Soil Science in Digital Agriculture

Where does the data for such a model come from?

The basis of DSM is the formalisation of the classical factor approach proposed by V. V. Dokuchaev, but expressed in quantitative form. In pedometrics, this became known as the “scorpan” model (McBratney et al., 2003). The name is an acronym for the factors that influence soil properties:

• s (soil) – other already known properties of the soil itself;
• c (climate) – climatic parameters;
• o (organisms) – vegetation and human activity;
• r (relief) – topography;
• p (parent material) – parent rock;
• a (age) – soil age;
• n (spatial position) – spatial position.

The idea is simple: if we measure soil properties at a number of points (training the model) and relate them to “predictor variables” (relief, satellite data, climatic maps) at the same points, we can build a statistical dependency model. Then, applying this model to all points in the landscape where we only have the predictor variables, we obtain a predictive map of the desired property (e.g., organic carbon content, pH, or available phosphorus) (McBratney et al., 2003; Weil and Brady, 2017).

Why is this important for management?

1. Quantitative assessment. We get not a soil name but a precise value of properties. To decide on fertiliser application, we need to know how much phosphorus is available, not just that the soil is “leached chernozem”.

2. Spatial continuity. Instead of assuming the whole field is uniform, we see zones with high and low nutrient contents (Weil and Brady, 2017).

3. Accounting for uncertainty. Modern statistical methods (e.g., regression kriging) allow us to estimate not only the predicted value of a property but also the probability of error at each point on the map (Huang et al., 2012). This means we understand where our data are reliable and where additional sampling is needed.

Thus, digital soil mapping is the first and critically important step on the path from generalised knowledge to a specific management decision. It transforms soil from a classification object into an object of quantitative digital modelling, creating the spatial foundation for all subsequent precision agrotechnologies.

In the next chapter, we will discuss what data we can obtain remotely and with field sensors to feed these models.

Physical principle: spectral reflectance

Most remote sensing methods are based on measuring reflected electromagnetic radiation from the Earth’s surface. Each object (soil, vegetation, water) has its own characteristic reflectance spectrum – a kind of “digital fingerprint”. Different soils reflect and absorb light differently in the visible, near‑infrared (NIR), and short‑wave infrared (SWIR) ranges (Weil and Brady, 2017; Scheffer et al., 2018).

For example:

• Organic matter strongly absorbs light in many ranges, so soils with high humus content look dark and have low reflectance (Weil and Brady, 2017; Huang et al., 2012).
• Iron oxides (hematite, goethite) give soils reddish and yellowish hues, which are clearly detected in the visible spectrum (Scheffer et al., 2018).
• Soil moisture also reduces reflectance, especially in the short‑wave infrared range (Scheffer et al., 2018; Doolittle, 2012).
• Vegetation has a pronounced “red edge” – a sharp increase in reflectance in the near‑infrared due to leaf structure. This allows calculation of vegetation indices (e.g., NDVI), which serve as indicators of biomass, plant stress, and, indirectly, nutrient availability and soil water regime (Weil and Brady, 2017).

Types of sensors and their data

Different types of sensors are used for soil research:

1. Multispectral sensors (Landsat, Sentinel‑2, SPOT). These are the most common systems. They record reflectance in several (usually 4–13) broad spectral bands (e.g., blue, green, red, NIR, SWIR). The resolution of such images can reach 10–30 m, allowing discrimination of soil cover heterogeneities at the field scale (Weil and Brady, 2017). They are widely used for mapping soil types, estimating organic carbon content, and moisture (Huang et al., 2012).

2. Hyperspectral sensors (AVIRIS, HyMap). These sensors measure reflectance in hundreds of narrow spectral bands. This allows creating “spectral fingerprints” of individual minerals and compounds (e.g., kaolinite, montmorillonite) (Weil and Brady, 2017; Doolittle, 2012). Such detail opens possibilities for direct mapping of soil mineralogical composition, but the cost and complexity of data processing limit their widespread use in agronomy, although they are actively used in geology and ecology (Weil and Brady, 2017; Doolittle, 2012).

