Training Field Soil Data for Annual Mapping of Soil Organic Carbon Stock (0–30 cm) in Brazil, 1985–2023 (MapBiomas Soil Collection 2, Beta Version)

10.60502/SoilData/SXCSDKMapBiomasMapBiomasTraining Field Soil Data for Annual Mapping of Soil Organic Carbon Stock (0–30 cm) in Brazil, 1985–2023 (MapBiomas Soil Collection 2, Beta Version)SoilData2024Agricultural SciencesEarth and Environmental SciencesSoil organic carbonTopsoilBrazilian biomesMapBiomasMapBiomasFerreira, LaerteLaerteFerreiraSamuel-Rosa, AlessandroAlessandroSamuel-RosaHorst, Taciara ZborowskiShimbo, JuliaJuliaShimboRosa, MarcosMarcosRosaAnjos, Marcos Alexandre dosMarcos Alexandre dosAnjosHuf dos Reis, Aline MariAline MariHuf dos ReisKerber, Débora LirielDébora LirielKerberPeruzzi, VitorVitorPeruzziSerena, Gabriele Vitória SilvaGabriele Vitória SilvaSerenaPretto, Ana CarolineAna CarolinePrettoPontes, David OliveiraDavid OliveiraPontesSilva, Luciana da LuzLuciana da LuzSilvaCardoso, Marcos Vinícius SouzaMarcos Vinícius SouzaCardosoCardoso, Maria Eduarda FerreiraMaria Eduarda FerreiraCardosoAlencastro, Mary Evelyn SantosMary Evelyn SantosAlencastroSobucki, LisianeParizzi, TalitaTalitaParizziCapoane, VivianeVivianeCapoaneSouza, DeórgiaTornquist, Carlos GustavoCarlos GustavoTornquistRocha, Anderson Sandro daAnderson Sandro daRochaOliveira, SergioSergioOliveiraMunsberg, GlaucoGlaucoMunsbergAzevedo, TassoTassoAzevedo2024-12-062024-12-062025-09-241958/2023survey data13685781653957application/pdftext/tab-separated-values3.0CC BY 4.0Collection of 12,575 georeferenced points containing field data used to train regression models for mapping soil organic carbon stock (0–30 cm) in Brazil. These data were utilized to prepare the annual maps of soil organic carbon stocks in Collection 2 of the MapBiomas Soil project, covering the period 1985–2023 (beta version). Each record includes spatial and temporal coordinates, the year of soil collection, and the soil organic carbon stock (grams per square meter) calculated for the 0–30 cm soil layer.Detailed description: Soil data were collected from 262 studies published in the Brazilian soil data repository SoilData (<a href=\"https://soildata.mapbiomas.org\">https://soildata.mapbiomas.org</a>). After analysis, a total of 250 datasets was selected for the spatial-temporal modeling of soil organic carbon stock. These data represent 15,729 points distributed across Brazil, with information on 29,881 soil layers. Some properties, such as the amount of coarse fragments and soil density, were missing in many samples. These missing values were estimated using statistical models known as pedotransfer functions. After this step, 12,575 points and 19,028 layers from 236 datasets were considered suitable for the spatial-temporal modeling of soil organic carbon stock. To calculate the soil organic carbon stock, the percentage of coarse fragments (% by weight) was converted to volume (% by volume) using a density of 2.65 g/cm³. The soil organic carbon stock (g/m²) for each layer was then calculated using the following equation: carbon content × (1 - coarse fragments) × soil density × layer thickness. For sampling points with multiple layers up to 30 cm depth, the carbon stocks from each layer were summed to obtain the total soil organic carbon stock at the point. The dataset includes the following columns:<ul><li>point_id: An alphanumeric code that identifies the source dataset ("ctb" followed by four numbers, such as ctb0003) and the ID of the soil profile, observation, or sampling point (event) in that dataset.</li><li>latitude and longitude: Decimal geographic coordinates in the WGS84 coordinate reference system (EPSG:4326).</li><li>year: The year of soil collection in the field.</li><li>soc_stock_g_m2: Soil organic carbon stock in grams per square meter (g/m²) in the 0–30 cm layer.</li></ul> The code used to process the data is available at <a href=\"https://github.com/Laboratorio-de-Pedometria/mapbiomas-soc\">https://github.com/Laboratorio-de-Pedometria/mapbiomas-soc</a>.