A soil satellite spectral service (Sat4): A strategy for stakeholders

Soil spectral libraries (SSLs) are commonly built with expensive proximal sensors, while satellite spectral data are free and widely available. This study investigates how service providers can construct a satellite-based SSL to deliver soil information to stakeholders. We propose the Soil Satellite Spectral Service (Sat4-Service) strategy to estimate clay content and soil organic carbon (SOC) from farm to national scale in Brazil. Bare-soil reflectance from Landsat-8 and Sentinel-2 was compared with laboratory spectra from 7711 topsoil samples, showing strong correspondence between lab spectra resampled to the satellite bandpasses and the SySI composites, with a mean Pearson correlation across all multispectral bands and both sensors of r ≈ 0.72. We then simulated a provider progressively building a satellite-derived SSL as new clients and soil data were added. With each farm, the dataset expanded and predictive models were retrained. Model accuracy varied initially but stabilized after ~50 clients, reaching R 2 = 0.63 for SOC and 0.79 for clay using both satellites. In contrast, a national model based on >60,000 samples performed worse on the same farms (R 2 = 0.43 for SOC, 0.65 for clay). Findings show that multispectral imagery can reliably estimate soil clay and carbon, and that models built on regional or farm-level data often outperform broad continental-scale libraries. While global models provide useful first approximations—especially for newcomers or data-scarce areas—regionally focused spectral libraries ensure higher accuracy. The Sat4-Service framework highlights the potential of satellite data to deliver cost-effective, scalable soil information, with applications in precision agriculture, carbon crediting, and digital soil mapping.