3. Active radar systems (SAR, Radar). Unlike passive optical sensors, radar emits microwaves and measures the reflected signal. Microwave penetration into the soil depends on its dielectric properties, which strongly correlate with moisture. This makes radar data especially valuable for estimating soil moisture, even under cloudy conditions (Weil and Brady, 2017; Doolittle, 2012). Moreover, radar with long wavelengths can provide information about surface structure and even subsurface horizons (Weil and Brady, 2017).

4. LiDAR. This is an active sensor that measures distance to an object using laser pulses. Its main application is creating high‑precision digital elevation models (DEMs), which are critically important for assessing the relief factor (r) in scorpan models. LiDAR allows “seeing” micro‑relief that is inaccessible to other methods (Weil and Brady, 2017; Huang et al., 2012). In addition, LiDAR can be used to estimate vegetation height and structure, providing information on biomass and carbon stock.

How does remote sensing become soil data?

It is important to understand that we do not see the soil directly – we see reflected light. Therefore, remote sensing does not replace ground observations but complements them. In digital mapping, we use spectral data as “proxies” (substitute variables). For example, we can establish a statistical relationship between reflectance in certain bands and organic matter content in the topsoil (measured at reference points). Then, applying the obtained model to the whole image, we get a map of organic carbon content for the entire field (Weil and Brady, 2017; Huang et al., 2012). This approach, called “spectral mapping”, is particularly effective on bare soil fields or with sparse vegetation (Weil and Brady, 2017).

Connection to management decisions

Remote sensing data are a critically important source of information for variable management. They allow:

• Pre‑dividing the field into zones of different productivity (management zones).
• Assessing crop status and identifying areas with nitrogen deficiency or stress due to moisture shortage.
• Planning targeted soil sampling for laboratory analysis.
• Creating yield maps (with combine harvesters’ sensors), which are then compared with remote sensing data and soil maps to identify causes of low yield (Weil and Brady, 2017).

Thus, remote sensing is not just a “pretty picture”. It is a powerful and cost‑effective way to obtain spatially continuous information about soil and plant status, which serves as the basis for precise and timely agronomic decisions.

In the next chapter, we will move to an even more local level – we will consider what field soil sensors can tell us about the soil “live”.

1. Geophysical sensors: electromagnetic induction and electrical resistivity

These methods measure the apparent (or effective) electrical conductivity of the soil (ECa – apparent electrical conductivity) (Doolittle, 2012; Huang et al., 2012). ECa is an integral characteristic that depends on several important soil properties:

• Soil moisture. The higher the moisture, the higher the conductivity, as water is the main conductor (Doolittle, 2012; Scheffer et al., 2018).
• Content and type of clay minerals. Clays have high cation exchange capacity and ability to retain ions, which increases conductivity. Smectites and vermiculite are particularly influential (Doolittle, 2012).
• Salt concentration in the soil solution. The higher the salinity, the higher the conductivity (Doolittle, 2012; Huang et al., 2012).
• Depth and density of different horizons. Sensors have different sounding depths, allowing differentiation of layers with different conductivity (e.g., topsoil and subsoil horizon with high clay content) (Doolittle, 2012).

Instruments operating on this principle:

• Electromagnetic induction meters (EMI, e.g., EM38). They generate an electromagnetic field that induces a secondary field in the soil. By measuring the ratio of primary to secondary fields, ECa is computed. These instruments are non‑contact, rapidly scan large areas, and are sensitive to depths up to 1.5–2 m (Doolittle, 2012; Huang et al., 2012).
• Resistivity meters (Veris, OhmMapper). They use contact electrodes inserted into the soil (usually as discs or blades pressed while moving). Electric current is passed between electrodes and resistance is measured. This gives similar information about ECa but with different depth sensitivity and resolution (Doolittle, 2012).

What does this give for management? ECa maps are among the most popular and useful in precision agriculture. They allow:

• Dividing the field into homogeneous zones (management zones) based on indirect indicators (texture, horizon depth, drainage) (Weil and Brady, 2017; Huang et al., 2012).
• Directing sampling for laboratory analysis to minimise variability within zones.
• Identifying areas with salinity, compaction, or poor drainage problems (Doolittle, 2012).
• Calibrating models for predicting other properties (e.g., organic matter content, which often correlates with ECa) (Huang et al., 2012; Weil and Brady, 2017).

