Remote sensing: A few flyovers isn’t nearly enough

It takes years of data (and boots on the ground) to really understand what’s happening in a field

More and more, farmers are using drones in their farm operations. But are they getting the most bang for their buck to assist them with cropping decisions?
Reading Time: 3 minutes

So you’ve bought yourself a drone and a spiffy new multispectral camera. Sounds fun. But what now?

It’s true that images from camera-equipped drones can offer a moment-in-time snapshot of yield or emergence. But what if you wanted to use remote sensing for something a little more robust, such as measuring nutrient variability in your soil?

Although this tech is moving ahead, there’s no quick or simple way to do that.

“Farmers are throwing drones up there and getting pictures but I think what some of them don’t understand is that data needs to be analyzed,” said Kristina Polziehn, a northern Alberta-based agronomist and remote sensing guru.

Although a strong advocate of the technology, it needs to be put in perspective, she said.

“I think the important part to understand is that remote sensing is just one tool for understanding what’s happening in your soil and on your farm,” she said.

“I think the reason why remote sensing hasn’t taken a huge uptake is because it’s not a case of ‘I have a picture, it tells me what I need to know, and easily gives me a solution.’

“Sometimes it only starts telling you a little bit about what’s going on in the field.”

The (long) process

Nutrient variability is a good example of that.

The key tool developed from remote sensing data is a variable-rate (VR) map. The first, long step in developing this map is using a satellite, manned aircraft or a drone with a multispectral camera to take near-infrared photos of your crop over several years.

You will also need multiple years of yield data.

Both the camera and yield data are comprised of georeference points (with more generally being better).

“Let’s say you’re measuring your yield data every five seconds. Each one of those geographic points has a yield associated with it called a georeference point,” said Polziehn.

“You can then layer that in — or compare it — to your drone imagery which also has georeference points. We build our yield maps out of these data points. If you have soil data from grid soil sampling or zone soil sampling that allows you to identify an exact georeference point, you know exactly what is happening at that point.”

From there, you will want to take the results to an analyst like Polziehn. She’s a professional agronomist and she said more and more of her colleagues are able to do this type of work, which involves processing camera or satellite data through a vegetation index, most likely NDVI (Normalized Difference Vegetation Index). In a nutshell, NDVI (which uses infrared and red light) defines the vegetative makeup of your field.

The analyst will also use your yield data to create a historical yield map, compare it to the NDVI results, and proceed to ‘layer’ in as much information as possible, including soil maps, water content and digital information models. From all these resources, management zones can be identified and a VR map created.

But even after all that, you’re still not done.

Even if the VR map offers a pretty good idea of yield variability, you still don’t know for sure if low-growth areas are the result of nutrient deficiency. You need to rule out other possible causes, said Polziehn, who owns and operates Axiom Agronomy in Sturgeon County.

“(Vegetative indices) give us an idea that there’s something happening in that part of the field. When it comes to the nutrient side of things, they don’t offer anything specific that will say, ‘That is a nitrogen deficiency.’ You have to eliminate everything else,” she said.

“Maybe the reason there’s poor production in one area is because there’s lodging, heavy disease or insufficient drainage. Ground-proofing can help you understand how much more data you actually need. Do you need more historical data? Do you need to do some more soil samples to understand what is causing some of the variability?”

The road ahead

In some ways, Canada is behind much of the world when it comes to streamlining remote sensing for farmers, said Polziehn, who travelled to several countries on a Nuffield scholarship to look at approaches to on-farm remote sensing.

It would help if there was more co-operation between the government and private industry in sharing soil data, she said, adding government funding would also help.

“When you look at countries like the U.K. or Australia, you have more of these private groups that have a large chunk of government funding helping them develop these projects,” she said. “But even if they’re being driven from the private sector, they’re intended as tools meant for the entire country.

“There has to be some collaboration with the federal government in order to develop farmer-friendly, useful tools.”

There are Canadian companies that are starting to innovate and there are also global players, such as Climate FieldView (a division of Bayer) that have platforms for adding yield and soil fertility data with remote sensing imagery to generate fertilization recommendations, she said.

But even remote sensing tech at its most robust will never be infallible, said Polziehn. It’s only as good as the data you put in.

“There is always the question of whether you trust a computer generating your variable-rate maps or do you want someone familiar with your fields put a little bit of a personal touch to it saying, ‘Yes, that makes sense.’”

About the author



Stories from our other publications