Mishra et al. (2025) Artificial intelligence and soil conservation: An overview

Identification

• Journal: Journal of Scientific Agriculture
• Year: 2025
• Date: 2025-11-18
• Authors: Harshit Mishra, Fredrick Kayusi
• DOI: 10.25081/jsa.2025.v9.9661

Research Groups

• Department of Agricultural Economics, College of Agriculture, Acharya Narendra Deva University of Agriculture and Technology, Kumarganj, Ayodhya, Uttar Pradesh, India
• Department of Environmental Studies, Geography and Planning, Maasai Mara University, Narok, Kenya

Short Summary

This review article synthesizes the transformative role of Artificial Intelligence (AI) in soil conservation, detailing its evolution from traditional methods to advanced data-driven solutions for soil health assessment, degradation monitoring, and fertility prediction, while also addressing associated challenges and future prospects.

Objective

• To identify the transformative role of AI in the monitoring, assessment, and management of soil health, focusing on how data-driven technologies are reshaping traditional conservation practices.
• To critically evaluate the integration of AI into policy frameworks and on-ground decision-making systems.
• To address the technical, ethical, and institutional barriers that may hinder the widespread adoption of AI in soil conservation.

Study Configuration

• Spatial Scale: Global, covering applications from site-specific field management (e.g., variable rate technology) to regional and national policy frameworks (e.g., land use planning, compliance monitoring).
• Temporal Scale: Historical, tracing the evolution of soil conservation techniques over centuries, through current real-time monitoring applications, and projecting future innovations and trends.

Methodology and Data

• Models used:
  • AI/Machine Learning: Machine Learning (ML), Deep Learning, Computer Vision, Expert Systems, Artificial Neural Networks (ANNs), Convolutional Neural Networks (CNNs), Support Vector Machines (SVMs), ensemble methods, Multivariate Adaptive Regression Splines (MARS), Gaussian Process Regression (GPR), Long Short-Term Memory (LSTM) networks, Random Forest Regression (RFR), Deep Neural Networks (DNNs), Gradient Boosting Machines (GBMs), XGBoost, k-Nearest Neighbours (kNN), Decision Trees (DTs), Recurrent Neural Networks (RNN), Bayesian networks, Support Vector Regression (SVR), Regularized Logistic Regression (RLR), Principal Component Analysis (PCA), t-SNE, Shapley Additive Explanations (SHAP), Federated Learning.
  • Traditional/Physics-based (often enhanced by AI): Universal Soil Loss Equation (USLE), Penman-Monteith equation, Modified Morgan–Morgan–Finney (MMMF) model, CENTURY Model, Richards Equation.
• Data sources:
  • Satellite imagery (e.g., Sentinel-2, Landsat 8, Planet Labs; multispectral, hyperspectral)
  • Unmanned Aerial Vehicle (UAV) imagery and sensors
  • Internet of Things (IoT) sensors (e.g., soil moisture, temperature, pH, electrical conductivity, NPK levels)
  • Ground-truth data and field observations
  • Historical climate and land-use data
  • Agronomic records (e.g., crop yield history, microbial data, previous input applications)
  • Geospatial data (e.g., GIS platforms, topographical variables)
  • Metagenomic datasets
  • Weather forecasts

Main Results

• AI integration marks a transformative phase in soil management, offering data-driven solutions for soil health assessment, degradation monitoring, and fertility prediction.
• AI-enabled models have achieved up to 92% accuracy in predicting soil organic carbon levels and 85% efficiency in mapping soil moisture patterns.
• Machine Learning models, trained on remote sensing and ground-truth data, have achieved over 90% accuracy in identifying areas at high risk of erosion and nutrient leaching.
• AI-driven Decision Support Systems (DSS) facilitate site-specific planning through Variable Rate Technology (VRT), adaptive tillage, and irrigation management, leading to a 20-25% improvement in input use efficiency.
• AI tools support policymakers with real-time dashboards and compliance tracking, enabling evidence-based policy formulation and regulatory oversight.
• AI-powered tools have improved soil fertility prediction accuracy by over 85% and AI-driven erosion models have reduced land degradation risks by nearly 30% in pilot regions (according to a 2023 FAO report).
• Traditional methods like mulching can reduce water evaporation by 25-50% and soil loss by up to 90%, while leguminous cover crops can increase nitrogen fixation by 30-60 kg/hectare.

Contributions

• Provides a comprehensive, up-to-date overview of the transformative role of AI in soil conservation, integrating traditional methods with advanced data-driven technologies.
• Systematically categorizes and explains various AI techniques (e.g., ML, Deep Learning, Computer Vision, Expert Systems) and their specific applications in soil property mapping, health assessment, and management.
• Highlights the critical role of AI in enhancing decision-making for site-specific soil management, risk assessment, and forecasting.
• Critically evaluates the integration of AI into policy frameworks and governance, including land use planning, compliance monitoring, and automated reporting.
• Discusses the environmental and ecological implications of AI use, focusing on enhancing ecosystem services (e.g., biodiversity monitoring, carbon sequestration) and mitigating negative externalities (e.g., ethical concerns, algorithmic bias, sustainability).
• Identifies key technical, infrastructural, operational, institutional, economic, and ethical challenges hindering AI adoption in soil conservation and proposes potential solutions.
• Explores future prospects and innovations, including federated learning, edge AI, digital twins, and multi-disciplinary integration, offering a roadmap for sustainable and resilient soil ecosystems.

Funding

No specific funding projects, programs, or reference codes were provided in the paper text.

Citation

@article{Mishra2025Artificial,
  author = {Mishra, Harshit and Kayusi, Fredrick},
  title = {Artificial intelligence and soil conservation: An overview},
  journal = {Journal of Scientific Agriculture},
  year = {2025},
  doi = {10.25081/jsa.2025.v9.9661},
  url = {https://doi.org/10.25081/jsa.2025.v9.9661}
}