2. Ground‑Penetrating Radar (GPR)

Ground‑penetrating radar is an active geophysical method based on emitting high‑frequency electromagnetic pulses (30 MHz – 1.2 GHz) and recording reflected signals from interfaces between media with different dielectric permittivity (Doolittle, 2012). Dielectric permittivity strongly depends on water content: water has high permittivity (~80), while soil minerals have low permittivity (from 2 to 10) (Doolittle, 2012; Scheffer et al., 2018).

What can be seen with GPR?

• Boundaries of soil horizons (e.g., transition from sandy A‑horizon to clayey B‑horizon, or depth to a dense horizon, ortstein, fragipan) (Doolittle, 2012; Weil and Brady, 2017; Huang et al., 2012).
• Groundwater level, especially on sandy soils (Doolittle, 2012; Weil and Brady, 2017).
• Subsurface sediment structure, presence of boulders, roots, voids (Doolittle, 2012; Scheffer et al., 2018).

Application in management: GPR allows rapid and non‑destructive “looking” into the soil to:

• Map the depth to bedrock or impermeable horizons.
• Assess topsoil thickness.
• Study spatial distribution of drainage conditions.
• Assist in planning reclamation measures and well drilling (Doolittle, 2012; Weil and Brady, 2017).

3. Optical sensors and near‑infrared (NIR) spectroscopy

While remote sensing uses the same spectral principles but from great heights, proximal optical sensors work directly on the surface or even in contact with the soil (Weil and Brady, 2017; Huang et al., 2012). These can be:

• Portable (or machine‑mounted) spectroradiometers measuring reflectance from the soil surface in visible and near‑IR ranges.
• Specialised sensors mounted on coulters that contact the soil and measure reflectance from the furrow wall.

The main advantage over remote sensing is the ability to obtain data regardless of cloud cover, time of day, and even in the presence of vegetation (if the sensor works at depth) (Weil and Brady, 2017). In the laboratory, NIR spectroscopy has long been used for rapid assessment of carbon, nitrogen, moisture, texture, and even mineralogical composition of samples. Development of field versions is active, and commercial solutions for “on‑the‑go” measurement already exist (Weil and Brady, 2017; Huang et al., 2012).

For management this means:

• Ability to obtain high‑resolution soil property maps in real time.
• Significant reduction in laboratory analysis costs, especially for parameters such as organic matter, moisture, nitrogen (though accuracy for nitrogen is still questionable) (Weil and Brady, 2017).
• Using these data as “inputs” for variable rate application (VRA) systems for fertilisers and amendments.

4. Contact sensors for physical properties

These include sensors that measure:

• Moisture and temperature (e.g., TDR probes, capacitive sensors). These are standard tools for monitoring soil water and thermal regimes within drip irrigation systems or for scientific purposes (Scheffer et al., 2018; Weil and Brady, 2017).
• Density and mechanical resistance (penetrometers). Although this is more of a “mechanical” method, it provides information on the degree of compaction, which is critical for assessing soil structure condition and predicting root system development (Scheffer et al., 2018).
• Electrical conductivity (already mentioned above) and pH (using special ion‑selective electrodes that can be built into penetrometers).

Application: These sensors are indispensable for monitoring soil condition dynamics (e.g., how moisture and temperature change during the season), for assessing soil suitability for tillage, and for identifying zones at risk of over‑compaction.

Thus, field soil sensors are our “magnifying glass” and “hand” that can “feel” the soil on‑site. They provide high‑precision, local, and often continuous data, which serve as a bridge between the global view from satellites and point laboratory analyses. The combined use of remote sensing (for coarse zoning), field sensors (for detail and calibration), and point sampling (for validation and laboratory analysis) allows creating a complete and reliable digital twin of the field’s soil cover.

In the next chapter, we will see how these collected and processed data are turned into specific management instructions for machinery.

Working principle: from data to instruction

The entire VRA process can be represented as a closed loop consisting of several key stages (Weil and Brady, 2017):

1. Data collection (Information stage). At this stage, we obtain information about the field. This can include:

• Soil property maps created using digital mapping (e.g., available phosphorus, potassium, pH, organic matter) (Weil and Brady, 2017; Huang et al., 2012).
• Sensor data (electrical conductivity, spectral data) that correlate with fertility, moisture, or texture (Doolittle, 2012; Weil and Brady, 2017).
• Remote sensing data (vegetation indices indicating plant stress) (Weil and Brady, 2017).
• Yield maps from previous years (Weil and Brady, 2017).

