NDVI imagery, otherwise known as Normalized Differential Vegetation Index, has become one of the most commonly discussed topics when it comes to UAV collected imagery for agricultural purposes. Plants reflect very strongly in near-infrared (NIR) and less strongly in blue green and red. Green is the the next most commonly reflected wavelength which is the reason plants usually appear green, however even the green reflectance of plants is significantly lower than the NIR reflectance.
The original NDVI method was developed by NASA in the late 1970s for use with the Landsat Earth monitoring satellite constellation and the original formula was (NIR – RED)/(NIR + RED). The reasoning behind using the Red band as the visible spectrum to use was that our atmosphere refracts blue light very strongly (hence the reason the sky appears blue), on satellite collected imagery this was a major concern and red was chosen as was the least impacted by atmospheric factors.
The amount of light reflected across all the visible bands is relatively small compared to the amount reflected in the NIR band and as such it doesn’t really matter which band is used on UAV collected imagery. Due to the low altitudes at which UAVs fly we don’t have to use the red band at all which makes it possible to convert most regular cameras so that they can capture actionable data.
Camera sensors have multiple receptors that capture blue, green and red light. However each of those sensor sites do react to light in neighboring wavelength ranges as well, the red sensor is particularly sensitive to not only red light but also near infrared light. Most cameras have what is called an IR block filter installed to filter out NIR light, but by replacing this filter with one that only allows blue and near infrared light to hit the sensor we can get a pure NIR channel on the red sensor and use the blue (or green since it also reacts to blue light) as the other side of the equation. In our experience it has actually become apparent that it is advantageous to use the green channel instead of the blue channel since the blue receptor reacts more to the NIR light let through the filter than the green channel does which results in a new “NDVI” equation:
(NIR – Green)/(NIR + Green)
Since we don’t need to normalize the data like a satellite captured image does we forego dividing by the NIR + Green term as we find that this step actually hampers the data collected using a single camera system. So in the end the result is a relatively simple formula:
NIR – Green
How does NDVI help your farm operation?
NDVI’s primary use in agriculture is identifying areas where plants may be struggling or where crop growth density is higher or lower than average. This can be difficult or impossible to see on the ground and may not be apparent in a normal image. Essentially NDVI allows for false-color imagery to be generated that highlights areas of high growth, poor growth and bare ground.
Other uses for UAV systems in agriculture
Crop damage assessment
Another major use of aerial imagery in farming applications is crop damage assessment after a hail storm. We can collect high-resolution images using our UAVs and calculate the exact percentage of your fields that have been severely damaged by hail or other weather. Using industry leading image analysis techniques and GIS technology we generate maps of your entire field and identify areas that are visibly damaged, making it easy to estimate crop damage for insurance purposes.
Dugout volume change and flood plain prediction
Another major application of photogrammetry in agriculture is the ability to calculate volume change over time for dugouts and ponds. Using direct volumetric measurements from the digital surface model created using our UAV we can tell you exactly how much water you are losing to evaporation over the summer and any low-lying areas that may become water collection zones or flood plains.