Look around a bit, and you will find lots of information and ideas about when and how to use imagery for row crop management on the Internet. Whole books have been written about this subject, and lots of papers have been published. And, every imagery company has information available. Unfortunately, much of the Internet information is misleading or simply wrong.
In this article, we’ll go back to the drawing board and try to unravel some of the complexity by looking at how imagery has gotten to this point, where we are today, and what areas imagery will most likely fit in row crop management systems.
A Bit of History
Imagery in agriculture can be traced back to the Depression Era and the use of aerial photography in general. The earliest kinds of aerial photography were black-and-white photos taken of most agricultural areas from an aircraft using a mapping camera with panchromatic film. This kind of imagery began in the 1930s with the Soil Conservation Service. Pan imagery was useful to photo interpreters who looked for patterns, tones, shapes, sizes, textures, shadows, and other features in the pictures in association with other mapped information such as terrain maps. In fact, stereo photos were processed to produce such elevation and terrain maps.
These pan photos were most useful when the soils were exposed (no or little vegetation cover) and when the field was relatively dry at the surface. Based on what photo interpreters saw, lines were drawn to identify homogenous areas called soil management zones. With such soil zone maps, farmers then would decide what inputs should be applied to each zone.
In the 1940s, aerial color photography became available. These photos were really just an extension of what you see with your own eyes … but from a bird’s perspective. Then, features such as soil color became something that helped define the soil management zones better as they might change over time or from year to year.
If vegetation had emerged in a field of interest, then plant colors would show where the plants were healthy and where stress was occurring. Stressed plants might be more yellow than the normal green color. One might even see areas that are diseased.
Usually, all of these pan or color aerial photos came with map coordinates or clear references to features such as row numbers, so that you could walk out to a place where something unusual was seen in the aerial photo.
After World War II, a special kind of aerial photography was available to replace or supplement ordinary color photography. This was called false-color camouflage photography. Today, this kind of imagery is called color infrared imagery. The term “false” was used to indicate that the colors that you see in a color infrared photo are not the same as the “true” colors that you see in a natural color photo.
Starting with Landsat 1 in 1972, satellite-based imagery became available as digital images that have data that can be processed by software on a computer. In the beginning, these images were crude, featuring just four bands (green light, red light, red edge, and near infrared) with large pixels about 260 feet across for each pixel. And, the precision of the brightness scale was very limited — at most, 64 shades of gray. Nevertheless, this lead to a number of spectral indicator (index) formulas being used to combine digital imagery in two or more spectral bands.
One of the most popular and often abused kinds of indicator maps is NDVI. NDVI stands for Normalized Difference Vegetation Index. NDVI values are made from only two bands … red light and near infrared. If done correctly, an NDVI map will have numbers from -1 to +1. However, for soils and vegetation, the NDVI numbers are all positive … say from 0.1 to 0.9 or up to 1.0. These numbers then are represented on a map by colorization, using a rainbow of colors. But they really are just numbers. The colors in NDVI imagery can lead you to believe that something is wrong (a red color for low NDVI values) when there is nothing wrong at all.
Now there are several kinds of NDVI-like spectral index formulas and maps that you can get from your image vendor. Some are better at handing the issues associated with real NDVI maps, such as soil noise or the lack of sensitivity at the top of the NDVI scale. Some prefer to use green light instead of red light for making an NDVI map called a “Green NDVI” map.
Some now like to use red edge instead of red light to make an NDRE map (NDVI based on red edge and near infrared). The NDRE maps are often called a chlorophyll map, but this is really just another kind of overall vigor and plant density map.
In the early days, Landsat 1 was the only global operational digital four-band camera, and you could only get an updated image every 18 days — even less often if it was cloudy on that revisit day. Later, more Landsat satellites were put into operation so the revisit interval got better — today, that’s perhaps every eight days. Landsat resolution also improved.