2. Analysis and creation of prescription maps (Analytical stage). Based on the collected data, a specialist (agronomist, soil scientist) or specialised software creates a prescription map. This is a digital map of the same field, but where each point (or zone) is assigned a specific application rate for the given resource (e.g., kg/ha of nitrogen fertiliser). This rate is calculated based on agrochemical recommendations that consider the current nutrient content in the soil and the planned yield (Eash et al., 2016; Foth, 1990).

3. Execution (Technological stage). The prescription map is uploaded to the onboard computer of agricultural machinery (seeders, sprayers, fertiliser spreaders). While moving across the field, the machinery uses GPS/GNSS to determine its position with high accuracy and, consulting the prescription map, automatically adjusts the application rate in real time (Weil and Brady, 2017). For example, in one area the seeder sows more seeds, in another less; the spreader applies more fertiliser where it is deficient and less where it is excessive.

Why it works: management zones

The basis for creating prescription maps is often management zones. These are parts of the field that are relatively homogeneous in key properties affecting yield: soil type, topography, water availability, nutrient content (Weil and Brady, 2017). For each such zone, its own management strategy is developed.

It is important to understand that VRA does not always mean variable‑rate application “at every point”. In practice, the zone management approach is often used, where the field is divided into several larger zones, and each zone is assigned its own rate (Weil and Brady, 2017; Eash et al., 2016). This is a simpler and cheaper option, but it is less accurate than continuous VRA.

Examples of VRA application in the context of soil science

• Variable nitrogen application. Based on vegetation index (NDVI) data or field sensor data reflecting leaf chlorophyll content, it is possible to identify areas where plants suffer from nitrogen deficiency. More nitrogen is applied as a top‑dressing to these areas, and less to areas with good plant development (Eash et al., 2016; Weil and Brady, 2017).
• Variable liming. Using a pH map created through digital soil mapping, lime can be applied at variable rates: more where pH is low, and less or none where pH is optimal (Weil and Brady, 2017). This saves money and prevents over‑liming.
• Variable seeding. Based on productivity and soil property maps, seeding rates can be varied: on more fertile areas with better water supply, the rate can be increased, and on less productive areas decreased (Weil and Brady, 2017).

Important notes

1. Data are the key element. The quality of VRA directly depends on the quality of the initial data. Poor maps lead to poor management decisions (Weil and Brady, 2017).

2. VRA is not a panacea. It is a tool that should be used in combination with other precision agriculture methods. VRA alone does not solve problems of soil compaction, diseases, or moisture deficit.

3. Economic efficiency. VRA requires additional costs for equipment and data analysis. It becomes economically justified on large fields with high spatial variability (Weil and Brady, 2017).

4. Environmental aspect. VRA is an important step towards sustainable agriculture, as it allows reducing overuse of fertilisers and pesticides, thereby decreasing their input into the environment (Weil and Brady, 2017).

Thus, variable rate application is the bridge between digital soil data and real agronomic actions. It turns information into specific instructions for machinery, implementing the principle of “right product, right place, right time”.

In the next chapter, we will summarise by discussing what role the soil scientist plays in this complex system and how he or she interacts with other specialists in the modern agroteam.

What is pedometrics and why is it important?

Pedometrics is a branch of soil science that deals with quantitative description and modelling of soil properties and processes using mathematical and statistical methods (McBratney et al., 2003; Huang et al., 2012). If classical soil science answers the questions “what?” and “why?”, pedometrics answers “how much?”, “where?”, and “with what probability?”.

It is pedometrics that is the theoretical and methodological foundation of digital soil mapping and the entire precision agriculture system. It provides the bridge between empirical observations and management decisions (Weil and Brady, 2017; Foth, 1990).

How does pedometrics turn data into knowledge?

The transformation process can be represented as several sequential steps:

1. Data collection and structuring. This is the first but extremely important step. We collect data from different sources: laboratory analysis results of samples, field sensor readings, remote sensing spectral data, weather data, digital elevation models. All these data must be georeferenced (have coordinates) and brought to a unified format (Weil and Brady, 2017; Huang et al., 2012).

