Soil Spatial Inference System of Clay and Soil Organic Carbon Content in Agricultural Lands of Queretaro Municipality, Mexico

The Municipality of Querétaro is one of the 18 municipalities that make up the state of Querétaro, Mexico. The municipality is located in the southwestern part of the state, with extreme coordinates between 20°30' and 20°56' N latitude, and 100°17' and 100°36' W longitude, covering a territorial extension of 690 km². Sampling campaigns were conducted to collect 30 soil samples from agricultural fields in the Municipality of Querétaro between August and October 2023. Sampling points were pre-selected using an algorithm to create spatially balanced points, specifically the Reversed Randomized Quadrant-Recursive Raster (RRQRR) algorithm, which establishes a correspondence between the area to be represented and the number of samples. This algorithm is used to cartographically represent a 2D space in a 1D framework, where successive samples constitute a spatially balanced sampling design. Sampling locations were chosen toward the interior of the plots to avoid edge effects, removing the top 5 cm of soil before collecting samples at a depth of 10–30 cm. The aforementioned soil database is armonized with topographic attributes (terreno.tif), satelital imagery (landsat.tif) and spectral indexes (indices_qro.tif) to generate a regression matrix. The regression matrix is used to model the soil organic carbon (%) and soil clay content (%) across agricultural (irrigated and rainfed) fields of Queretaro Municipality, in central Mexico. We provide the soil database along with the terrain and spectral datasets. Also we provide the code used to generate the models and the spatial inference of soil organic carbon and clay content. We provide the code in html files as digital notebooks to ensure reproducibility. We provide independent html files for each model, comparing two independent validation strategies (leave one out cross validation and 10 fold cross validation). File: BD4.csv - Is the database used. It contains information of 30 agricultural sampling sites in Querétaro, México. Has coordenates (X, Y), type of crop, system used (Irrigated/rainfed), wetness percentage (%W), Bulk density (D.A. g/cm3), Sand, Clay and Silt (%Arena, %Arcilla, %Limo), pH (H2O), soil organic carbón (%C), cation exchange capacity (CICT), year, database name (DB) and the identificator (ID). The samples were analyzed at the Soil Science Laboratory of the Institute of Geosciences, National Autonomous University of Mexico (UNAM), Juriquilla campus, following the methodology established in NOM-021-SEMARNAT. using the hydrometer method of Bouyoucos for soil textura analysis. Files: Emergencia_mcpo_qro.shp - Seven archives of the shapefile of the agricultura áreas in Querétaro, México. Divided in Irrigated and Rainfed. Files: landsat.tif - Raster file with de spectral images of the misión Landsat 8, from august 1, 2023 to october 31 of 2023, 30 m resolution, eigth bands ("A" "B" "G" "R" "N" "S1" "S2" "T1"). Downloaded with R Project, library "rsi" and comand "get_landsat_imagery". https://gisgeography.com/landsat-8-bands-combinations/ terreno.tif - Raster file of the topographic attributes of the sampling site. Calculated using SAGA program 9.3.1. Using: Geoprocessing - Terrain Analysis - Basic Terrain Analysis ("Analytical Hillshading" "Aspect" "Channel Network Base Level" "Channel Network Distance" "Closed Depressions" "Convergence Index_1" "demqro" "LS-Factor" "Plan Curvature" "Profile Curvature" "Relative Slope Position" "Slope" "Topographic Wetness Index" "Total Catchment Area" "Valley Depth"). https://saga-gis.sourceforge.io/saga_tool_doc/9.3.1/index.html indices_qro.tif - 138 spectral indexes. Calculated with R Project, library "rsi" and comand "calculate_indices". https://github.com/awesome-spectral-indices/awesome-spectral-indices Files: Arfrepeatedcv.html - R script of Clay (irrigated and rainfed) using Random Forest algorithm with repeated cross validation. Crfrepeatedcv.html - R script of Carbon (irrigated and rainfed) using Random Forest algorithm with repeated cross validation. Arrf.html - R script of Clay (irrigated) using Random Forest algorithm with leave one out cross validation. Ctglm.html - R script of Carbon (rainfed) using Generalized linear model algorithm with leave one out cross validation. File: Fig 1. Agri- Macrolocation of the sampling sites divides in irrigated and rainfed fields. This files support the study of Torres-Gómez et al., "Soil Spatial Inference System of Clay and Soil Organic Carbon Content in Agricultural Lands of Queretaro Municipality, Mexico" submitted in European Journal of Soil Science. March 2026. Authors acknowledge support from UNAM PAPIIT, (Grant / Award Number: 'IV200223'). Authors acknowledge support from the Instituto de Ecología del Municipio de Queretaro. Mariano Torres-Gómez is grateful for the funding of his postdoctoral stay carried out by the Postdoctoral Fellowship Program at the National Autonomous University of Mexico (UNAM-DGAPA).