Stenberg, Bo
- Department of Soil and Environment, Swedish University of Agricultural Sciences
Abstract
This review provides an overview of the accuracy of soil property predictions using the most common proximal soil sensing (PSS) techniques in precision agriculture (PA), both standalone and in combination with one another or with environmental covariates. Based on 114 scientific papers, the accuracy of soil property estimates was evaluated by calculating the normalized root mean square error (NRMSE) using root mean square error (RMSE) values and the range of the predicted soil property. Soil properties, PSS techniques, covariate types, and the type of model employed for predictions were the factors around which accuracy results were sorted. Additionally, we estimated PSS service costs based on both the literature and on a market study with questionnaires for private companies operating in the PA sector. Our literature analysis indicates that diffuse reflectance spectroscopy (DRS) was able to estimate the greatest number of soil properties with a high accuracy compared to the other PSS techniques. The most popular applications of DRS are to determine soil organic matter, nutrients, and soil texture, although most of the applications are primarily lab-based. X-ray fluorescence (XRF) is the second most popular technique for soil property estimation; however, in contrast to DRS, most estimations are in-field applications with portable XRF sensors. The use of XRF is widespread in determining elemental concentrations. Onthe-go techniques such as electromagnetic induction (EMI) or gamma-ray spectroscopy (gamma-ray) accounted for lower accuracy compared to point-based techniques (e.g., DRS, XRF, time-domain reflectometry). However, they are widely used by companies, as they have vast potential to delineate PA management zones in the field, and are suitable for on-the-go mapping of soil properties such as mineralogy, texture, salinity, water content, cation exchange capacity, and soil depth. The combined use of PSS techniques generally doesn't outperform the singular application, although the number of samples collected for calibration, and specific combinations of sensors, covariates, and modeling techniques, combined correctly, may enhance the predictions of soil properties using PSS techniques applied singularly. However, these outcomes tend to depend on local site characteristics. Differences were found between the analysis of costs collected from the literature and from the companies' survey. The estimated cost of surveying a hectare with PSS oscillates between 15.5/ha and 130/ha, according to research data, whereas our company survey resulted in an interval between 142 and 362/ha. Price variability was influenced by personnel costs, fieldwork, data and reporting, sample analysis, and equipment. Besides, increases in the final prices can be attributed to accessibility and difficulties related to field work, as well as traveling to the area of interest. This review aims to serve as a reference for encouraging the adoption of current and available sensing technologies by farmers, policymakers, and companies by providing helpful insights into the suitability of different PSS techniques for mapping various soil properties, their associated costs, and what is available in the market. We foresee that availing PSS services will become cheaper with technological advances. Thus, it will become a standard approach in the future, as it is the most feasible way for producing high-resolution maps and affordable soil property information.
Keywords
Electromagnetic sensors; Covariates; Data fusion; Precision agriculture mapping; Cost analysis; Soil; Sensing; Proximal
Published in
Computers and Electronics in Agriculture
2026, volume: 243, article number: 111378
SLU Authors
UKÄ Subject classification
Soil Science
Publication identifier
- DOI: https://doi.org/10.1016/j.compag.2025.111378
Permanent link to this page (URI)
https://res.slu.se/id/publ/145531