OneSoil data_cover

How we collect agricultural data

Big data makes precision farming possible. The more information available, the more accurate analytics and forecasts become. Our team collects data from satellites, sensors, agricultural vehicles, drones, in addition to the results of laboratory analyzes.


Various satellites have been photographing Earth for more than 50 years. Our main sources of information are images from the US Geological Survey and the Sentinel-2 satellite of the European Union's Copernicus programme, shot in the visible and infrared ranges. This data is free of charge and is available to anyone, but the images are stored in a large file size and specific format.

To analyze them, special tools are required. The OneSoil team developed its own algorithms for searching, downloading, processing and storing satellite images in a convenient format on our servers. Thanks to that, farmers receive information about the state of their field in a simple and accessible way, through our applications.
OneSoil_field from satellite
Display of a vegetation index for a field in a satellite image
The two main limitations of this type of satellite imagery are low resolution and dependence on cloudy weather. To resolve the first issue, we use modern methods of machine learning. To address the second one, we analyze images over a long period of time. We also use data from the Sentinel-1 radar satellite; its quality is not reliant on cloudy weather.

Satellite data can be used in a variety of ways in precision agriculture. We analyze images over several years to locate fields around the world, determine their boundaries, and define crops and the stages of their development. Also, satellite imagery allows you to monitor field status in real time. Our free OneSoil Scouting app is based on this principle.


In precision agriculture, many sensors are used. For example, weather stations can measure a whole range of features: air temperature and humidity, wind direction and strength, precipitation rate, soil density and acidity. Sometimes we ask farmers for this data, sometimes we measure these indicators ourselves. Our OneSoil Sensor measures soil and air moisture, their temperature, and also determines the level of light intensity for a specified portion of a field.
OneSoil Sensor in a field
OneSoil Sensor in a field
This kind of data helps plan agricultural work more efficiently. For example, it allows you to optimize an irrigation schedule and its frequency, and determine the optimal time for fertilizer application. Currently, we are developing a web application that will provide farmers with a deep analysis of the state of their fields.

Big data will allow us to make an accurate local weather forecast, predict plant diseases and pests, and measure other field indicators.

Agricultural vehicles

Tractors, combines, and seeders equipped with onboard computers are a valuable source of information. Data about crop yield is one of the most interesting for us. First of all, it is about wheat, triticale, barley, corn, peas, and soy.
We collected one of the largest data sets on yield in Eastern Europe. In 2015, we surveyed 600 hectares; in 2016 — 8 thousand hectares; and in 2017 — 11 thousand hectares.
Information from the onboard computers allows us to analyze a vehicle’s route, technical errors, engine speed, the amount of fuel, seeds, and fertilizers that have been used. For example, we can analyze the GPS track of a seeder that applied fertilizers and then compare data on the inputs rate with yield level. Also, data from agricultural machinery allows us to see how often varied agricultural operations were conducted.
Harvesting_instagram OneSoil
Photo by Slava Mazai. See more photos at Instagram
For the last two years, we have been working on a wireless modem that transmits data from onboard computers to our online system. It will help farmers streamline another process — data from their agricultural vehicles will be collected and delivered remotely. Also, the modem will let users control the work of machine operators and analyze implementation of agricultural procedures without leaving the office.

Now, our free OneSoil platform simplifies another important process: there you can calculate nitrogen variable rates and create a task for onboard computers in a couple of minutes.

Drones and aerial photography

Drone photography can be utilized to view areas captured from satellite imagery more clearly. The drone is airborne for 15−30 minutes and takes high-resolution images, which then need to be combined into an orthophoto map (aerial photographs corrected to scale so that geographic measurements can be taken directly from prints — OneSoil). Working with it, you can detect weeds and pests, determine the height of shoots, assess plant health and analyze other field processes.

In the beginning, we used drone images to determine the field relief and its boundaries, to analyze the overwintering of plants and their condition throughout the season, and also to test our hypotheses about various field processes. The main drawback of drone filming is its locality. It cannot be scaled to the whole world, so now we rarely use this method.
Team_instagram OneSoil
OneSoil team. See more photos at Instagram

Laboratory analysis

Agrochemical analysis of soil is one of the traditional estimation methods to determine necessary fertilizer rates. It allows you to detail the chemical composition of a field with a number of indicators; the key ones are the content of humus, phosphorus and potassium, and soil acidity.

We examined the fields of Belarus in order to understand whether this method is suitable for the differentiated application of fertilizers. Our tests showed that it is ineffective; this method doesn't allow us to estimate the variability of nutrient content accurately in different parts of the field.

Our hypothesis states that it is necessary to calculate the nutritional rate on the basis of yield indicators. If the area is of a high yield, then the depletion of nutrients is high and more fertilizers need to be applied. If the yield is low, then the depletion of nutrients is also low, and the fertilizer rate should be reduced.

Field studies validated this assumption. We will use this method when developing programs for differentiated application of phosphorus and potassium fertilizers.
Do you like this post?
Related Articles