SWAT CAM vs. Human: The Future of Plant Stand Counting

An important step in the SWAT MAPS process and step 8 in the SWAT ECOSYSTEM, is viewing the available data that actualizes plant stand counts in the field. This helps verify how well the crop is establishing in each SWAT zone, and it’s done using one of SWAT CAM’s most revolutionary features of automating the manual process of driving fields.

The two figures below show manual plant counts from two fields on the same farm. The manual counts for Field 1 averaged 5 plants / sq ft and Field 2 averaged 5.2 plants / sq ft. This is great because we achieved the target plant stand of 5 plants / sq ft for both fields!

Field 2 Manual count = 5.2 plants / sq ft

However, SWAT CAM data reported Field 1 averaging 4.5 plants / sq ft and field 2 averaging 3.1 plants / sq ft as shown in the SWAT CAM reports below. As well, Field 2 had the lowest average population on the farm. If we’re looking to improve variable rate strategies on a farm, this data suggests we should be focusing on why Field 2 population was low. Keep in mind the manual counts suggested these fields had similar population and no further investigation was necessary.

Field 1 SWAT CAM Report

Field 2 SWAT CAM Report

Notice a trend? In Field 1, good crop establishment was apparent in SWAT zones 1 to 8, but zones 9 and 10 were below target. This demonstrates how SWAT CAM data comes in useful once again by indicating high weed pressure in SWAT zones 9 and 10. Once these images are automatically uploaded to our SWAT RECORDS software, we can zoom in on specific images within the weedy areas which reveals moisture loving weeds. When we view the map in 3D, we see zones 9 and 10 are in low lying areas of the field indicating a water management issue.

{image link}

Field 1 Weed pressure map and summary by SWAT Zone

So, did our field technician do a poor job at counting plants? The simple explanation is we just can’t expect representative data for large fields counting in just a few locations. Of course, it’s not practical to manually count thousands of locations across every field of the farm either. As well, humans can be biased. We are drawn to counting plants in areas of the field that look good so information from manual counts is often skewed to the positive.

SWAT CAM counted 2,800 locations in Field 1 and 1,900 locations in Field 2 using a consistent non-biased method. It’s hard to compete with that resolution and consistency. It’s also hard to compete with the economics of field technicians spending most of their day behind a wind shield to collect low resolution data versus automated cameras getting a free ride on a sprayer across the whole farm.

Interested in how SWAT CAM autonomous plant stand counting works? Give this a read.

Interested in getting your own full farm report card on crop establishment and weed pressure by SWAT zone? We’re happy to chat and get you the tools you’re looking for. Get in touch today.

High resolution plant stand assessments with SWAT CAM

Having a consistent and even plant stand across a field is beneficial in achieving even maturity and maximizing yield. Consistent maturity is critical for fungicide staging (especially for fusarium head blight in wheat) as well as harvest timing. At Croptimistic we use SWAT MAPS to not only vary rates of fertilizer but also to vary seeding rates. This allows us to have a nice, even plant stand. Conditions in the defined SWAT zones will affect seedling mortalities, therefore we can increase seeding rates in those zones to even out plant stands (Figure 1). Typically, zones 1-2 and zones 9-10 will have the highest mortality depending on the field’s soil properties and moisture conditions. These zones will have the highest seeding rates. Zones 1-2 are drier areas of the field that are water-shedding and can often have lighter textured soil, tend to have shallower seeding, and can dry out when it is hot and windy, which increases mortality. Zones 8-10 can often be “mudded in” as these areas are wet when seeding, can have permanent soil issues such as salinity, peat, or sodicity, and these factors again, will increase seed mortality. In the end, we typically prescribe the same amount of seed as we would for a flat rate, but redistribute it to optimize the final plant stand.

Figure 1: Example of VR canola, wheat, and pea seed prescriptions from left to right. Increasing the seeding rates in Zones 1-2 and 8-10, as these zones have higher mortality.

