For some time now I’ve been watching uneasily as DJI Phantom quadcopters are increasingly marketed to farmers as agricultural survey tools. With the greatest of respect to some of the people who are selling these drones, mostly with good intentions, I feel they might be doing more harm than good.
We fly the DJI Phantom 3 Professional in our business operation and it’s a nice entry-level aerial photographic platform. Reasonably easy to use, reasonably realiable, but perhaps a bit fragile for on-farm use where it’s likely to go in the back of the ute and turn upside down now and then on landing.
The Phantom 3 out of the box is useful for taking photos of the harvester in the paddock, or making happy snaps of your contract pickers. It takes nice photos and videos if you know how to use a camera. But as it comes out of the box it’s definitely NOT an agricultural drone.
We use two Phantom 3 UAVs in our commercial agricultural crop survey business. One of these has a Canon S100 near-infrared camera mounted under it, facing down, to take high resolution images of crops. But it’s not an easy machine to operate and it takes a lot of work to extract the imagery and process it to NDVI. But if you’re interested in owning one of these, we can organise one for about $2500 including a sturdy travel case.
Our second Phantom 3 UAV has a Sentera near-infrared sensor mounted on the front. This is a LOT easier to use although the lower resolution sensor does not produce the same high quality NDVI images as the Canon S100 camera. The NDVI images are still very good and you can achieve the same kind of resolution just by flying your mapping missions at a lower altitude.
If you want to fly your own crop health drone, the Sentera DJI Phantom is a good package, sold and supported in Australia, with packages starting from just $5450.
Check out this video to see how the Sentera Phantom UAV works for you.
Using a stock DJI Phantom UAV on a farm
There are only two ways a DJI Phantom UAV can be used for crop health surveys. Either you replace the standard X3 RGB camera with a modified NIR camera, in which case it can’t take RGB photos and videos any more, or you use what’s called “false NDVI” based on RGB images.
Kyle Miller from AgEagle recently posted a great comparison of false NDVI vs real NDVI on LinkedIn and I’ll summarise what he said. If you want to see what Kyle had to say, you can read the full article on LinkedIn.
RGB images are just the normal photos we take with any camera – they capture red, blue and green light, otherwise known as natural color. These image capture what we would see with our unaided eyes.
Crop health surveys are usually carried out using Color Infrared cameras (CIR or NIR). These cameras are modified (by changing lens and filter) to capture the “near infrared” wavelength of light that you can’t see with your unaided eyes. Plants reflect this near infrared light very strongly, dependent on how healthy they are. This reflectance is much stronger than the visible reflectance we see as green colour in the leaves.
False NDVI is called false because there is no Near Infrared light captured by the camera, so the processing fakes the NIR by comparing the blue-green light to the green-red light and computing the difference between them. It looks “official” and it looks a lot like a real NDVI image, but it’s much less sensitive than a real NDVI image (as you’ll see below) … and worse than that, it’s sometimes just plain misleading.
While some areas of the False NDVI map correlate at least partly to the true NDVI image above, there is a huge difference in the sensitivity which means the False NDVI images will often miss emerging crop stress and will sometimes exaggerate the amount of stress in the crop.
Markus Weber from Landview Drones, who also flies the AgEagle, did a neat comparison of False NDVI vs Real NDVI which is captured in the image above … and this is where False NDVI can become really misleading. The red areas showing in the False NDVI map at the top, rendered from RGB images, indicates crop stress. But the real NDVI map, computed from the NIR images, tells a very different story. The so-called stress areas are actually the healthiest parts of the field!
That’s why real NDVI is so important, and it’s also why we always stress the need for ground-truthing crop health imagery.
And it’s not just Kyle or Markus or myself saying this. Leading experts in remote sensing are speaking out about False NDVI …
If you still want to know more, check out the research by NASA’s Earth Observatory.