Submitted to: Smart Agricultural Technology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/16/2026
Publication Date: 7/24/2026
Citation: Ruan, G., Tian, F., Reinbott, T., Aloysius, N., Hunt, S., Zhou, J. 2026. Developing high-resolution digital soil health maps for soybean field using multiple-source sensor data. Smart Agricultural Technology. 14: Article 102408. https://doi.org/10.1016/j.atech.2026.102408.
DOI: https://doi.org/10.1016/j.atech.2026.102408

Interpretive Summary: Healthy soil is important for growing crops and improving agricultural sustainability. However, farmers and researchers still lack affordable and trustworthy methods to measure soil health. In this study, we explored whether soil health could be mapped across a field by combining soil samples with information collected from sensors and drones. A field study was conducted in 1.2-hectare soybean field over two growing seasons. We collected 69 soil samples at two depths to analyze their soil properties and describe soil health and fertility using 23 indicators. Soil apparent electrical conductivity was collected using an on-the-go sensor system. Drone images were used to develop a digital elevation model (DEM) of the field and calculate the vegetation indices of plants. Based on the acquired digital information of soil and plants, we tested four different computer mapping approaches to estimate soil health across the entire field. Our results show that the drone imagery data, together with soil data and yield could explain 62% of the variation in soil properties across the field. Soil electrical conductivity and field elevation were found to be especially important for explaining where soil health changed within the field. Overall, this study develops a potential method to map soil health within a field using sensor data. These maps can help farmers better understand soil differences, manage nutrients more efficiently, improve long term soybean productivity, and build a sustainable agricultural system.

Technical Abstract: Mapping soil health is essential for understanding the spatial and temporal variations of soil properties. However, there are no credible, verifiable, and cost-effective frameworks to quantify cropland soil health at the field level. This study aimed to test the feasibility of quantifying soil health by incorporating sensing-based information with digital soil mapping (DSM). A total of 69 georeferenced soil samples were collected at two depths (0-5 cm, and 5-15 cm) from a 1.2-ha (hectare) experimental soybean field after harvesting in two consecutive years. Twenty-three soil health and fertility indicators measured in laboratory were analyzed. Soil apparent electrical conductivity (ECa) was collected using an on-the-go ECa mapping system. Soybean grain yield maps were collected using a yield monitor system of a combine harvester. Digital elevation model (DEM) and vegetation indices were derived from Unmanned Aerial Vehicle (UAV)-based multispectral imagery. Four spatial modeling approaches, i.e., Empirical Bayesian Kriging (EBK), EBK Regression (EBKR), Geographically Weighted Regression (GWR), and a hybrid GWR-EBK, were implemented in ArcGIS pro for DSM using ECa, yield map, DEM, and Vector Image (Vis) as predictors. Model performance ranked as GWR-EBK>GWR>EBKR>EBK, with mean R² values of 0.62, 0.49, 0.40, and 0.26, respectively. Wet aggregate stability, particle size distribution, pHs, neutralizable acidity, cation exchange capacity, and manganese were the most predictable properties (R2 > 0.4 in all corresponding models). GWR coefficient analysis identified ECa-shallow and elevation as the driving factors for DSM in the most predictable indicators. This integrated DSM framework offers a promising approach for sensing-based soil health assessment, facilitating the characterization of in-field soil heterogeneity and promoting long-term soybean profitability, sustainability, and resilience to climate change.