Drones equipped with multispectral imaging help precision agriculture

Every year, Peterson Farms Seed, an independent corn and soybean seed company in the United States, researches and tests innovative methods of seeding and fertilizing. Field conditions such as soil composition and plant health must be monitored to determine seed and fertilizer application levels for each season. In order to achieve this goal, the traditional method is to use satellite images, and new technologies such as multi-spectral imaging are used to provide a faster and more accurate method for drones.

Peterson Farms Seed must determine the level of seed and fertilizer application. A process called "variable seeding / fertilization" that applies the optimal amount of seeds or fertilizer to a field based on its historical yield, soil health, altitude, and satellite or drone images. For example, Nolan Berg, an expert in precision agriculture systems at Peterson Farms Seed, said that if the traditional planting density of a corn field (see Figure 1) is 34,000 corn seeds per acre, then adjust the number of seeds sown to historically higher yields, Reducing the number of seeds in the lower yielding fields is positive.

Figure 1: DJI's quadrotor drone provides aerial images of cornfield harvesting for Peterson Farms Seed

Seasonal variable seeding / fertilization is another way for Peterson Farms Seed to reduce input costs while maintaining high yields.

"Maize will absorb more nitrogen fertilizer at the end of the growing season, so if all nitrogen fertilizer is applied to the field at the beginning, it will cause nitrogen loss due to environmental factors such as rain and hot weather." Berg said, "So At the beginning of the corn growing season, first apply a small amount of nitrogen fertilizer, and then re-evaluate the soil nutrients and the growth of the corn plants, and according to the actual situation, carry out variable fertilization as needed in the subsequent growth period, thereby helping to reduce nitrogen fertilizer loss risk."

A plan for variable fertilization is called a variable fertilization plan (see Figure 2). Berg said that traditional satellite imagery can help inform (show) what kind of variable fertilizer regimen should be specified.

Figure 2: Using the data collected by the RGB camera and multispectral camera on the drone to create a variable seeding / fertilization scheme

"Satellite images are very helpful for creating variable fertilization programs, but satellite images can only be acquired at most once a week or once every two or three weeks." Berg said, "If you want to apply fertilizer immediately, you need to use the latest Satellite images, so satellite images collected a week or weeks ago may not be a good representation of the local field and crop conditions at the time. "

Berg explained that by using drone airborne imaging technology, Peterson Farms Seed can now accurately evaluate fields and crops within a few inches, allowing them to change the application / fertilization of the field based on the actual assessment. Program.

"There is a large evaluation variation in the fields in different regions, so the application levels of seeds and fertilizers will be constantly readjusted based on the evaluation results." He said, "Some lands are barren and the annual yield is very low, so in It is meaningless to apply a large amount of seeds and fertilizer in such land. This is a waste of money, so the investment in these land will be reduced, but more will be invested in the land with better yield in order to obtain higher Yield. "

For this project, Peterson Farms Seed chose Sony Electronics Professional Solutions' US-based vegetation analysis platform, which includes a multispectral camera and the included Fast Field Analyze offline software.

Installing Sony's system on a DJI Phan-Tom4 quadrotor drone can create a flight plan covering the target area. once the plan is determined, the drone will take off on its own and begin to take images of the target field. After the drone lands, remove the SD card from the camera and place it in a laptop for processing. With this system, a map can be generated in minutes after the flight without wher to rely on any internet connection on site.

After Sony Fast Field Analyze software generates a map, the map is exported and imported into the SMS Farm software from Ag Leader. The input data includes color drone data and data input layers, including soil types, soil test numbers, standardized differential vegetation index (NDVI) charts (see Figure 3), and multi-year yield maps. These are run through software and combined to create a variable seeding / fertilization management area.

Figure 3: Standardized Differential Vegetation Index (NDVI) map provides images and data to help create variable fertilization programs

In a recent case, drones flew over several fields in a single day. By the afternoon of the next day, Berg used both software packages to process NDVI maps and create variable fertilization maps to fertilize the next day. The red part of the map shows the area of ​​drowned corn without fertilization. In the south (lower part of the map), the area with better drainage on the map, the company applied more fertilizer for better yields.

"The overall amount of investment in this farmland generally remains the same, but instead of wasting seeds or nitrogen fertilizer in low- or no-yielding areas, more fertilizer is applied in high-yielding areas to increase yields."

The MSZ-2100G sensor unit (see Figure 4) consists of a multispectral camera and an RGB camera. The RGB camera is equipped with a 12-megapixel Sony Exmor R CMOS image sensor for capturing aerial images, and the 2-megapixel NDVI multispectral camera is used to capture images in the red area and the NDVI range.

Figure 4: Sony's MSZ-2100G sensor unit is equipped with a 12-megapixel RGB camera and a 2-megapixel multispectral camera for aerial imagery

According to the description of NASA, NDVI is calculated by the following methods: near infrared radiation minus visible radiation, and the difference between the two is divided by the sum of near infrared radiation and visible radiation, that is:

NDVI = (NIR-VIS) / (NIR + VIS)

Most satellite vegetation indices use this formula to quantify the density of plant growth on Earth.

In addition to Sony's image sensor, the device also includes a Global Navigation Satellite System (GNSS) sensor, which is used to provide accurate geolocation data for imaging. Sony's Fast Field Analyzer software processes NDVI and RGB data and geographic location information, generates detailed maps, and enables fast on-site analysis without the need for an Internet connection.

Berg said: "A lot of work can be done with just one color image, but NDVI images can provide more scientific and accurate repeatable data. If somewhere is displayed in red on the NDVI scale, this is the field situation Exactly. "

He added, "Basically, NDVI images can show things that are not visible to the naked eye, while also showing the various hazards that crops may face earlier."

When Berg first started working at Peterson Farms Seed, the traditional method was to apply all planned fertilizers to the field at once before planting. Later, after research, they chose to apply about 70% -75% of the fertilizer first, and the remaining fertilizer was seasonally fertilized based on images taken by the drone, which gave them positive results. Peterson Farms Seed has spent almost the same amount of input on seeds and fertilizers as it did in the past, but its total output has increased.

Berg said, "In the final analysis, our current planting methods are definitely more efficient than before, and we hope to use less investment than before to achieve higher returns and ultimately improve profitability."