SUELO — AI-Powered National Soil Intelligence Platform

Overview

The platform combines Artificial Intelligence (AI), remote sensing, geospatial analytics, and field-based soil observations to deliver actionable insights into soil properties, support sustainable agricultural practices, and improve land-use planning in the context of climate change.

A major technological achievement of SUELO is the development of a national soil database and digital soil mapping framework for Algeria, producing soil property maps at a 700-meter spatial resolution. The resulting maps were evaluated and validated by soil science experts from the National Institute of Soils, Irrigation and Drainage (INSID) and the Ministry of Agriculture (MADR), supporting their scientific reliability and operational relevance.

Key Components & Technical Implementation

1. Soil Data Collection & Knowledge Extraction

• Aggregation of soil profile observations from national institutions, including INSID and BNEDER.
• Integration of additional soil observations extracted from doctoral dissertations, master’s theses, and scientific documents.
• Exploration of Generative AI techniques to automate the extraction and structuring of relevant soil information from unstructured scientific documents.
• Consolidation and harmonization of soil profile data into a unified national dataset.

2. SCORPAN-Based Geospatial Data Warehouse

• Design and development of a geospatial data warehouse based on the SCORPAN digital soil mapping framework.
• Integration of soil observations, environmental covariates, and Sentinel-2 satellite imagery.
• Preparation and standardization of spatial predictors for large-scale soil property modelling.

3. AI & Machine Learning Model Development

Design, implementation, and evaluation of multiple machine learning and deep learning architectures:

• Machine Learning: Decision Trees, Random Forest, and LightGBM.
• Deep Learning: Multilayer Perceptrons (MLP) and Multi-Task Convolutional Neural Networks (MTCNN).
• Quantum Machine Learning: Hybrid quantum-classical Multi-Task CNN architectures, exploring quantum computing approaches for soil property prediction.

4. National Digital Soil Mapping

• Training and evaluation of predictive models for multiple soil properties.
• Selection of the best-performing models for spatial prediction.
• Generation of national-scale soil property maps at 700 m resolution.
• Validation and assessment in collaboration with national soil science experts.

5. Soil Profile Clustering & Similarity Analysis

• Application of clustering techniques to identify groups of soil profiles with similar characteristics.
• Exploration of soil similarity patterns to support soil classification, characterization, and the identification of spatial relationships.

Project Impact

SUELO contributes to the development of a national digital soil intelligence infrastructure for Algeria, bridging soil science, artificial intelligence, and Earth observation technologies.

It provides a foundation for data-driven agricultural decision-making, soil resource management, and climate change adaptation.

Key Achievements

• Development of a national soil database containing over 2,200 soil profiles.
• Production of national digital soil property maps at 700 m spatial resolution.
• Integration of institutional soil observations and information extracted from academic research.
• Development and benchmarking of classical machine learning, deep learning, and hybrid quantum-classical architectures.
• Scientific evaluation and validation in collaboration with INSID and MADR experts.
• Recognition through the Innovative Project Label.