Predictive layers of trace elements in soil in the conterminous United States

This dataset provides posterior mean predicted rasters of the distribution of trace elements important to vertebrate health (cobalt, copper, iron, manganese, selenium, and zinc) in the soil across the conterminous United States. Rasters were generated using a Bayesian modeling framework implemented in R with the R-INLA framework (Integrated Nested Laplace Approximation) and the Stochastic Partial Differential Equation (SPDE) approach. Geochemical data were derived from the U.S. Geological Survey survey of soils of the conterminous United States, and spatial modeling incorporated environmental covariates including soil properties (percent clay, organic matter, saturated hydraulic conductivity, pH), topography (elevation, slope), climate (temperature, precipitation), and land cover. The resulting rasters, at approximately 2450 m resolution, can be used to visualize spatial patterns, extract geochemical data for modeling, and support ecological, environmental, or agricultural analyses. The raster files represent posterior mean predicted concentrations and are provided for download.

本数据集提供了美国本土范围内土壤中对脊椎动物健康至关重要的微量元素（钴、铜、铁、锰、硒及锌）分布的后验均值预测栅格（posterior mean predicted rasters）。该批栅格通过R语言实现的贝叶斯建模框架生成，具体采用了R-INLA（集成嵌套拉普拉斯近似，Integrated Nested Laplace Approximation）框架与随机偏微分方程（Stochastic Partial Differential Equation，SPDE）方法。地球化学数据源自美国地质调查局（U.S. Geological Survey）针对美国本土土壤开展的普查，空间建模过程纳入了多类环境协变量，包括土壤属性（黏粒占比、有机质含量、饱和导水率、pH值）、地形要素（海拔、坡度）、气候因子（气温、降水量）以及土地覆盖类型。最终生成的栅格分辨率约为2450米，可用于可视化空间分布格局、提取建模所需的地球化学数据，以及支撑生态、环境或农业相关分析。本次发布的栅格文件均为后验均值预测浓度结果，可供下载使用。