2. Analysis of spatial structure. Pedometric methods such as variography and spatial autocorrelation analysis allow us to understand how strongly soil property values at different points in the field are related and at what distance this relationship persists (Huang et al., 2012; Eash et al., 2016). This answers the question: “How often should samples be taken to get a reliable picture?”.

3. Building predictive models. At this stage, we use methods of regression analysis, machine learning (neural networks, random forest, support vector machines), and geostatistics (kriging) to create continuous maps of soil properties (Huang et al., 2012; McBratney et al., 2003). For example, we can build a model that predicts organic matter content based on topography, NIR reflectance, and electrical conductivity. This is the essence of digital soil mapping.

4. Uncertainty assessment. Any model gives a probabilistic, not perfect, prediction. Pedometric methods allow estimating the prediction error for each point on the map (Huang et al., 2012; Weil and Brady, 2017). This is extremely important for management decisions: we understand how much we can trust a given recommendation.

5. Integration of data into Decision Support Systems (DSS). Predictive soil property maps, supplemented with uncertainty estimates, are loaded into specialised software (agronomic platforms). These systems combine soil data with plant status data (vegetation indices), weather forecasts, and economic indicators to calculate optimal application rates for fertilisers, amendments, and plant protection products (Weil and Brady, 2017; Eash et al., 2016).

The role of pedometrics in various aspects of digital agriculture

• Fertility management. Pedometric models allow accurate calculation of the spatial distribution of available nutrients (N, P, K) and, on this basis, create prescription maps for variable fertiliser application (Foth, 1990; Eash et al., 2016; Weil and Brady, 2017).
• Water regime management. Models that account for textural heterogeneity of the soil (obtained from digital maps) help plan irrigation systems and assess risks of waterlogging or drought (Eash et al., 2016; Foth, 1990).
• Assessment of environmental risks. Pedometrics allows modelling the migration of pollutants (nitrates, pesticides) in the soil profile and assessing risks of groundwater contamination (Weil and Brady, 2017; Foth, 1990). This is critically important for compliance with environmental regulations.
• Yield prediction. Integration of soil data with weather and crop status data allows building yield forecast models that account for spatial variability (Eash et al., 2016).

Soil data as the foundation for “smart” agriculture

Thus, soil data are not just information for a report. They are an asset that creates value for the agricultural producer. Pedometrics is the key tool to extract this value, turning scattered measurements into a knowledge system that:

• Increases the precision and efficiency of resource use (fertilisers, water, plant protection products) (Weil and Brady, 2017; Eash et al., 2016).
• Reduces negative environmental impact (Foth, 1990; Weil and Brady, 2017).
• Increases economic efficiency of production (Weil and Brady, 2017).
• Allows shifting from reactive management (responding to problems) to proactive management (preventing problems) (Eash et al., 2016).

That is why, in the modern agroteam, the soil scientist who masters pedometric methods becomes not just a “cartographer” but a key analyst and strategist, providing the digital foundation for all agrotechnological decisions.

In the next chapter, we will look at how the soil scientist interacts with other specialists in such a team to make that interaction effective.

New role of the soil scientist: “translator” and “integrator”

In a digital agroteam, the soil scientist performs three key functions:

1. “Translator” of complex soil processes. He or she understands how physical, chemical, and biological soil properties interact with each other and how they affect plant growth. He or she translates these complex interrelations into language understandable to the whole team (e.g., “here we have low cation exchange capacity, so potassium leaches quickly and we need to apply it more often” or “on this plot, bulk density is high, so roots cannot penetrate deeply and plants suffer from drought”).

2. “Integrator” of data. The soil scientist knows what data need to be collected, how to interpret them, and how to integrate them into a unified digital management system. He or she understands how to combine sensor data, laboratory analysis results, satellite images, and agrochemical databases to create a holistic picture of the field’s soil cover status (Weil and Brady, 2017; Huang et al., 2012).

3. “Strategist” of fertility and soil health management. He or she does not just give recommendations but develops a long‑term strategy for managing soil resources, aimed not only at obtaining high yields in the current season but also at preserving and enhancing soil fertility and health in the future (Weil and Brady, 2017; Eash et al., 2016).

