Proximal Soil Sensing in Agriculture: A Bibliometric Analysis and Narrative Review of Research from 1998 to 2025

Proximal soil sensing (PSS) provides high-density soil data for precision agriculture and soil monitoring. This study combines a bibliometric analysis with a narrative synthesis to describe how agricultural research that identifies itself as PSS developed between 1998 and 2025. Records retrieved from the Web of Science Core Collection were screened against explicit eligibility criteria, which yielded 174 studies that were analyzed with the bibliometrix R package, and the sensors and soil properties identified by dictionary matching were validated by manual coding of all abstracts. Brazil, Australia, Germany, the United States and Italy contributed the most studies. Vis–NIR spectroscopy and electromagnetic induction or apparent electrical conductivity sensors were each used in 37% of the studies, and soil organic carbon (SOC; 40%) and texture (35%) were the most frequently targeted properties, whereas soil nitrogen (9%) and salinity (4%) were rarely targeted. The main reported limitations are the effect of soil moisture on field spectra and the poor transferability of calibration models. Satellite imagery was combined with proximal data in 18% of the studies, mainly for management zones and field-scale mapping, deep learning was applied in 4%, and monetary costs or returns were reported in 5%. These results describe the literature that uses the term proximal sensing and point to the need for calibration transfer, economic evaluation and communication of sensor-derived soil information to farmers, advisors and policymakers.