It is important that we assess our VR seeding rate strategies to ensure that we are hitting our plant stand targets across SWAT zones. Before SWAT CAM, the process for doing this was by manual plant stand counts, which involved counting a few different areas in each zone to come up with an average plant stand count across zones and across the field. This method was very labor intensive and low resolution, since we are counting limited locations in a few fields across a farm to assess our VR seed strategy. As we know, different fields will have different seeding conditions, soil types, and therefore different mortalities. Manual plant stand counts are not able to capture all variability within every SWAT zone across a farm.

Figure 2: SWAT CAM system mounted to sprayer boom

SWAT CAM is a camera system that is mounted to a sprayer that is automatically capturing 1000s of images on every field, running those images through a machine learning algorithm, and getting plant stand counts for several crops. As of today, SWAT CAM can count individual plants in canola, corn, soybean, dry bean, faba bean, and potato. In 2024 we will be beta testing plant stand counts for cereals and lentils.

Figure 3: Example of SWAT CAM image with canola plant stand count model.

SWAT CAM allows us to accurately assess our VR seed strategies on the crops that we have plant count models for. As shown in Figure 3, we can count individual canola plants and calculate the average plants per square foot in that image; in this example the average was 5.5 plants per square foot. The seeding rate in this zone was 4.3 lbs/ac and calculated mortality was 44%, which is slightly better than the industry standard for canola seed.

Figure 4: Canola plant stand count by zone with calculated mortalities.

With 1000s of counts across the field analyzed by SWAT zone we can then get an average of the whole field and an average plant stand for each SWAT zone. This data can help us to adjust our VR strategies. For example, if SWAT zone 1 has a higher plant stand than we were expecting, and zone 1 received the highest seeding rate, then in the future we can consider decreasing the seeding rate resulting in reduced seed costs. Another example may be that certain fields on the farm tend to have lower mortality across the whole field and the plant stand is always higher on those fields. This is another opportunity to decrease the seeding rate. Alternatively, fields with higher-than-expected mortality should be investigated further. Do seeding rates need to increase or are there other factors causing the high mortality that need to be altered, such as residue management, fertilizer placement and seed-placed rates?

SWAT CAM data allows us to calculate mortalities across SWAT zones by comparing the seed rates applied to the actual plant counts calculated by SWAT CAM. Automation of this process allows us to capture optimized plant stand counts across full farms at a low cost. SWAT CAM increases our confidence in our plant counts such that we can make better decisions and adjustments to variable rate seed strategies.

Kochia control with SWAT CAM: A step-by-step guide

Kochia (Kochia scoparia) has grown to be a problematic weed in much of the semi-arid cropping areas of the US and Canada. The pressure it has on crops is most notable in saline and marginal areas, which are commonly found in SWAT zones 9 and 10. The rapid growth of this weed, its ability to tolerate high salinity, drought and heat, as well as growing herbicide resistance problems make it a massive threat to crops.

If kochia is left uncontrolled, at maturity and with wind the tumbleweed will spread seeds throughout the field. With each kochia plant producing at least 15,000 seeds, dense kochia patches with low crop competition can produce millions of seeds per square metre (Canola Council, 2023). However, with proper management, growers can reduce the soil seedbank and prevent kochia from reaching maturity. On top of that, farmers who bring precision ag tools like SWAT CAM into their operation find managing kochia to be a whole lot easier.

SWAT CAM is an autonomous imaging system designed to capture, assess, and evaluate crop establishment and weed pressure. It is a proprietary product for full farm SWAT MAP users that are clients of Croptimistic Technology or any premium SWAT MAPS service provider. Here’s how it works:

Step 1: The Capture
SWAT CAM is mounted onto sprayer booms (one SWAT CAM on each boom) which will then auto capture images of weeds and crop every 50 to70 feet each time the equipment enters the field.

Step 2: The Analysis
These photos are then processed using machine learning technology. During this step, crop and weeds are identified and various maps are generated. Crop and weed data are divided and presented separately, unlike satellite imagery where NDVI could represent both crop and weed biomass. This allows SWAT CAM to determine percent weed cover, percent crop cover, and specifically identify kochia. Once kochia has been identified in the field images, SWAT CAM can generate kochia models indicating the kochia coverage across the field.

