Never fail, as spring turns to summer we get questions about algal growth in water tanks. There are lots of suggested solutions, but questions about pH antagonism and phytotoxicity seem to linger. In 2021/22 we ran trials to explore how well home-grown algicides like copper, bleach, and ammonia work, and whether they cause antagonistic responses when that carrier water is used on crops. You can see the results here.
Anticipating those results, we wrote this parody on a Stone’s classic. Not long after it was brought to life by the brilliant minds behind Michigan’s epic podcast “The Vegetable Beet” (Go subscribe right now!). You can hear it in the link below, where Ben Phillips is accompanied by his daughter’s toy tambourine. Crank it up to 11!
I see a white tank and I want it painted black No algae anymore, I want my water back I see the cart drive up while spraying summer rows I have to dump it out until the algae goes
I see my neighbours’ tanks and they’re all painted black With copper sulfate they claim algae won’t come back I see them turn their heads and quickly look away They see my algae grow in sunlight every day
I look inside the tank to see if it is black I could park it in the shed, or in the shade out back Maybe then it’ll fade away and I can face the facts It’s not easy filling up when your filters plug with crap
I wish that my green sea would turn a deeper blue If you try chlorine pucks it will clear up for you
I’ll store my tanks away from that bright summer sun Then I’ll spray algae-free before the mornin’ comes
I see a clean tank ‘cause I had it painted black No algae anymore, I got my water back
I see the cart drive up while spraying summer rows Clear water coming out, and no more clogging woes
I want to see your tank painted black! Black as night! Black as coal! I wanna see the bugs, knocked right out of the sky I wanna see it painted, painted, painted black, yea!
Some years ago, a friend recommended that I read The Tipping Point by Malcolm Gladwell. In this book, Gladwell tries to understand why some things catch on, and others don’t. It’s a compelling read full of Gladwell’s trademark stories and his knack to deftly interpret scientific studies. He talks of connectors, mavens, and salesmen, as well as the “stickiness factor”, a measure of how memorable something is, as keys to success of products and ideas. I think of the book often as I ponder the many good ideas in agriculture, many of which never see widespread adoption.
One of these good ideas is spot spraying. Green-on-brown detection was first introduced in the early 1990s. Anyone remember the Concord DetectSpray? It was great but had bad timing, as resistance wasn’t a big issue and glyphosate prices were about to slide. Green-on-brown grew to the NTech (later Trimble) WeedSeeker a few years later. Rometron’s WEEDit built on Trimble’s success and found widespread adoption in Australia in the past ten years. Spot spraying did not gain any traction in North America during this time.
Australia is unique in many ways, not the least of which is their summer spraying practice. Summer is the hot, dry season where land is typically fallow and weeds are kept in check with herbicide sprays (aaaah, the serenity). Making several passes over a field, combined with the need to control some larger and hardy plants, is expensive, and a spot spray saves much of the cost. The savings can be put to use with more effective herbicide tank mixes that delay the onset of herbicide resistance. Spot sprays pay for themselves in short order Down Under.
It’s more of a challenge in the northern plains of North America, where the fallow season involves snow cover and burnoff occurs in a short window before seeding and sometimes after harvest. But nonetheless, spot sprays have a fit for many of the same reasons.
WEEDit is the first system to make serious inroads in North America, with several dozen systems having been retrofitted to high-clearance sprayers. High detection accuracy and hardware reliability is proven in three seasons.
On March 2, 2021, John Deere entered the Green-on-brown spot spray area with See & Spray Select. This not to be mistaken as competition. Instead, the entry of a major brand provides validation of the concept like only a large manufacturer can. Yes, we’ve reached a tipping point.
While the first Green-on-brown units are becoming established, Green-on-green, the ability to detect weeds within a crop, continues to be developed around the world. French startup Bilberry has made enough gains in Australia to bring its product to market with Agrifac, where it’s called AIC Plus. In farmer field trials, they have achieved 90 per cent detection accuracy of wild radish in Western Australia, and claim that they are ready for broadleaf weed identification in wheat, barley and oats. Bilberry’s technology will also be seen on Australia’s Goldacres and France’s Berthoud. Other startups, notably Israel’s Greeneye Technology, plan to introduce a Green-on-green system in the U.S. in the near future. Amazone, the German farm equipment giant, partnering with Xarvio and Bosch, announced plans at Agritechnica to have a commercial unit for sale by 2021.
This technology will have significant impact on sprayer design philosophy. At present, productivity is synonymous with capacity, and large tanks with commensurate heavy and powerful tractor units dominate. Spot spraying savings will depend on weed density and hardware resolution, but 50 per cent to 90 per cent reductions in spray volume can be expected. A 1,600-gallon tank would no longer be necessary. The savings in frame weight and horsepower would be significant, as would the time savings from less intense tendering demands. These savings would offset the lower driving speeds that accompany sensing technologies, and, overall, provide a lower bar for autonomous operation. We may see lighter specialty spot sprayers.
The savings in brute size will be countered by increased sophistication. Better boom height management is essential for spot spraying, not just for the sensor to properly see the target and estimate the time needed for the boom to reach that spot, but also for the spot spray itself to deliver the right dose. In any fan spray, band width at ground level changes with height, and that, of course, is related to dose. Trailed booms can address this issue easily.
But not everyone wants a specialty spot sprayer that would require an extra pass over the field. With growing utility of soil residual herbicides, dual tank sprayers—small tank for the spot spray, large tank for the broadcast residual—make sense. Large sprayer frames can accommodate an additional smaller tank, second pump, and plumbed boom easily.
Plant detection and identification bring other opportunities. Adjusting dose for plant size is one of the first, or for harder to control weed species.
Spot sprays rely on fast, precise response of the nozzle, and this provided by fast-reacting solenoids that are part of pulse-width modulation (PWM) systems. On a broadcast sprayer, these solenoids can change the emitted dose instantly, within a certain envelope, by altering the duty cycle of the pulse. This, however, works best in the context of a boom with overlapping spray patterns. A single band spray would not change dose with duty cycle as easily.
Higher dosing would be an opportunity for multiple nozzle bodies that are able to spray one, two or more nozzles in the same spot simultaneously. These are already widely available and popular in Europe.
This also brings direct injection into play. Current systems introduce the active ingredient into the boom upstream of the nozzles, affording it time to mix into the water. For true spot spray utility, though, direct injection ought to be at the nozzle. Only then can custom mixes and rates be applied on a spot basis. It’s been done before, if only to show how difficult it would be to deliver uniform doses to a spot spray machine.
Spot spray sensors have agronomic benefits. By recording the location sprayed, weed patches can be mapped. As plant identification becomes possible, it’s conceivable to obtain plant species and stage distribution maps from the spray pass That would turn the sprayer into a high-resolution crop scouting tool. As machine learning and sensor sophistication grows, other plant and soil parameters can be mapped. The agronomic value of such maps, especially if created over the course of the growing season, is immense. Of course, data density, handling, storage, and analysis will constrain this.
If the past has taught us anything, it’s that there seems to be a appetite for investment in farm equipment. Sprayers have been the most-used implement on the farm for some time, and their popularity continues despite sharp price increases. These new capabilities will only add value to these implements. Prepare for sticker shock, followed by acceptance and adoption.
What will a future spot sprayer look like? Although it will have tanks and booms, the level of electronic sophistication will make it so much more versatile we can’t yet imagine all the ways in which it might be used. But it seems to me the situation has tipped and we’re already accelerating toward that future.
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.
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.
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.
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.