Interaction of the soil scientist with other specialists

Let us consider how the soil scientist interacts with each of the key members of the modern agroteam:

1. With the agrochemist: The agrochemist is a specialist in plant nutrition. He or she determines optimal doses, timing, and forms of fertiliser application to achieve target yields (Eash et al., 2016; Foth, 1990). The soil scientist provides fundamental information: maps of available nutrients (N, P, K), pH maps, and cation exchange capacity (CEC) maps. The soil scientist explains to the agrochemist why in some zones fertilisers will work effectively and in others not, and helps develop a variable application system that accounts for spatial heterogeneity (Weil and Brady, 2017). Joint work allows not only increasing yields but also significantly reducing environmental pollution risks from over‑fertilisation.

2. With the agronomist (tillage and crop rotation specialist): The agronomist is responsible for choosing the tillage system, crop rotation, and sowing technology. The soil scientist provides critically important information about soil physical properties (particle‑size distribution, structure, bulk density, water‑holding capacity, infiltration rate) (Scheffer et al., 2018; Weil and Brady, 2017). He or she explains how different tillage systems affect soil structure and its ability to accumulate and retain moisture. Together they develop a strategy that minimises risks of compaction and erosion, preserves and improves soil structure, and increases water use efficiency. For example, the soil scientist may recommend a transition to minimum or no‑till on areas with low organic matter content to prevent its mineralisation (Weil and Brady, 2017).

3. With the reclamation specialist: The reclamation specialist deals with improving soil water regime (drainage, irrigation, groundwater level control). The soil scientist provides maps of soil hydraulic properties: water retention curves, maps of depth to impermeable horizons, infiltration rate data (Eash et al., 2016; Foth, 1990). The soil scientist helps the reclamation specialist understand where and which reclamation measures will be most effective and predicts their impact on soil processes. For example, he or she can show that on areas with high clay content and low permeability, drainage will be less effective than on light soils.

4. With the microbiologist: The microbiologist studies soil biota (bacteria, fungi, actinomycetes) and its role in organic matter decomposition, nutrient cycling, and pathogen suppression (Eash et al., 2016). The soil scientist helps the microbiologist understand which physico‑chemical soil properties (pH, aeration, moisture, organic matter content) determine the activity and structure of microbial communities. Joint work allows developing approaches to managing microbial activity to improve plant nutrition and reduce disease incidence.

5. With the breeder: The breeder creates plant varieties adapted to certain conditions. The soil scientist supplies the breeder with data on spatial variability of soil properties. He or she helps the breeder understand which soil conditions (e.g., acidity, salinity, moisture deficit) are limiting on a given field and which varieties will be best adapted to these conditions (Eash et al., 2016). This allows more precise variety testing and creation of varieties adapted to specific soil‑climatic conditions of the region.

6. With the precision agriculture specialist: This is the soil scientist’s key partner in digital agriculture. The precision agriculture specialist is responsible for implementing and operating VRA technologies, GPS navigation, and prescription mapping software. The soil scientist is the main “data supplier” for the precision agriculture specialist (Weil and Brady, 2017). He or she interprets sensor data, builds digital maps of soil properties, and develops algorithms for variable management. Without data created by the soil scientist, the precision agriculture system works “blindly” and loses its effectiveness.

7. With the ecologist: The ecologist assesses the impact of agricultural activities on the environment (water bodies, biodiversity, climate). The soil scientist plays a key role in this work, as soil is the main filter and buffer in the landscape (Weil and Brady, 2017; Eash et al., 2016). The soil scientist models the migration of pollutants (nitrates, pesticides, heavy metals) in the soil profile and assesses risks of their leaching into groundwater (Foth, 1990). He or she helps the ecologist develop measures to reduce the negative impact of agriculture on the environment and to preserve soil biodiversity.

The soil scientist as a “systems analyst” of the field

Thus, in the modern agroteam, the soil scientist is not just a soil specialist, but a systems analyst of the field who understands how all components of the soil system (physical, chemical, biological) interact with each other and how they affect plant growth and environmental conditions. He or she masters modern digital tools (GIS, pedometrics, remote sensing) and can translate complex soil processes into data language understandable to the whole team. It is the soil scientist who connects fundamental knowledge about soil with practical agribusiness tasks, ensuring scientific validity and environmental safety of agrotechnological decisions.

