Exploding the myths about NDVI and crop health mapping
Queensland Drones has been providing crop scouting and crop health mapping services to farmers in South East Queensland and Northern NSW for several years, partnering with a range of grower groups, special interest groups, universities and government agencies to test different methods and approaches. In that time we have become aware that many of the current approaches to crop health mapping are just too simple to be effective.
NDVI crop health mapping has been marketed to farmers and agronomists as the holy grail for years now, but there are many flaws in the most common approaches to this technology.
The principle behind NDVI crop health mapping is not new. In fact it’s been in use since the 1960s, but primarily using low resolution satellite imagery. With the emergency of cheap unmanned aircraft it has become more timely, more cost-effective and much more details, but not necessarily any better (at least not the way most people are doing it).
As plants grow, they absorb light and use it to produce chlorophyll (photosynthesis). The cells on the underside of the leaves absorb sunlight and convert the parts of it they need to energy to power growth. Healthy leaves reflect some of that light back in the visible spectrum because they can’t use it all, and that’s why we see healthy leaves as green. These healthy leaves are also able to reflect almost all the sunlight they receive in the infrared (IR) spectrum, because they don’t need it to grow.
In NDVI crop health mapping we capture the reflected IR light (actually we collect light in the Near Infrared (NIR) part of the spectrum using a modified camera) and use that to map the health of the leaves according to how well they are reflecting the IR light.
Problem is, this is uncalibrated data. This is a problem because (a) the amount of light being reflected will change day by day and even during the day according to the light levels, time of day and time of year, (b) the amount of light reflected will vary with cloud cover, so patchy cloud during the flight may appear as poor reflectance (so poor plant health) in the final crop health map, and (c) the most method of calculating NDVI (the method used by Pix4D, DroneDeploy and many other online platforms) is flawed because it tries to measure NIR light in isolation – whereas it should actually be measured in the context of other light reflection sources.
Even worse, some growers are being encouraged by unscrupulous retailers (who I won’t name) to think they can measure plant health using unmodified drone cameras. This is a concept called “false NDVI” or “synthetic NDVI” and it works by measuring the amount of red light being reflected by the plants and basically pretending that this is IR light. It’s not.
The problem here is not just that the method is not as good as “real NDVI”, but that it can show health issues where there are none and healthy plants where there are problems. It’s basically misleading, like trying to work out the colour of a rose from a black and white photo.
NDVI is not bad science. It’s just been deployed badly in many cases by UAV operators, UAV resellers and the online platforms created to process UAV imagery automatically. Real NDVI crop health imaging requires subtraction of the visible light from the NIR light captured by the sensor, something the online platforms are not capable of calculating. Queensland Drones has always used this method to calculate NDVI for its agricultural clients.
Queensland Drones has worked extensively with organisations like Growcom and the University of Queensland to understand how NDVI imaging can be best used to benefit growers and agronomists. One proven way to realise the benefit of real NDVI crop health imaging is to overlay it with electromagnetic (EM) soil mapping data – read about one grower’s experience with this approach in this Growcom Hort360 media release and listen to the grower’s story in his own words in this Growcom Hort360 Youtube video.
If you would like to find out more about NDVI crop health imaging combined with EM soil mapping data, give us a call to discuss your specific needs.