Exploratory Evaluation of UAV-Guided and Conventional Soil Sampling Strategies in a Semi-Arid Agroecosystem

The transition toward Agriculture 4.0 requires soil sampling strategies that really characterize spatial heterogeneity. This study compared conventional five-point composite (5P) and zigzag (ZZ) sampling methods with multispectral image-guided sampling (MI) using a DJI Mavic 3 unmanned aerial vehicle (UAV) (SZ DJI Technology Co., Ltd.), Shenzhen, Guangdong, China) in a semi-arid agroecosystem of northeastern Mexico. In situ measurements included penetration resistance (PR), volumetric water content (VWC), and apparent electrical conductivity (ECa), while laboratory analyses included bulk density (ρb) and soil pH (H2O). Statistical analyses revealed significant differences among sampling strategies for penetration resistance and volumetric water content (p < 0.05). The MI approach captured greater spatial contrasts in penetration resistance within the 10–20 cm depth interval, with PR values reaching up to 5076 kPa, whereas conventional methods recorded substantially lower PR values. UAV-guided sampling identified sampling locations associated with contrasting VWC and PR measurements relative to conventional composite sampling strategies. Soil pH (H2O) and ECa remained stable among sampling methods; however, highlighting the need for methodological standardization and local calibration. Multispectral image-guided sampling facilitates spatial stratification and the characterization of within-field variability in dynamic soil physical properties, supporting stratified soil management and more efficient resource use in semi-arid agricultural systems.