Multi-Sensor AI Modeling of Soils

Rapid advances in sensing technologies and artificial intelligence are transforming how we observe, understand, and manage soils. Many of soil properties once accessible only through labor-intensive laboratory analyses can now be accurately and instantly using proximal or remote sensors supported by powerful data-driven models. This Research Topic, “Multi-sensor AI Modeling of Soils,” aims to showcase innovative research that integrates multiple sensors with AI to improve soil characterization, prediction, and decision support across agricultural, environmental, and ecological domains. Multi-sensor approaches use combinations of various sensors using a single or multiple stimulus — including but not limited radioactive radiation (e.g., gamma ray, X-ray), electrical current (e.g., microwave, radio wave, radar), optical radiation (e.g., microscopes, RGB, NIR, MIR, hyperspectrum), thermal radiation, mechanical stimulus (e.g., position, movement, force, stress), acoustic stimulus, and chemical and biological stimuli.These technologies provide diverse, complementary perspectives of soil characteristics. AI models can fuse these heterogeneous data streams to create robust, scalable predictions that outperform single-sensor approaches. Such models enhance our ability to estimate soil physical, chemical, and biological properties; monitor soil health; detect constraints in current management practices for agro-ecosystem functions; and enable real‑time precision agriculture applications. A central challenge in digital soil sensing is transferring models across instruments, environments, and scales. This collection welcomes contributions on multi-sensor approaches, calibration transfer, and harmonization techniques that support deployment of sensor-based soil models in variable conditions, including field-to-lab, benchtop-to-handheld, or region-to-region transfer or across sensors. Studies leveraging multi-sensor fusion to overcome limitations of individual sensors, reduce uncertainty, or improve resilience to moisture, texture variation, or other confounding factors are particularly encouraged. The Research Topic also invites research that connects multi-sensor AI soil modeling with practical applications: precision nutrient management, soil carbon monitoring, soil health assessment, mapping of biogeochemical characteristics of soil, carbon assessment, regenerative agriculture, environmental monitoring, and digital extension tools. We welcome both theoretical advances and applied studies demonstrating how multi-sensor systems may reduce costs, increase spatial and temporal resolution, and support scalable soil information systems. Potential topics include, but are not limited to:-AI-driven fusion of proximal, remote, and laboratory sensors-Calibration transfer-Multi-sensor prediction of soil properties, functions, or processes-Uncertainty quantification in sensor-based soil models-Real-time or on-the-go sensing for precision agriculture-Multi-sensor data integration for soil mapping and monitoring-Hybrid physics informed or integrated modeling-Multi-modal datasets and benchmark frameworks (e.g., soil health indices)-Case studies using multi-sensor systems in diverse landscapes-Advances enabling low-cost, deployable soil diagnostic tools- Physics-informed AI modeling- Hybrid AI modeling of soil dynamics- Soil modeling of sensor data using machine learning and deep learning Topic editors: Katsutoshi Mizuta, Ebrahim Babaeian Source: https://www.frontiersin.org/research-topics/78583/multi-sensor-ai-modeling-of-soils