Then, a whole host of commercial (not free like Landsat) Earth observation satellites came into operation. This created much better revisit opportunities. Also, the number of bands increased and the radiometric precision got better. Starting in 2016, free imagery from the European Space Agency’s Sentinel 2A system began its regular operations. This provided for revisits that are as often as every 10 days. With Sentinel 2B now in operation, the revisits are as short as every 5 days and even every 2 to 3 days if you happen to be in the overlap between orbits.
The launch of Sentinel two years ago delivered four bands that show 33-feet details … blue light, green light, red light, and near infrared solar radiation, which is the number of bands necessary to produce any of the usual spectral indicators such as NDVI, Green NDVI, NDRE, and others that are better for handling soil noise and saturation issues.
The Fit For Row Crops
Row crops, usually annual crops, require yearly planning and decisions on seed selection, field preparation, planting, irrigation (in some regions), management during early stages of growth, management during reproductive stages of growth, preparing for harvest, harvesting, and post-season preparations for the next year or cycle. This all involves lots of time in the field — boots on the ground — with tractors, implements, and equipment for each phase of the farming cycle.
All of these management activities are best done by having good and frequent information about the soils that are in the field of interest — hence why one of the most common uses of imagery is for soil mapping and defining soil zones.
Service providers and growers also want to know about how the crop is doing throughout the growing season. A lot of this information can be gathered by using direct sampling methods – taking soil and plant samples, and using in field sensors such as the Greenseeker for leaf and yield (harvester) information. In-field photos might already be a part of what you do when you go into the field. Imagery can be an integral part of scouting throughout the season.
With a new image being taken every few days, you can now watch your crop emerge, grow (in terms of NDVI-like “vigor” values), and mature as you approach harvest. And, the better spatial resolution of these data let you see small management units that need attention at that scale of precision farming. You might even see that some parts of an otherwise healthy crop are lagging behind or turning more yellow. We use these frequent revisits of great satellite data to map both vigor (a combination of biomass density, a.k.a., leaf area index, and leaf health) and leaf health (pigmentation) as a “second option.” Using two spectral index maps instead of just one can reveal spatial patterns that are not seen in one or the other.
The image above provides an example of both a vegetation vigor index map and a pigmentation map for a set of fields in California.
If the vigor and pigmentation values are calibrated, then plots of these values for any field or place in a field will be useful for making maps of plant events or conditions such as emergence date (germination success), growth rates during vegetative growth stages, and changes during reproductive stages leading to being harvest ready. And, spatial patterns can indicate places of stress that you need to visit to determine what’s wrong and what needs to be done to fix the problem (if you can).
Solutions Coming to Every Field
Many satellite-based commercial companies too numerous to name are putting up relatively cheap satellites with multispectral cameras — even hyperspectral and/or thermal cameras — that will revisit at least every day and perhaps several times per day. This will make more reliable the notion that you can monitor the emergence, growth, and development of a crop of interest in a field of interest as often as every day, as often as every week for areas that are prone to be cloudy.
Such frequent revisits will also allow for better calibration of the data that so that change features can be derived from the time series of various spectral indices taken from the imagery. And, the spatial resolution of the imagery will be as good as 3 feet even for the various bands of blue light, green light, and red light — and perhaps red edge radiation and near infrared as well.
In the coming years, as we see the shift to the digital farm, data derived from imagery will persist in the overall farm management planning regimen. And the technology will be advancing, the prospects of what imagery analytics will yield in the future are exciting and the opportunities for improved decision-making will only get better.
One last thought about what will benefit the functionality of imagery: Using indices that are more bulletproof to the possibly confusing effects of foreground (sparse) vegetation. Spectral indices are most useful early in a season of growth when early patterns of health and stress are present — factors that affect potential yields (not actual yields, because in-season weather will have a significant future impact).
Optimizing the use of imagery really requires a team approach, with a trusted service provider working closely with grower-clients to ensure real problems are matched up with real solutions. The farmer will know what questions they want answered on their farm, and as the service provider you can advise as to whether imagery might help answer those questions.