The imagery provided by UAVs (aka RPAS or drones) has many benefits to farmers. While there is much focus on NDVI at present, there are many more applications enabled by timely, high resolution imagery and 3D modelling.

Aerial imagery has been used by some agronomists for decades. Until recently, virtually all of the imagery obtained came from satellite and manned aircraft. While there are advantages and disadvantages to each platform for imagery acquisition, UAVs are unique in that they tend to have the highest resolution, can be dispatched quickly, are affordable, and allow for accurate 3D modelling of crops, topography, and other features (although, 3D modelling is more a function of the image processing than the UAV platform itself).

In the hands of a properly trained agronomist, aerial imagery may be used to:

  •        Enable Variable Rate in-season fertilizer, particularly Nitrogen
  •        Enable Variable Rate fall fertilizers, such as Potash and Phosphorus
  •        Enable Variable Rate Lime and amendment applications
  •        Enable Variable Rate planting
  •        Guide scouting and in-season tissue sampling for spot-treatments of fertilizers and pesticides
  •        Guide late season yield estimates for more accurate assessments of yield
  •        Delineate and quantify areas of crop injury in order to document the impact of specific events
  •        Create topography maps, watershed analysis, delineate depressions and hilltops, and plan for drain tile installation
  •        Measure stockpiles of: limestone, manure, silage, etc.
  •        3D crop modelling and mapping the height of the crop as it varies across a field, which is an entirely new way for agronomists to visualize crop health

Note: Many of the maps discussed are coloured using a rainbow colour scheme. For the purpose of this article, the absolute values are not relevant and are left out for aesthetic reasons. For all these maps, red represents low values, while purple is the highest values. For management zone maps, red is the lowest soil organic matter and purple is the highest.

RGB imagery and NDVI

Conventional photography uses red, green, and blue bands of light. NDVI requires a near-infrared band in addition to red or green. While there is much emphasis in the media on the value of NDVI, there is a lot of information present in conventional (RGB) photography that may provide value to growers and agronomists alike.

NDVI is probably the most often discussed with regards to UAVs and precision Ag applications and it is true that there are many crop health applications for this layer. NDVI may in fact be the best layer for variable rate nitrogen applications, for example. However, NDVI alone is often ill-suited for answering important agronomic questions and may lead to misapplication of other inputs. Consider the example below:

The two areas above are classified the same (red) by NDVI analysis, only when we overlay topography, do we see why each is stressed. NDVI has classified completely different micro-environments as being similar and may have resulted in misapplication of inputs based on this layer. For example, turning this layer into a variable rate application map for planting corn next season would result in the hilltop receiving the right rate of seed, but the depression would be dramatically under-seeded; yield maps will present similar issues for this purpose.

In addition to NDVI, RGB imagery may be used to document crop injury events and, if appropriate methods are used, are nearly as effective as NDVI at identifying dramatic crop health issues. NDVI, however, is especially effective at identifying areas of crop stress that may be more subtle.

True NDVI bottom left, false NDVI (not using NIR) bottom right. The same general patterns are established with both methods, but the distribution of intensity is markedly different on the best and worst areas.

RGB photography may also be used to do a stand-gap analysis for assessing areas that may require re-plant.

Counting individual plants with an RGB analysis. Converting this raster (left) to a contour shapefile and selecting the polygons with the signature of the plants provides a means of rapidly counting plants within a known area. 

Drown-out of new seeding alfalfa. This crop suffered from several inches of rain shortly after planting, which left most of the field with a poor stand. The variability of the field made it difficult to assess the extent of the damage from the ground. Using RGB imagery and analysis, we were able to delineate and measure the area that needed to be replanted.

Under certain conditions, along with appropriate methods, RGB imagery may also be used to identify, count, and measure weed infestations. Many cameras modified for NDVI lack a blue band of light, which was necessary for the analysis shown below.

1,008 clumps of volunteer corn per acre, which accounted for 4% of the field area. Areas with volunteer corn averaged 4-6 bushels less than areas without this weed. The cost of treatment would have been approximately half a bushel of beans (at that time). 

Relative Biomass Modelling and 3D Layers

The 3D layers produced by the photogrammetry process that turns individual pictures into full-field orthomosaics may also be used to map out the height of the crop. This enables many agronomists to visualize the condition of the crop in an entirely new dimension!

Subtracting the Crop Surface map from the Topography layer leaves just the height of the crop left. 

Ground-truthed relative biomass map (crop height, not NDVI). Plants were pulled from the same row, same NRCS soil type, but different biomass zones. The wet-weight of the blue zone was twice that of the red zone.

Relative biomass mapping complements layers such as NDVI and helps to explain what is actually happening in the field.

NDVI has classified the pivot overlap, which received too much water and has increased disease pressure, the same as the dryland corners (mostly green).  

Overlaying the relative biomass map with NDVI shows that there are marked differences in plant height, even though reflectance-based NDVI classified them as being similar in plant health. The ‘potholes’ in this map are areas of crop lodging, which was especially severe at the pivot overlap.

Enabling Variable Rate Technology (VRT)

Management zones allow growers to variable rate apply inputs such as seed, fertilizer, lime, amendments, and more. Some zones in a field may need no lime, while others may need two or three tons per acre, for example. Management zones are an alternative to grid sampling, which has been the default method of enabling variable rate applications. Recently, many precision agronomists have realized that grid sampling on conventional 2.5 acre grids tends to leave a lot of spatial variability unaccounted for.

There are numerous methods for generating management zones. Many layers have value depending on what needs to be managed. Example layers for management zone creation may include NDVI maps, yield maps, electrical conductivity (EC) maps, topography, and soil survey maps, to name a few.

There are many potential layers that may be considered for use in zone creation, depending on what needs to be managed. Some layers are indispensable, others will add a little value, and some may even lead to less precise applications of inputs.

Many agronomists look to define management zones by historic yield averages, topography, and soil survey maps. These can be helpful layers to identify spatial patterns, but are often ill-suited in for zone creation where we wish to enable VRT applications of seed, fertilizer, lime and amendments.

Management Zones from aerial imagery are represented by different colours. The solid black lines delineate soil types as defined by the soil survey. It is clear that there is spatial variability that is left out of the soil survey.

Management zones created by well-timed, high resolution aerial imagery allow for spatial delineation and classification of areas by plant available water, organic matter, and exchange capacity.


There is immense potential for aerial imagery to aid farmers in managing their inputs more efficiently and effectively. Using site-specific treatments of fertilizer, lime, amendments and even pesticides means that the environmental impacts of these treatments can be reduced when compared to conventional application methods. Furthermore, site specific treatments are good for the farmer’s bottom-line; they may save farmers money by reducing the volume of inputs required to grow a crop and/or increase the growers bottom-line by putting more inputs into the areas that need them and increasing yields as a result.


Quality imagery from UAVs is heavily sensor/camera dependent. It is highly recommended that the camera be equipped with an Incident Light Sensor (ILS) sometimes referred to as a sunshine sensor. This will make images acquired in varying levels of light, suitable for reflectance-based modelling of soils and vegetation. Without an ILS, uneven lighting can skew NDVI results and lead to inaccurate spatial delineation of crop health, for example.