A global soil spectral grid based on space sensing

Abstract

Soils provide a range of essential ecosystem services for sustaining life, including climate regulation. Advanced technologies support the protection and restoration of this natural resource. We developed the first fine-resolution spectral grid of bare soils by processing a spatiotemporal satellite data cube spanning the globe. Landsat imagery provided a 30 m composite soil image using the Geospatial Soil Sensing System (GEOS3), which calculates the median of pixels from the 40-year time series (1984–2022). The map of the Earth's bare soil covers nearly 90 % of the world's drylands. The modeling resulted in 10 spectral patterns of soils worldwide. Results indicate that plant residue and unknown soil patterns are the main factors that affect soil reflectance. Elevation and the shortwave infrared (SWIR2) band show the highest importance, with 78 and 80 %, respectively, suggesting that spectral and geospatial proxies provide inference on soils. We showcase that spectral groups are associated with environmental factors (climate, land use and land cover, geology, landforms, and soil). These outcomes represent an unprecedented information source capable of unveiling nuances on global soil conditions. Information derived from reflectance data supports the modeling of several soil properties with applications in soil-geological surveying, smart agriculture, soil tillage optimization, erosion monitoring, soil health, and climate change studies. Our comprehensive spectrally-based soil grid can address global needs by informing stakeholders and supporting policy, mitigation planning, soil management strategy, and soil, food, and climate security interventions.