How Do You Measure Plant Health With A Drone?

You may already have heard something about drones in agriculture, and how they are being used to map the health of crops and to identify diseases and pests in crops. If you haven’t heard about this yet, a quick Google search on “crop health mapping” will turn up lots of examples.

So drones take pictures from the sky, right? How do aerial photos measure crop health? Sounds mysterious. Well, it’s not rocket science, as they say, but it is science. The science of “reflectance”.

You see, when the sun shines on your plants, even the ones in your garden, the leaves of the plant soak up the sunshine and convert it to energy for growth. But you know that, hey? Otherwise you probably wouldn’t be a gardener (or perhaps you’d be trying to grow dead plants in the dark).

What you might not know as much is that the plant can only use part of the light provided by the sun. Mostly, healthy plants want to absorb visible light, the light we can see with our eyes. But sunlight also contains loads of light we can’t seen, down in the ultraviolet spectrum and up in the infrared spectrum, which the plants generally can’t use.

So what do the plants do with the light they can’t use? They put it in a postbag and mark it “return to sender”. Hmmm, no they don’t of course, but they do send it back to where it came from. The light they cannot turn into chlorophyll gets reflected back to the atmosphere. This is where the drones come in, because they can be equipped with sensors that measure how much non-visible light (most near-infrared) is being reflected back by the plants, which is a way of measuring how healthy they are.

How’s that? What they can’t use measures how healthy they are? That sounds a bit like saying what’s left on our dinner plate is a measure of how healthy we are. And in a way that’s also true. If you cut off the fatty bits of your lamb chops, and if you only eat as much as you need, you might be healthier than me. At least my doctor would say you are.

But back to the plants. It helps to understand that the process of turning sunlight into chlorophyll (how plants get energy from sunlight) involves absorbing lots of light in the blue and red wavelengths, but not so much in the green wavelength (or the infrared). Healthy plants reflect most of the green wavelength back, which is why we see them as green. In fact, the greener we perceive them, the more green light they’re reflecting, so the healthier they are. That’s the short version.

To understand the long version, we need to understand a little of the science. But only a little.

Even using a normal consumer drone camera (RGB colour – red, blue, green – making sense?) we can get some idea of how healthy a plant might be based on how much green light it reflects compared to red and blue light. There is a light index generally known as “false NDVI” which can display areas of a crop where there is more or less green light being reflected. It’s not an accurate measure, and sometimes it gives a false clue, but it has its uses.

A more scientifically valid version of “false NDVI” called VARI (Visible Atmospherically Resistant Index) offers a better way of measuring plant health from RGB images, but it still isn’t that great.

Thankfully there are much better ways that drones can measure plant health. NDVI (Normalised Difference Vegetation Index) uses RGB plus light in the near-infrared (NIR) spectrum above the red light to get a better idea of how the plants are using the light. The less NIR light reflected by the plant, in general, the better the health. Bare ground or a dead leaf will generally reflect most of the NIR light it receives, whereas a healthy leaf won’t reflect as much. This is a short version, but I’m sure you get the general principle.

The technique of measuring NIR light is by removing the red filter in an RGB camera, the filter that normally stops photos being too red. This allows more light from the red end of the spectrum into the images, and this light is what NDVI is measuring.

It’s not so much that NDVI imaging can see more plant health than RGB imaging. It’s that NDVI can see much more subtle differences in the light (what we call variability), which allows us to focus in on smaller, more subtle changes in the health of the plant and see emerging problems earlier.

There is also light above the NIR band, called “RedEdge”, which is even better at measuring crop health. Modern “multispectral” drone cameras have Red, Blue, Green, NDVI and RedEdge sensors built into them, each done tuned to a specific wavelength of light. RedEdge NDVI is much more sensitive than standard NDVI, so it can see even more subtle signals from the way the plants reflect the light.

But the real power of these multispectral sensors is in combining different amounts of the red, blue, green, near infrared and rededge light to create more complex indicators that can eliminate factors which might otherwise mask a problem. SAVI (Soil Adjusted Vegetation Index), for example, removes the very large reflectance difference between the leaves and the soil, which means the plant reflectance variability can be shown much more intensely.

But how does all this identify pests and diseases in plants? Well, pests and diseases usually manifest first in damage to the leaves or stems of the plant. This damage means the plant is less healthy, so it changes how the plant reflects the light. Being able to see this change earlier means a pest or disease can be treated while it’s still at a very early (and often very contained) stage.

Drone mapping operators like Queensland Drones can provide their clients with maps that show them precisely where the problems are occurring and, in combination with input from an agronomist, precisely how much pesticide or fungicide or other treatment should be applied and where it should be applied. We call this a “prescription map” and it can be used to automatically drive a GPS controlled tractor attachment or, more often these days, a crop spraying drone.

We also provide maps based on variability of crop health to show or guide farmers about where and how much fertiliser or other nutrients to apply to their crop, based on how the crop is performing (instead of just applying the same amount across an entire crop).

Related Articles