Step 3: The Application
Since kochia is an early germinating annual weed, it requires an aggressive approach to spraying. Incorporating a previous kochia map into your early season spray jobs can get you off to a phenomenal start with your weed control.

Effective control of kochia requires use of premium herbicide products which are expensive to apply at a flat rate across a field. However, SWAT CAM can dramatically reduce the costs of Kochia control by accurately targeting specific areas that are affected, providing a much stronger return on investment than flat rate applications. A custom spray requisition can be created using the kochia model to build an on/off herbicide application, spraying only where kochia has been found. SWAT based prescriptions can also be combined with the SWAT CAM kochia models to ensure all high risk areas are covered. Through this process, there can be huge savings as a of result only applying product to the areas with kochia patches compared to applying at a flat rate across all acres. After spraying, SWAT CAM imagery can also be used to check the effectiveness of the herbicide application.

It's important to note that reliance on herbicides alone should not be the only tool used to control kochia, or any other weeds. Crop competition is the most important tool for weed control, so variable rate seed is another valuable tool to use in these problem soils that often have high mortality and allow kochia to thrive with little competition. Increasing seeding rates 20 to40% can be an effective cultural control tactic, increasing crop competitiveness that helps reduce weed escapes.

Kochia maps are a great example to start with for SWAT CAM use cases, although it has much more to offer to your farm! Connect with your SWAT MAPS agronomist or consult our sales team to find out more about SWAT CAM’s variety of applications.

Outside Source:

Canola Council. (2023, November 30). How to contain herbicide-resistant Kochia. Canola Council of Canada. https://www.canolacouncil.org/canola-watch/fundamentals/how-to-contain-herbicide-resistant-kochia/.

SWAT CAM: Croptimistic introduces the world’s first map using fully autonomous plant stand counting

SASKATOON, SK – NOV 21, 2023: SWAT CAM technology developed by Croptimistic Technology Inc. (Croptimistic) is a revolutionary crop monitoring tool for farmers and service providers. It has recently been updated to include a new plant stand count feature which automates the manual process of driving fields and performing subjective plant counts at multiple locations or flying a drone with image capture.

SWAT CAM mounts on sprayer booms and autonomously captures images every 60 feet. With no interface in the cab, there’s no action required by the farmer. The images are automatically uploaded to Croptimistic’s SWAT RECORDS software where machine learning models assess and segment crop versus weed data, creating an auto-generated map that summarizes these populations by SWAT zone. This allows farmers and agronomists to evaluate and adjust variable-rate seed/planting, fertilizer, or herbicide strategies to optimize crop establishment and develop SWAT-based prescriptions that target weed pressures across the field.

With the new plant stand count feature, the machine learning model within SWAT CAM analyzes each image and calculates individual plants per acre or per square foot, as well as the average measurements of plant spacing. The tool is available for corn, soybean, canola, and potatoes. More to come with the continuous future development.

“With SWAT CAM now having the ability to perform plant stand counts, it completely eliminates the need for human involvement in this labour-intensive process. It provides a 1000x higher resolution of counts and is completely objective and consistent.” says Derek Rude, VP of Research and Development at Croptimistic. “SWAT CAM uses constantly improving machine learning models to conduct these assessments. This is the pinnacle of observing what’s happening in the field.”

“This is a day we had marked on the calendar,” says Cory Willness, CEO of Croptimistic Technology. “We are very excited to be able to spend our time analyzing the profitability of seed applications rather than driving around for a month to capture a few observations. In an era of agriculture where greenhouse gas emissions are scrutinized and farmers need the best data possible to make decisions, this technology becomes a no-brainer. We can now collect 1000x the plant stand data with substantially lower GHG emissions and do it for the same cost as people with trucks.”

SWAT CAM pre-orders for 2024 made before December 14, 2023 will receive a 10 per cent discount off SRP of $3,950 CDN. That makes their hardware cost only $3,555 CDN per system, plus the subscription fee of $0.50/acre per year. All SWAT CAM orders must be made through a SWAT MAPS service provider and is only available to farmers who have mapped or intend to map their entire farms in the spring of 2024. To view a full list of service providers, please visit here.

