High-resolution mapping of soil mechanical properties by integrating geophysical sensors and terrain attributes
Year: 2026
Type: article
Abstract: Soil mechanical properties are essential for evaluating its strength and stability when subjected to stress from farm machinery. Conventional methods for measuring these properties are often slow, labour-intensive, and destructive. This study presents a novel approach that utilizes on-the-go proximal soil sensors (PSS), specifically electromagnetic induction (EMI) and gamma-ray spectroscopy (GRS), alongside terrain attributes, to evaluate their effectiveness in estimating three key soil mechanical properties: (1) precompression stress (σ pc ), (2) compression index (C C ), and (3) strain at 100 kPa (Ɛ 100kPa ). For this, a geophysical survey and soil sampling were conducted across three arable fields, yielding 129 bulk soil samples and 516 intact soil cores (100 cm 3 ) collected from 69 sampling points at two depths (0.15 and 0.40 m). Uniaxial confined compression tests (UCCT) were conducted to measure these mechanical properties, with multiple linear regression (MLR) applied for their estimation and cross-validation with the leave-one-out method (LOOCV). Both site-specific and unified datasets were analyzed, revealing higher prediction accuracy at 0.15 m compared to 0.40 m depth. Among the examined soil mechanical properties, C C was estimated most accurately, followed by σ pc and Ɛ 100kPa . Estimates of σ pc derived from on-the-go PSS combined with terrain attributes substantially outperformed those obtained from the existing pedotransfer function. Furthermore, digital maps of these properties were successfully generated to visualize their spatial variability at the field scale. This study shows that on-the-go PSS provide a rapid, field-scale and non-destructive framework for estimating soil mechanical properties, supporting improved soil compaction assessment and monitoring. • On-the-go PSS and terrain attributes (TA) estimated soil mechanical properties. • Compression index (C C ) was predicted with the highest accuracy. • Precompression stress (σ pc ) was the second most accurately estimated property. • PSS and TA based σ pc estimates outperformed the pedotransfer function (PTF). • Digital maps were created to show spatial variability of mechanical properties.
Source: Soil and Tillage Research
Institution: Aarhus University
Language: English
Subfield: Environmental Engineering
Field: Environmental Science
Domain: Physical Sciences
SDG: Life on land
Open Access status: hybrid
Estimated APC paid: $4,570
Award: G3241098762
OpenAlex ID: W7118160400