The effectiveness of using information technologies in agriculture to support climate change adaptation and mitigation depends largely on the qualitative and quantitative characteristics of data collected through sensors, satellite monitoring, GIS technologies, drones, and other information- gathering tools. This article presents the development of a methodological toolkit for assessing the potential and forecasting carbon sequestration by agricultural crops using artificial intelligence. The ground-based installations were used to obtain the database-sensors that capture geolocation and sensors of physical characteristics such as humidity, temperature, and insolation, etc. Using the method of correlation and regression analysis, mathematical models were developed to predict the possible volumes of carbon dioxide sequestration depending on the size of sown areas and crop yields in the Ternopil region. It has been investigated that increasing the productivity of field crop agrocenoses is an important way to reduce the CO₂ content in the atmosphere and prevent further global warming on a planetary scale. To verify the accuracy of the obtained data and generate predictive analytics, the XGBoost gradient boosting method was applied. The application of this approach made it possible to increase the accuracy of predicting the volume of carbon dioxide assimilation by agrocenoses, taking into account the variability of yields and crop areas in different years. The results obtained allow us to predict potential changes in the ability of crops to fix CO₂ depending on climatic and farming factors, which is important for developing a strategy for sustainable agricultural production. The obtained results are the basis for further research on the use of artificial intelligence in carbon farming management.
The application of climate change adaptation and mitigation measures in agriculture represents a set of innovative approaches to farming. In particular, the introduction of climateneutral innovations in agricultural natural resource management are nature-based solutions for the development of precision agriculture, regenerative agriculture, and low-carbon agriculture itself. Such innovations are a synergy of carbon-neutral technologies and digital technologies to ensure the sustainable use of agricultural resources in the context of strengthening both food and climate security.
"Information technology in agriculture is used to generate yield maps, machinery movements;
calculate the need for seeds, planting material, fertiliser; draw up a scheme of sown areas for future
years; assess soil conditions; create an electronic field log with the ability to sort by harvest year;
forecasting of technological operations for the next season or several years; preparation of reports
with diagrams on the presence of diseases and pests, as well as weeds in the fields; division into
groups of diseases [
The "GIS technologies are a set of digital techniques that allow analysing the physical
characteristics of the planet's surface. For better perception, the data obtained with their help is
visualised. This information is used to create multi-layer maps, atlases, graphs, charts and
interactive applications. The benefits of using GIS technologies for agriculture include increased
yields and the development of precision farming, which increases the average yield
of farms by 22% while reducing clean water consumption by 20%. In addition, the automation of
machinery reduces the use of manual labour (for example, with the help of GIS solutions, one farm
worker can operate four machines at the same time), saves fertilisers and plant protection products,
and optimises associated costs (for example, the use of GIS reduces the cost of purchasing fuel,
maintaining machinery and storing materials by 7-9%)" [
- "Threat of drought" displays days with temperatures above +30 ℃ to assess damage from heat
stress" [
One of the consequences of anthropogenic activities of mankind on the environment is global
warming, which began in the middle of the second half of the twentieth century and continues to
this day. Ecologists and climatologists believe that climate change is caused by an increase in the
content of greenhouse gases in the atmosphere, in particular carbon dioxide. "Over the past 66
years, the CO₂ content has increased by 34.7%, from 315.2 ppm in 1958 to
424.6 ppm in 2024" [
One of the ways to reduce the CO₂ content in the atmosphere is through agricultural production,
as agrocenoses can accumulate up to 1 Gt/year of carbon on a global scale [
To address the issue of agricultural decarbonisation, artificial intelligence technologies are being
applied. "Precision farming technologies based on artificial intelligence allow farmers to use
resources such as water, fertilisers and pesticides more efficiently. In order to minimise the overuse
of resources, reduce emissions associated with their production and application, using machine
learning and data analytics algorithms, artificial intelligence can analyse various factors such as
soil conditions, weather conditions and crop requirements. For example, AI- driven irrigation
systems can significantly reduce water use and energy consumption, contributing to a lower
carbon footprint. By preventing crop losses and reducing the need for chemical treatments,
artificial intelligence can help reduce emissions associated with excessive pesticide use and freight
transport" [
The study [
Another way to prevent climate change goes with [
A review of scientific papers confirms the relevance of carbon sequestration in agriculture. Of particular significance is the integration of remote sensing technologies and artificial intelligence to create and process a database. Attention is drawn to employ new technologies and the selection of plants with deeper root systems to mitigate climate change. However, in the context of climate change adaptation and mitigation in agriculture, an open challenge remains: how to accurately quantify and predict the amount of carbon dioxide, its removal and conservation by crops through carbon farming.
