Digital soil mapping for precision agriculture using multitemporal Sentinel-2 images of bare ground

High-detail soil mapping is fundamental to apply precision agriculture approaches. Although, soil maps are available in many countries at a regional scale for land planning purposes, there is a need to increase the detail in the most important agricultural areas for application of site-specific agriculture practices and soil monitoring. New data from proximal and remote sensors, as well as quantitative digital methods provide the right tools to obtain these maps with sustainable costs. Digital soil mapping (DSM) includes many tools to generate spatial soil information and provides solutions for the growing demand for high-resolution soil maps worldwide.This study aims at producing soil property maps (organic carbon, clay, sand, total nitrogen, and total carbonates) at a local level (the Rieti agricultural plain, about 4000 ha), using digital soil mapping methods combining punctual soil observations, Digital Elevation Model (DEM) and related covariates (slope, topographic wetness index, etc.) and a Synthetic Soil Image (SYSI) derived by multitemporal derived bare-soil images from Sentinel-2 satellite. Regression kriging with forward stepwise regression provided reliable results for interpolation of clay, sand, soil organic carbon (SOC), whereas total carbonates $\left(\mathrm{CaCO}_{3}\right)$ and nitrogen (TN) were better interpolated by universal kriging.