About Croptimistic Technology Inc.:Croptimistic Technology Inc. is an international AgTech company providing SWAT MAPS, a turn-key variable rate process that prioritizes Soil, Water, and Topography factors of fields for the creation of management zones. Their SWAT RECORDS software powers the entire SWAT ECOSYSTEM of products, all of which are synced with the app for real-time viewing. Learn more about SWAT products and how you can leverage SWAT MAPS by visiting swatmaps.com.

-30-

For more information, contact:

Alexandra Blackwell
Marketing Coordinator
marketing@swatmaps.com

Have a kochia problem? Save thousands of dollars by using SWAT MAPS and SWAT CAM to VR residual herbicides

SWAT MAPS and SWAT CAM have proven to be useful tools for reducing weed populations not only across the field, but also in specific management zones. Kochia is an example of a weed that can be targeted using VR spray prescriptions to limit applications of expensive residual herbicides, therefore providing savings of thousands of dollars by not blanket applying the product on every acre.

Prevalence of Kochia

Kochia is a problem weed across the Canadian prairies and US great plains. It thrives in salinity and areas with less crop competition (Figure 1). Kochia establishes itself very quickly in these areas and is a prolific seed producer (15,000-25,000 seeds per plant).

Figure 1: Kochia dominating saline area of field

Herbicide resistance is a major problem in controlling kochia. Kochia resistance to Groups 2,4,9, and most recently Group 14 herbicides has been confirmed. Using residual herbicides has been an effective way to help reduce kochia populations that emerge in-crop. Residual herbicides are expensive, therefore, if we can limit our applications of these products only to the areas of the fields where kochia is growing, it will be more cost effective.

Using SWAT MAPS to VR Apply Granular Herbicide and Sulphur

Since kochia is most prevalent in saline areas of the field, we can use SWAT MAPS to delineate where these saline areas are and manage them. We can use this data and only apply residual herbicides in the areas where kochia is growing. This strategy would need to be ground-truthed to confirm that there is kochia growing in these management zones. Some of the products that we may use are Edge (ethalfluralin) prior to canola, or Authority (sulfentrazone) prior to sowing wheat or peas. In Figure 2, we are applying both variable rate (VR) Edge and sulphur with an air boom applicator that has two separate tanks, using SWAT MAPS as our map source. The average rate in this example is 1.8 lbs/ac. We are applying Edge on just 14% of the field, which works out to a cost savings of $3200 on this 124 acre field.

Figure 2: Prescription for VR Edge and VR Sulphur based off SWAT MAP

Using SWAT CAM to Develop On/Off Herbicide Spray Prescriptions

SWAT CAM is an automated camera system mounted to a sprayer boom that captures images throughout the field. These images are then run through a machine learning model and can quantify weeds and crop separately. One of the models that is available is a kochia model, which allows us to create kochia maps on a field (Figure 3). We can view individual pictures taken across a field as represented by the black dots in Figure 3. Figure 4 demonstrates a photo from one of those dots, where you can see the kochia pressure in that spot. We can then use this map to develop an on/off herbicide prescription that can be applied to the field.

Figure 3: Example of a SWAT CAM kochia map, showing kochia coverage % across the field

In the field in Figure 3, if we apply Authority in only the areas with kochia growing, we would save approximately $9500 compared to blanket applying the herbicide on every acre. This cost savings also increases the likelihood that a grower would apply a residual herbicide utilizing a multiple mode of action layering strategy to control the weed.

Figure 4: Example of SWAT CAM Image showing kochia within a canola crop

SWAT MAPS and SWAT CAM can be very useful tools in managing kochia in a cost-effective way by only targeting the areas of the field where kochia is present. Using data form SWAT MAPS and SWAT CAM, we can create variable rate or on/off prescriptions for applying many different pre-emergent herbicides for sprayers or air-boom applicators. Ground truthing is still necessary before using maps to apply VR herbicides. For more information on SWAT MAPS, SWAT CAM, or the value they can provide your operation, please find your local SWAT MAPS Service Provider or contact us.