In view of this, the aim of the article is to develop a methodological toolkit for assessing the potential and forecasting the carbon sequestration of agricultural crops using artificial intelligence.
Under martial law in Ukraine, there is a ban on the use of equipment installed on aircraft (e.g. drones). In view of this, ground -based installations were used to obtain the database-sensors that determine the coordinates and sensors of physical characteristics such as humidity, temperature, and insolation. The analytical data obtained were processed using correlation and regression analysis. For this purpose, a graphical model of the dependence of air temperature changes on a global scale on the content of carbon dioxide in the atmosphere was built in Figure 2.
A close direct correlation was found, as the correlation coefficient between the independent and dependent variables is 0.9439.
The regression equation Y=-3.6378+0.0111*X, where X is the content of carbon dioxide in the atmosphere, ppm, and Y is the deviation of air temperature, °C, reliably describes these relationships, since the probability of the null hypothesis (p) is 0.0000, which is less than 0.05.
Using the method of correlation and regression analysis, mathematical models were built to predict the possible volumes of carbon dioxide sequestration depending on the size of sown areas and crop yields in the Ternopil region. In order to determine the carbon sequestration potential of the main agricultural crops in the Ternopil region, the relevant calculations were made. The data from the State Statistics Service of Ukraine were used [16] on their sown areas and yields of main products.
To verify the accuracy of the data obtained and generate predictive analytics, we used the XGBoost gradient boosting method, which is one of the most effective approaches for working with tabular data. Its application is based on the ability to take into account complex dependencies between variables, high resistance to multicollinearity, and the ability to adaptively improve forecasting results through consistent training.
The total yield of absolutely dry biomass was calculated taking into account the dry matter content of the grown products. The amount of sequestered carbon was determined based on its average content in the plant biomass of field crops - 47% [17], and the amount of sequestered CO₂ was calculated using a coefficient of 3.7.
The study found that agrocenoses of winter and spring wheat in the Ternopil region can accumulate 169-21.4 t/ha of carbon dioxide, and their total crops are 4288.3 thousand tonnes (Table 1). For winter and spring barley, these figures are 13.5 -15.1 t/ha and 294.2-363.5 thsd tonnes, respectively.
Corn is the leader in carbon dioxide sequestration among the crops grown in Ternopil region. During the growing season, it can accumulate 29.1-35.5 t/ha of CO₂ in biomass, and, taking into account the size of the harvested area, 955.4-1544.7 thsd tonnes. 17,5 18,7 16,7 20,4
Sugar beet 10,8 12,1 13,5 12,5 Sunflower seeds 10,9 11,5 11,0 11,9
5,55 6,30 5,68 7,93
Rapeseed 9,32 12,16 12,60 9,58
Potato 15,9 18,9 17,4 17,5 8,23 8,78 7,87 9,59 5,08 5,68 6,33 5,89 5,14 5,40 5,17 5,61 2,61 2,96 2,67 3,73 4,38 5,71 5,92 4,50 7,46 8,87 8,17 8,20 30,5 32,5 29,1 35,5 18,8 21,0 23,4 21,8 19,0 20,0 19,1 20,8 9,66 10,9 9,87 13,8 16,2 21,1 21,9 16,7 27,6 32,8 30,2 30,3 1185,7 1544,7 955,4 1094,7 91,5 94,8 127,8 132,5 461,7 449,3 542,7 562,2 196,0 247,6 253,7 351,7 272,6 392,5 456,6 461,4 421,2 481,4 444,4 447,8
Oilseeds - sunflower, soybean and rapeseed - have a much lower sequestration capacity, as they assimilate 19.0-20.8, 9.66-13.8 and 16.2-21.9 tonnes of carbon dioxide per hectare, respectively.
