The market for drones, sensors, and agronomic software that quickly turns images into actionable intelligence is racing forward at a breakneck speed.
It’s difficult to know which systems will save agronomists or producers time and provide a return on investment, so Glacier FarmMedia had Markus Weber provide a presentation during Farm Forum.
Webber heads up LandView, a company that sells drones with agronomy packages, and he’s helped pioneer the use of drones in western Canadian agriculture.
He said agricultural technology is too focused on producing better maps, including normalized differential vegetation index or normalized difference red edge index, and others.
“Yes there is some value there. Much of this you can do by satellite, but drones can do a better resolution,” Webber said during the FFE webinar.
“But where the real value is in a drone is being able to see individual plants, parts of plants and really getting an up-close view.”
He said software is being developed that will help drone and satellite systems work together more efficiently, though using the strength of each system.
The resolution satellites offer is constantly improving, but today they typically offer five to 10 meter resolution, while drone systems commonly used in agriculture for high resolution mapping are usually in the one to 10 centimetre resolution range and they can be quickly deployed to any part of any field.
Webber said a company out of the United Kingdom called Skippy Scout is launching a service this spring that uses data from the Sentinel 2 satellite constellation to build NDVI maps and then locates points of interest.
“So rather than flying randomly around or mapping the entire thing and waiting 30 minutes for that map to be completed, you get sight inspection from very low altitude,” Webber said.
“You can do it manually. But this is done autonomously, the capture of high resolution imagery, and it will return to the same place throughout the season if you want to recheck the same benchmark.”
He said Skippy Scout can identify crop, weeds, and bare soil, and can calculate the green area index, which is useful to identify early crop establishment.
The program does this in two ways. The first is to use about a dozen points in a field to create a heat map of relative emergence that is useful early in the year.
The other way is to use a satellite map to create zones, and then use the drone to perform benchmark samples within the zones.
“If you’ve established zones by some other way, whether that’s by imagery or soil, you can sample the amount of vegetative cover for your variable rate application of fertilizer,” Weber said.
He said drones can help agronomists cover more acres by getting an aerial view of problems not obvious from the ground, save them time because it’s much faster than walking, provide information from tough areas to access in the field, and help quantify it and geolocate issues.
“It lets you measure the spatial extent, or to figure out what part of the field is different from another before you’re taking samples,” Webber said.
To demonstrate how drones can fit an agronomist’s workflow, Weber included a video in his presentation of him deploying a small Mavic 2 with a scouting package.
“It has everything you need for your first season,” Weber said.
“It’s a very small drone, lots of batteries, charging systems, and cases and landing pads to keep your investment safe.”
He said this small drone saves a lot of time because it’s easy to transport, deploy, and it can fly up to 60 km-h, so it can quickly provide images from any point on a field within a matter of seconds.
Operators can control the camera independently from the direction of travel, and take images from either taking the camera right down to the ground or by using the 4X zoom on the camera.
“You can identify individual plants and get a good look at what’s happening at that location. And once you get good at this you just move it around from one location to the next,” Weber said.
He said there will be instances where agronomists and farmers will wish they had a NDVI or NDRE multispectral camera, later on in the season to perform mapping.
“For most people, unless they are going to be doing thousands of acres a season. It’s probably best to farm that out to someone who has this type of equipment,” Webber said.
He displayed a DJI Matrice 200 series drone, which still has the largest variety of cameras you can attach to it even though Matrice 300 series was released this year.
The Matrice 200 series is a large drone that flies with two batteries on the back and has a FPV camera on the front that gives the pilot their position.
It also has two gimbals attachments called sky ports where a multitude of different cameras can be attached.
Webber said there are a few good options of visible spectrum cameras available including the Zenmuse H20 that has a 20-megapixel camera and is used for mapping.
“During flight, it will tilt straight down and just take picture after picture, and that will give you visible spectrum light, that is great for mapping and surveying,” he said.
“The Zenmuse 5S is more of a camera you would use for video, or high resolution pictures for marketing purposes.”
He said if users want to use visible spectrum to get a close looks at things, a Zenmuse 30 works for them.
Its sensor is only two megapixels, but it has a 30 times optical zoom, which is useful for weed inspection.
“When you have this type of optics you can hover with the drone and take a quick panel along the quarter line. It’s a very efficient way to get a quick view on something, without then also having the time to fly to that quarter line because you can simply zoom in,” Weber said.
He said plants reflect differently depending on if they are healthy or unhealthy and also on their phenotype.
For instance, a thistle looks different from a barley plant based on its reflectance, not just on its shape or features.
To capture these differences, multispectral cameras have been developed to let in near infrared light, and to capture light in very narrow and very distinct bands so that algorithms can be used to create different indices.
“NDVI is the one you’ve probably heard of the most and it’s probably the one that you would want early in the season,” Webber said.
“If you as an agronomist have some issue you need to document, whether that’s spray drift, hail, anything you want to document early in the season NDVI is the go-to index.”
He said there are many indexes and that there will soon be specific algorithms built for specific issues.
For instance, he’s taking part in a project that is developing a vegetative index specifically for identifying Canada thistle in cereals.
“You can do things like quad-band vegetation stress, yield potential, vegetative fraction, multiple NDVI variants, chlorophyll index. All of these are done with just doing the math differently on those four or five spectral channels,” Weber said.
“You can do some very unique work flows like plant population. And then because you are finding the plants you can also identify weeds and do weed densities, emergence fractions, even plant sizing for some crops.”
MicaSense is the leader in this field, especially for researchers, he said.
“Rather than just one sensor that’s filtered, it has multiple sensors. In this case blue, green, red, red edge, and near infrared are being collected by those five sensors. It’s mounted on the bottom of the drone and takes pictures straight down. The case with all of these sensors.”
One of the sensor companies that Webber likes best for infield agronomy or for on-farm use is SlantRange, primarily because of its software.