Precision Agriculture Technologies and Trends: Variable Rate Application

Precision agriculture is a step up from traditional farming. Also known as satellite farming, the concept is based on observing, measuring, and responding to inter and intra-field variability in crops. And variable rate application (VRA) is one of the most promising trends of modern precision agriculture. That’s why we’re so excited about it.

VRA is not just about fertilizing, seeding, and applying pesticides. For us, it’s more about the technologies used to automatically apply different kinds of expendable materials on a field, above and beneath it. Let us tell you more about variable rate application, its advantages, and the technologies that power it.

Precision agriculture is a step up from traditional farming with variable rate application (VRA) being the most promising trend that helps optimize crop growth, reduce costs, save expendable materials, and more. In our latest article, we examine the benefits of VRA and its impact on the agriculture business.

Variable rate application benefits

Variable rate application is an excellent way to optimize crop growth, reduce costs, save expendable materials (things like fertilizers, chemicals, seeds, etc.) and the environment. All thanks to the VRA technologies used in modern agriculture:

• Variable rate fertilizer. This is the very basis of modern agriculture.
• Variable rate seeding. Increases the seeding rate in the productive areas of the fields and reduces it in less fertile zones to increase the crop yield.
• Variable rate irrigation. It allows the monitoring of water supply, saving time, water, fuel, and reducing the wear of agricultural machinery.
• Variable rate pesticide. It saves pesticides, improves utilization rate, and significantly reduces environmental pollution.

Reaping the benefits of VRA

Using various technological means is critical for variable rate application. This includes smart engineering vehicles, seeders, fertilizers, soil sensors, Global Navigation Satellite System (GNSS), and geographic information system (GIS) applications for field mapping. But to fully take advantage of them, you’ll need:

• Special agriculture machinery like tractors, seeders, fertilizers
• Satellite imagery, weather and other sensors, and artificial intelligence (AI) applications
• Supporting infrastructure to process and store different kinds of data coming from different data sources

We need all this because the basis of precision agriculture is data. It’s vital to know where what, and when you seed, fertilize, or harvest, what expendable materials are applied, and how much.

This data is coming from a massive amount of different sources:

• Sensors. Soil nutrients, compaction, moisture, weather stations (temperature, humidity, wind speed)
• GNSS. Coordinates of events as well as point and time for collecting time-series data
• Drones and satellite imagery. Field hyperspectral imaging
• Maps. Field boundaries, surface levels, soil type
• Spatio-temporal data sources. Spatio-temporal specific data (trajectories of agricultural machinery, event points, spatio-temporal points, time-series data)
• AI solutions. Plant disease detection, weather conditions prediction

But raw data isn’t as useful as we’d like it to be. You need to process that data and turn it into information for insights, decision making, and automatic alerts, as well as control signals from equipment like seeders and fertilizers. Thus, you need to be able to:

• Collect data
• The process that data to get useful information for precision agriculture appliances
• Load the data into agricultural machinery
• Download actual data from tractors, fertilizers, seeders, and other machinery

Data-related challenges in VRA

While all this sounds promising, there are challenges you need to overcome to truly enjoy the benefits of VRA.

Same data from different sources

The first challenge is data that is the same by nature but comes from different sources and has different accuracy and temporal resolution.

You need to decide what source is available and more suitable for a specific application, for example, to determine weather conditions (precipitations, wind). Should you get data from a weather forecast service or from a local weather station? Is local weather station data available or not?

Another question is, how do you combine satellite data with high temporal resolution and low (relatively) spatial resolution and drone imagery with low temporal resolution but high spatial resolution?

Domain-specific data structures

The next challenge lies within the domain-specific data structures to store measurement (factual) data.

We solve it with a soil fertility map. A field is broken down into blocks, and for each block, we do several measurements (agrochemical analysis). Every block receives a separate fertilizer recommendation based on the results from the soil test. All this data should be stored in a spatial-temporal database to be analyzed for a given field and time.

It’s also important to monitor the trajectory of seeders and the actual number of seeds planted to be able to fight such common issues as “double plant” and “planter skip.”

Data accuracy

Finally, data accuracy and adequacy are a challenge you’ll need to handle. Here are our tips.

The spatial resolution of satellite data should correspond to the tasks you need to solve using that data. For example, it’s not a good idea to detect field boundaries using the Sentinel-2 images with the spatial resolution of 10m/pixel for small fields since its accuracy is +/-10 meters. Instead, it’s better to use more precise data from Planet to detect field edges and less precise Sentinel-2 data (but with more sensors) to calculate the required agricultural indices.

Satellite data can contain pieces with cloudiness. Clouds shade the images of the surface, bringing errors into indices calculation and making them useless. So we need to define if there’s cloudiness at specific parts of the image and exclude it from the calculator or try to work it out using sophisticated cloudiness cleaning ML algorithms.

Another problem is an atmospheric aerosol. We need to detect data with aerosol and apply specific methods to avoid calculation inaccuracy.

Lastly, you should remember that there are accurate satellites (1 meter per pixel), but they don’t have NIR-based sensing. There are also less accurate satellites (10 meters per pixel), which have NIR sensors and other data channels. The former can be used to accurately determine the area of interest, while the latter is used to obtain data from a wide variety of sensors for calculating various indices.

Conclusions

Variable rate application is an important and financially feasible approach you should definitely consider. According to reports, VRA can save 10% when sowing and growing agricultural crops, depending on the type of soil.

However, you can only benefit from VRA once you master the technologies it uses: sensors, GNSS, satellite and drones imagery, digital maps, and spatio-temporal data sources. All of them provide farmers with valuable data they need to interpret and apply. But don’t worry if you can’t do it yourself — Quantum will gladly help you out. Contact us anytime!