For root crops (sugar beet) and tubers (potatoes), the volumes of carbon dioxide sequestration are 18.8-23.4 and 27.6-30.3 t/ha.
According to the averaged data on sown areas and yields, the potentially possible volume of CO₂ assimilation by agrocenoses of major crops in the Ternopil region is 7303.4 thousand tonnes.
Using the method of correlation and regression analysis, mathematical models were built to predict the possible volumes of carbon dioxide sequestration depending on the size of sown areas and crop yields in the Ternopil region (Table 2).
The regression equations, where Х(1) is the size of crops sown, thousand hectares, and Х(2) is the yield, t/ha, describe the dependence of the amount of carbon dioxide absorbed on these variables.
Thus, increasing the productivity of field crop agrocenoses is an important way to reduce the CO₂ content in the atmosphere and prevent further global warming on a planetary scale.
To verify the accuracy of the data obtained and generate predictive analytics, we used the XGBoost gradient boosting method, which is one of the most effective approaches for working with tabular data. Its application is based on the ability to take into account complex dependencies between variables, high resistance to multicollinearity, and the ability to adaptively improve forecasting results through consistent training.
XGBoost's algorithm involves a step-by-step adjustment of the model by building an ensemble of decision trees, where each subsequent tree learns from the mistakes of the previous ones. The process begins with the initialisation of the base forecast, after which the deviations between the calculated and actual values are determined. Each new tree is aimed at minimising these deviations, which ensures gradual improvement of predictions.
An important step is the optimisation of the model parameters, which includes adjusting the depth of the trees, learning coefficients and applying regularisation to prevent overfitting. Thanks to the built-in feature selection mechanisms, the algorithm allows us to identify the most significant factors affecting the CO₂ sequestration process. Accuracy is assessed by crossvalidation, which ensures the model's high generalisability.
import pandas as pd import numpy as np import xgboost as xgb import matplotlib.pyplot as plt from sklearn.model_selection import train_test_split from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score # Data loading (initial set for model training) data = { "Area_thousand_hectares": [200.8, 78.5, 114.2, 22.5, 100.2, 94.4, 102.5, 54.6], "Yield_in_tonnes_per_hectare": [5.95, 4.82, 10.78, 55.4, 3.47, 3.84, 3.01, 17.8], "CO₂_sequestration_in_tonnes_per_hectare": [21.4, 15.1, 35.5, 21.8, 20.8, 13.8, 16.7, 30.3] } df = pd.DataFrame(data) # Definition of input variables (X) and target variable (y) X = df[["Area_thousand_hectares", "Yield_in_tonnes_per_hectare"]] y = df["CO₂_sequestration_in_tonnes_per_hectare"] # Splitting into training and test sets X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42) # Creating and training the XGBoost model model = xgb.XGBRegressor(objective="reg:squarederror", learning_rate=0.1, max_depth=4) model.fit(X_train, y_train) # Function for entering new data and forecasts def predict_sequestration(): print("Enter the data for the CO₂ sequestration forecast:") area = float(input("Sown area (thousand hectares): ")) yield = float(input("Yield (tonnes per hectare): ")) new_data = np.array([[area, yield]]) forecast = model.predict(new_data)[0] print(f"Projected sequestration of CO₂: {forecast:.2f} tonnes per hectare") # Run a forecast for user-entered data predict_sequestration() # Visualising the importance of features xgb.plot_importance(model) plt.show() n_estimators=100,
The application of this approach made it possible to increase the accuracy of predicting the volume of carbon dioxide assimilation by agrocenoses, taking into account the variability of yields and crop areas in different years. The results obtained allow us to predict potential changes in the ability of crops to fix CO₂ depending on climatic and agronomic factors, which is important for developing a strategy for sustainable agricultural production.