Scheme: The path from soil formation to digital management

1. Soil formation (Pedogenesis) – “The Beginning”

Everything starts with soil formation. This is the process of interaction of five factors: climate, relief, parent material, organisms, and time (Dokuchaev‑Jenny factors) (Weil and Brady, 2017). They determine what kind of soil will form at a given site. This is the fundamental basis that sets all subsequent properties.

2. Soil profile – “Architecture”

The result of soil formation is the soil profile – a vertical sequence of genetic horizons (Eash et al., 2016; Scheffer et al., 2018). Each horizon has its own unique physical, chemical, and biological characteristics. Studying the profile gives us the first qualitative information about the soil: its type, thickness, particle‑size distribution, presence of illuvial or eluvial processes.

3. Soil physics – “Matrix”

Physical properties (particle‑size distribution, structure, density, porosity, water permeability) form the matrix in which all vital processes take place (Scheffer et al., 2018; Weil and Brady, 2017). They determine how much moisture and air the soil can hold, how quickly water infiltrates, how easily roots can penetrate deep. This is the “engineering basis” of fertility.

4. Soil organic matter – “Energy and nutrition”

Organic matter is the “blood” of the soil system (Eash et al., 2016; Weil and Brady, 2017). It is a source of energy for microorganisms, a reserve of nutrients (especially nitrogen), and improves structure and water‑holding capacity. The content and quality of organic matter are key indicators of soil health.

5. Soil chemistry – “Buffer and menu”

Chemical properties (pH, cation exchange capacity (CEC), content of available macro‑ and microelements) determine the nutritional value of the soil and its ability to buffer adverse effects (acidification, salinisation, pollution) (Eash et al., 2016; Foth, 1990). Soil chemistry is the “menu” from which plants select necessary elements.

6. Soil biota – “Engine”

Microorganisms, fungi, invertebrates are the “engine” of soil processes (Eash et al., 2016; Weil and Brady, 2017). They decompose organic matter (mineralisation), fix atmospheric nitrogen, convert phosphorus into available forms, structure the soil, and suppress pathogens. Biota is the living phase of the soil that processes the “matrix” into “nutrition” for plants.

7. Soil fertility – “Result of interaction”

Fertility is the integral result of the interaction of all the above components: physics, chemistry, biology, and organic matter (Eash et al., 2016; Foth, 1990). It is the ability of the soil to satisfy plant needs for nutrients, water, and air. Fertility is not a static characteristic but a dynamic state that can be managed.

8. Soil health – “System quality”

Soil health is a broader concept than fertility. It includes not only the ability to produce crops but also the ability to maintain biodiversity, purify water and air, and resist stresses (drought, diseases, pollution) (Eash et al., 2016; Weil and Brady, 2017). Health is an integral assessment of the sustainability of the soil system, its ability to perform all its ecological functions.

9. Digital management – “Implementation tool”

Finally, we come to the final link of the chain – digital management (Weil and Brady, 2017; Eash et al., 2016). It is here that knowledge about soil turns into concrete actions. Using remote sensing and soil sensors, we obtain quantitative data on soil properties (Doolittle, 2012; Weil and Brady, 2017). Using pedometrics and digital mapping, we create spatial models of these properties (McBratney et al., 2003; Huang et al., 2012). Based on these models, we develop prescription maps for variable resource application (VRA) (Weil and Brady, 2017). Finally, using automated machinery and GPS navigation, we implement these decisions in the field.

Schematic representation

This scheme is not linear and unidirectional. There are complex feedback loops between the elements. For example, management (9) affects biota (6) and organic matter (4), while soil health (8) is the result of the interaction of all previous components.

Why is this scheme important?

This scheme demonstrates that digital agriculture is not just a set of technologies, but a logical continuation and practical implementation of fundamental soil knowledge. Without understanding the “lower” levels (formation, profile, physics, chemistry, biota), the “upper” level (digital management) will work inefficiently or even harmfully. Digital tools are only a way to make management more precise, operational, and environmentally safe. But the basis of this management is a deep understanding of how the soil works as a living system.