in agriculture in our lifetimes. It was a great discussion, and we arrived at a few that included direct seeding (for its soil and moisture conservation as well as improved fertilizer placement), GMO crops (for slowing Group 1 and 2 herbicide resistance), and the abandonment of summer fallow in much of western Canada. Let’s apply this exercise to spray application to see what we come up with.

    What follows are my version of the most important spray technology developments in the last 50 years.

    1. Low-drift Nozzles. Spray drift is the biggest time management challenge and also perhaps the biggest public relations battle. These nozzles reduce drift, making more time available for spraying and doing it safely and effectively.
    2. Rate Controllers. I both love and hate these things. On the one hand, a rate controller matches sprayer output to travel speed. On the other, it has allowed spray pressures to go wherever they need, even beyond the optimum, to match travel speed, and that can lead to nozzle performance issues.
    3. Pulse Width Modulation. The pulsing nozzle fixes the rate controller problem mentioned above. Now, travel speed and pressure are independent. Plus, of course, a whole host of other flow management options, such as turn compensation and rate boosting, become available.
    4. Optical Spot Spraying. Once you see these in action, you can’t go back. Why would you spray a whole field when weeds only cover 10% of it? Products like WEEDit and WeedSeeker are proven green-on-brown performers after years of field success around the world.
    5. GPS Guidance. Some of us grew up with foam or disk markers, others learned to aim for brave family members perched on headlands. Achieving accuracy was stressful, overlap was insurance, and misses were common. The importance of this development is probably under-estimated.
    6. Sectional Control. The ability to adjust the spray width in individual nozzle steps makes sense, and this can come with or without PWM. In fact, that alone can save 5% of an annual chemical bill compared to conventional sections measuring about 10 to 15 feet. And it’s definitely better than the left boom or right boom options from the 70 and 80s.
    7. Operator Comfort and Safety. The refuge of the cab makes longer days bearable for all equipment, but for spraying it dramatically improves safety as well.

    But we’re far from done. We still need work in these areas.

    1. Cleaning and Waste Management. I can’t imagine another industry where managing potentially hazardous leftover materials are left to the discretion and circumstances of the applicator. Let’s make it easy and fast to thoroughly clean the sprayer and safely dispose of leftovers. Step 1 is smarter and simpler plumbing.
    2. Boom Stability. Booms are too high, resulting in more drift and poorer nozzle performance, and adding to operator stress. The sole reason is unsatisfactory levelling. It’s possible to solve this, but it seems to not be a priority.
    3. Weight. The road to productivity seems to be paved with larger, heavier machines. The side effects are fuel consumption, compaction, getting stuck. Let’s get smarter with frame design and logistics and talk acres/h rather than tank capacity and power.
    4. Cost. All farm equipment has seen cost increases that far outstrip inflation or any reasonable accounting of productivity and features. Sprayers lead the way. Yes, it’s possible to spins this as a value proposition. But it shouldn’t be necessary.
    5. Drift Management. Sprayer design continues to ignore drift management. We need sprayers that produce less drift by design, and this requires consideration of tractor unit, wheel, and boom aerodynamics. It’s more than a droplet size issue.
    6. Direct Injection. Although very handy for single product application, the plethora of product formulations and mixes has limited the success of direct injection systems. The complexity of injecting at the nozzle, and the resulting lack of available systems, has stymied some very attractive options, such as site-specific rate or product use.
    7. Ergonomics. If you need training, or to call someone before using your new sprayer for the first time, something’s wrong. Interfaces need to be intuitive and simple. The golden age of spray monitors was the 1980s. Those featured a main power toggle switch, a pump power switch, boom section switches, an agitation switch, and a simple way to enter the important information which was basically desired application volume. The screen can still be pretty, and you can still paint and monitor or tweak all the functions if you like that. But let’s at least have different tiers so beginners can also use the machine. Make interfaces using the philosophy Steve Jobs instilled in his trusted designer Jony Ive with the first iPod: no more than three clicks to achieve any desired outcome.

    A few areas show promise and may suit certain niches.

    1. In-Crop Weed Sensing. The green-on-green sensing that has been made possible by machine learning has shown some encouraging early success. Continuing improvements will eventually bring its reliability to within commercially acceptable standards. There is significant activity below the radar in this area, as all players recognize the enormous upside of a breakthrough.
    2. Autonomy. While dispensing a pesticide adjacent to sensitive areas isn’t exactly the low-hanging fruit of autonomy, such field sprayers will have a fit in the temperate plains of North and South America, Australia, and Asia and may help solve the cost and weight problem.
    3. Drone Application. The rapid pace of advancement in remotely piloted aerial systems, along with a seemingly low barrier to entry of new companies, will put pressure on the industry to make a decision on this alternate application method. If it can be done safely, it will have a dramatic impact.