The application of information technologies in agriculture serves as a foundation for building comprehensive databases, performing in-depth analysis, and generating accurate forecasts. Artificial intelligence plays a crucial role to contribute to this issue. This study aimed to develop a methodological toolkit for assessing the potential of, and forecasting, carbon sequestration by agricultural crops using artificial intelligence. The issue of carbon sequestration in agriculture is relevant. The integration of artificial intelligence to create and process a database is particularly significant.
The analytical data collected were processed using correlation and regression analysis. The mathematical models were built to predict the potential volumes of carbon dioxide sequestration depending on the size of sown areas and crop yields in the Ternopil region. The relevant calculations were also conducted to determine the carbon sequestration potential of major agricultural crops in this region. The findings indicate that increasing the productivity of field crop agrocenoses is an effective means of reducing atmospheric CO₂ levels and mitigating further global warming on a planetary scale.
In order validate the accuracy of the data obtained and generate predictive analytics, the XGBoost gradient boosting method, which is considered to be one of the most effective approaches for working with tabular data, was used. Its application is based on the ability to take into account complex dependencies between variables, high resistance to multicollinearity, and the ability to adaptively improve forecasting results through consistent training adaptively. The application of this approach made it possible to increase the accuracy of predicting the volume of carbon dioxide assimilation by agrocenoses, taking into account the variability of yields and crop areas in different years. The resulting models enable the anticipation of potential shifts in the CO₂ fixation capacity of crops under varying climatic and agronomic factors (conditions), which are essential for the developing a strategy for sustainable agricultural production. The findings of this study provide the basis for further research into the application of artificial intelligence in carbon farming management.
The research was conducted within the framework of the project on the topic ‘Information and communication technologies for increasing productivity and involvement of human capital in the agrosphere’ (state registration number 0125U000008).
The authors have not employed any Generative AI tools. [15] S. Ahuja, P. Mehra, Sustainable Artificial Intelligence Solutions for Agricultural Efficiency and Carbon Footprint Reduction in India, Agricultural Economics and Agri-Food Business, 2023. doi: https://doi.org/10.5772/intechopen.112996. [16] Statistical collection "Crop production of Ukraine", State Statistics Service of Ukraine, 2023.
URL: https://www.ukrstat.gov.ua/druk/publicat/kat_u/2023/zb/09/zb_rosl_2022.pdf. [17] N. Ocheretnyuk, I. Voytyuk, M. Dyvak and Y. Martsenyuk, "Features of structure identification the macromodels for nonstationary fields of air pollutions from vehicles," Proceedings of International Conference on Modern Problem of Radio Engineering, Telecommunications and Computer Science, Lviv, UKraine, 2012, pp. 444-444. [18] 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Agriculture, Forestry and Other Land Use 4(2006). URL: https://www.ipccnggip.iges.or.jp/public/2006gl/vol4.html. [19] M. Dyvak, "Parameters Identification Method of Interval Discrete Dynamic Models of Air Pollution Based on Artificial Bee Colony Algorithm," 2020 10th International Conference on Advanced Computer Information Technologies (ACIT), Deggendorf, Germany, 2020, pp. 130135, doi: 10.1109/ACIT49673.2020.9208972. [20] M. Dyvak, I. Voytyuk, N. Porplytsya and A. Pukas, "Modeling the process of air pollution by harmful emissions from vehicles," 2018 14th International Conference on Advanced Trends in Radioelecrtronics, Telecommunications and Computer Engineering (TCSET), Lviv-Slavske, Ukraine, 2018, pp. 1272-1276, doi: 10.1109/TCSET.2018.8336426. [21] Herman Fesenko, Vyacheslav Kharchenko, Anatoliy Sachenko, Robert Hiromoto and Volodymyr Kochan, "An Internet of Drone-based Multi-Version Post-severe Accident Monitoring System: Structures and Reliability" in book Dependable IoT for Human and Industry - Modeling Architecting Implementation, River Publishers, pp. 197-218, 2018. [22] N. Ocheretnyuk, M. Dyvak, T. Dyvak and I. Voytyuk, "Structure identification of interval difference operator for control the production process of drywall," 2013 12th International Conference on the Experience of Designing and Application of CAD Systems in Microelectronics (CADSM), Lviv, UKraine, 2013, pp. 262-264.