Pulse Brain · Growing Health Evidence Index
Tier 1 — Meta-analysis / systematic reviewPeer-reviewedRegenerative

What can artificial intelligence do for soil health in agriculture?

Stefan Schweng, Luca Giuliano Bernardini, Katharina Keiblinger, Hans‐Peter Kaul, Iztok Fister, Niko Lukač, Javier Del Ser, Andreas Holzinger

Computer Science Review · 2025

Read source ↗ All evidence

Summary

This systematic review synthesises the integration of artificial intelligence into soil science, evaluating diverse applications in soil health assessment, property prediction, and decision support systems for sustainable agriculture. The authors identify soil organic matter decline, compaction, and biodiversity loss as the most frequently modelled soil degradation issues, whilst highlighting substantial regional and methodological gaps—particularly for tropical, arid, and polar tundra systems. The paper provides practical guidelines on data preparation and model selection, positioning AI as transformative for soil science whilst emphasising the need for standardised, regionally distributed datasets and novel predictor–response combinations.

Regional applicability

Findings on machine learning approaches and soil property modelling are applicable to UK arable and mixed farming systems; however, the review's identification of underrepresented mild continental and temperate climates suggests UK-specific datasets and validation studies would strengthen applicability of AI-derived soil health models to British pedoclimatic conditions.

Key measures

Applications of AI algorithms (random forest, support vector machines, neural networks) for soil property prediction; soil degradation types addressed (organic matter decline, compaction, biodiversity loss); geographic and climatic representation in datasets; predictor–response combinations for soil parameter modelling

Outcomes reported

The review synthesised applications of AI in soil health assessment, predictive modelling of soil properties, and pedotransfer function development, identifying key trends in digital soil mapping, decision support systems, and regional data gaps. The study evaluated dominant machine learning approaches (random forest, support vector machines, neural networks) and identified underrepresented climatic regions and promising predictor–response combinations for future research.

Theme
Farming systems, soils & land use
Subject
Soil health assessment & monitoring
Study type
Systematic Review
Study design
Systematic review
Source type
Peer-reviewed study
Status
Published
Geography
Global
System type
Mixed farming
DOI
10.1016/j.cosrev.2025.100832
Catalogue ID
SNmojxd8i1-tf6vtk

Topic tags

Pulse AI · ask about this record

Dig deeper with Pulse AI.

Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.