    If you want to improve your sprayer, don’t ignore the small things you can do in your operation. Although we’re conditioned to look for game-changing technology, the most sustained improvements don’t come from a single innovation, but from a period of persistent evolution. A lot of small improvements add up. Spray application is no different.

  • Categorizing air-assist sprayers by air-handling design

    Categorizing air-assist sprayers by air-handling design

    Air handling systems

    Air handling systems can be specialists or generalists; some are designed to do one thing very well while others are more adaptable but not as precise. Fan type plays a big role in determining a sprayer’s abilities. Their native characteristics make them better suited to certain scenarios.

    This may seem contradictory, but we are not saying that the fan alone defines or limits the entire sprayer. Fans operate within a larger, engineered air handling system. Also, the operator has control over how that sprayer is configured and used. This means it is equally important to consider how the air exits the sprayer – not just the fan type that generated it.

    Fan types

    • Radial fans: Radial fans produce high volumes of moderately turbulent air, and relatively low static pressures. They are often associated with fixed vanes and straighteners inside the fan housing to reduce initial turbulence.
    • Turbines: Turbines may look like radial fans but they’re designed to spin faster and they have blades designed to compress air. They are used in sprayers that have ducts, towers, cannons, or other more complex volutes.
    • Straight-through axial fans: These fans produce high volumes of the most turbulent air. With their comparatively short throw and wide air wash, they should be positioned close to the target.
    • Tangential (aka Cross-flow) fans: Tangentials produce the most laminar air, forming a very high volume, low velocity jet sometimes called a “curtain” or “knife”. They have a comparatively long throw and rely on the canopy to induce turbulence.
    • Centrifugal (aka Squirrel cage) fans: Centrifugal fans have a side-discharge arrangement that turns air 90 degrees. They can produce high pressures and are nearly always paired with an air-shaping volute.

    We are proposing defining air-assist sprayers for perennial crops according to their air handling systems. Ultimately, the defining characteristic of each design is the net vector of the air they generate. We have provided silhouettes for clarity, but these generic designs are not intended to imply a manufacturer.

    Low profile radial

    The oldest and perhaps most recognizable air handling design, the Low Profile Radial (LPR) sprayer generates air in a radial pattern from one or more axial fans or a volute connected to some other fan style. This is the classic airblast sprayer.

    Defining characteristics

    • Wide range of adjustable air energies from virtually zero to high.
    • Minor adjustability of air vectors via deflectors and moveable outlets.
    • Net air movement is lateral and upward.

    Cannon

    The Cannon (CN) sprayer generates and channels air through a single volute and delivers the spray as a compact, point-source jet. 

    Defining characteristics

    • High air energy characterized by high velocity and low volume.
    • Extensive adjustability of air vector via a vertical duct with positional outlet and deflector(s).
    • Usually a single-sided sprayer used to spray over and through multiple rows.

    Fixed tower

    The Fixed Tower (FT) sprayer generates air from one or more axial fans, multiple straight-through radial or tangential fans. It may employ flexible tubes, tapered bags or solid ducts to redirect air laterally from a fixed central tower. It may feature additional flexible ducts or adjustable deflectors at the top of the tower to spray over and beyond the adjacent rows. 

    Defining characteristics

    • Wide range of adjustable air energies from virtually zero to high.
    • Minor adjustability of air vectors via deflectors and moveable outlets.
    • Net air movement is lateral compared to LPR sprayers.

    Targeting tower

    Similar to the FT, the Targeting Tower (TT) sprayer can focus air vectors with a wider range of adjustability, shaping the lateral air output more precisely to the canopy. TT generates air from one or more radial fans or multiple tangential or straight-through axial fans. It may employ flexible tubes or solid ducts to redirect air generally laterally. 

    Defining characteristics

    • Medium to high air energy.
    • Moderate to high adjustability of air vectors. Airflow can be subdivided into individually-adjustable sections.
    • When the tower exceeds canopy height, net air movement is lateral to slightly downward.

    Wrap-around

    The Wrap-Around (WA) sprayer surrounds the target rows with air sources. This creates multiple converging and/or opposing airflows within the row. 

    Defining characteristics

    • Straight-through axial fan systems are either electric or hydraulic with a wide range of air energies.
    • Low to high adjustability of air vector via deflectors, moveable air outlets, or fan position adjustments. May also have an adjustable frame.
    • Net air movement is ideally neutral to slightly downward.

    Summary

    In adopting this system of classification, we believe the process of optimizing sprayer configuration and calibration can be made less complicated. A universal language facilitates clear communication between growers, industry and consultants/specialists.

    We acknowledge that there may be rare sprayers that don’t fit these categories. There are commercial examples of air-assist sprayers that combine features from these air-handling designs (e.g. hybrids of LPR and FT designs)… but let’s keep things simple.