<?xml version='1.0' encoding='UTF-8'?><metadata xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:dcterms="http://purl.org/dc/terms/" xmlns="http://dublincore.org/documents/dcmi-terms/"><dcterms:title>Training Field Soil Data for Mapping of Soil Particle Size Distribution (0-100 cm) in Brazil (MapBiomas Soil Collection 3, Beta Version)</dcterms:title><dcterms:identifier>https://doi.org/10.60502/SoilData/OXSR2N</dcterms:identifier><dcterms:creator>MapBiomas</dcterms:creator><dcterms:publisher>SoilData</dcterms:publisher><dcterms:issued>2025-11-29</dcterms:issued><dcterms:modified>2026-04-02T02:00:00Z</dcterms:modified><dcterms:description>A collection of 15,798 georeferenced points with data from 60,883 soil layers, which were used to train regression models for mapping soil particle size distribution (0-100 cm) in Brazil. These data supported the development of maps of clay, silt, and sand content and depth to layers with >50% and >90% volume of coarse fraction in Collection 3 of the MapBiomas Soil project (beta version). Each record includes spatial coordinates, the midpoint depth of the soil layer, clay, silt, sand, and coarse fraction content, and three additive log-ratios.</dcterms:description><dcterms:subject>Agricultural Sciences</dcterms:subject><dcterms:subject>Earth and Environmental Sciences</dcterms:subject><dcterms:date>2025-11-29</dcterms:date><dcterms:contributor>Samuel-Rosa, Alessandro</dcterms:contributor><dcterms:dateSubmitted>2025-11-29</dcterms:dateSubmitted><dcterms:spatial>Brazil</dcterms:spatial><dcterms:license>CC BY 4.0</dcterms:license></metadata>