World Soil Predominant Texture 0-100cm

This layer uses sand, silt, and clay most likely values from soilgrids.org to create texture classes. Soilgrids.org sand, silt, and clay datasets are integers that give a weight in grams in each particle class. The weight we are converting directly into percent, for example soilgrids value of 500g of sand means 50% sand ((500g/1kg) * 100 = 50%). A 100cm depth was chosen because it matches many of the world's most important crops' rooting depths. A 0 to 60cm version of this is also available. Variable mapped: Predominant USDA texture class as derived from predicted percent sand, silt, and clay. Data Projection: Goode's Homolosine (land) WKID 54052 Mosaic Projection: Goode's Homolosine (land) WKID 54052 Extent: World, except Antarctica Cell Size: 250 m Source Type: Thematic Visible Scale: All scales are visible Source: SoilGrids.org Publication Date: June 14, 2021 NOTE: This layer uses the USDA texture classification system with international soil datasets, which use different particle size definitions than the USDA. Very little silt shows up in this layer, this could be a reason why. To determine the predominant soil texture we first classified texture for the following layer depths: 0-5cm 5-15cm 15-30cm 30-60cm 60-100cm Then we used focal statistics with the majority option to find the majority texture class of each pixel from the five layers, weighted as follows: 0-5cm * 1 5-15cm * 2 15-30cm * 3 30-60cm * 6 60-100cm * 7 (not 8, something had to break the tie and I reduced the multiplier by 1 to break ties, thinking of all soil depths the depth from 95-100cm may be the least significant in the stack overall.) ----------------------------------------------------------------- Raster functions were created to classify sand, silt, and clay using the following statements in raster calculator: Sand Con((( Silt + ( 1.5 * Clay )) < 150 ), 1, 0) Loamy Sand Con(((Silt + (1.5 * Clay)) >= 150) & ((Silt + (2 * Clay)) < 300),2, 0) Sandy Loam Con(((Clay >=70)&(Clay<200)&(Sand>520)&((Silt + (2 * Clay)) >= 300))|((Clay<70)&(Silt<500)&((Silt + (2 * Clay)) >= 300)),4, 0) Loam Con(((Clay>=70) & (Clay<270) & (Silt>=280) & (Silt<500) & (Sand<=520)),8 ,0) Silt Loam Con((((Silt>=500) & (Clay>=120) & (Clay<270)) | ((Silt>=500) & (Silt<800) & (Clay<120))),16 , 0) Silt Con(((Silt >= 800)&(Clay<120)),32 ,0) Sandy Clay Loam Con(((Clay>=200) & (Clay < 350) & (Silt < 280) & (Sand > 450)),64 ,0) Clay Loam Con(((Clay >= 270) & (Clay<400) & (Sand > 200) & (Sand <= 450)), 128, 0) Silty Clay Loam Con(((Clay >= 270) & (Clay < 400) & (Sand <= 200)),256 ,0) Sandy Clay Con(((Clay >= 350) & (Sand > 450)) ,512 , 0) Silty Clay Con(((Clay >= 400) & (Silt >= 400)), 1024, 0) Clay Con(((Clay>=400) & (Sand <= 450) & (Silt < 400)) , 2048 , 0 ) These conditionals were used on the "mean" soilgrids.org rasters for silt, sand, and clay on rasters representing the following depths: 0-5 cm below the land surface 5-15cm below the land surface 15-30cm below the land surface 30-60cm below the land surface 60-100cm below the land surface The conditionals were just summed together to create check rasters for each depth. All analysis was done in soilgrids.org own Goode's Homolosine projection (land) in ArcGIS Pro. The data were served in this same projection in ArcGIS Image for ArcGIS Online. --------------------------------------------------------------------------------------------------- At first, the classes were given a value of 1, 2, 4, 8, 16, 32 and so on, then were added together. This is so we could see if some classes were overlapping others. We continued to troubleshoot the above definitions until there were no overlaps and as few values of 0 as possible. Once the overlaps and misses were fixed, the dataset was reclassed into values of 1-13. An attribute table was built to drive popups and a legend. Read more in our blogs: Dsicoveries and Revelations: Exploring the World Soils 250m Layers Getting Started with Multidimensional Soil Layers in the ArcGIS Living Atlas of the World本图层采用soilgrids.org提供的砂粒、粉粒和粘粒估算值，生成土壤质地类别。soilgrids.org提供的砂粒、粉粒和粘粒数据集为整数值，代表各粒级的质量（单位：克）。我们将该质量直接转换为占比：例如soilgrids中砂粒含量为500g时，对应砂粒占比50%（计算方式为(500g/1kg)×100=50%）。本次采用100cm的土层深度，因其契合全球多数重要农作物的根系分布深度；同时也提供0至60cm深度的数据集版本。 映射变量：由预测的砂粒、粉粒、粘粒占比推导得出的美国农业部（USDA, United States Department of Agriculture）标准优势土壤质地类别。 数据投影：古德Homolosine（Goode's Homolosine）（陆地）投影，WKID 54052；镶嵌投影：古德Homolosine（Goode's Homolosine）（陆地）投影，WKID 54052 覆盖范围：全球（南极洲除外） 像元尺寸：250米 源数据类型：专题数据 显示比例尺：全比例尺可见 数据源：SoilGrids.org 发布日期：2021年6月14日 注：本图层采用美国农业部（USDA）土壤质地分类体系，但配套使用的国际土壤数据集的粒级定义与USDA标准存在差异；本图层中粉粒占比极低，或为上述差异导致。 为确定优势土壤质地，我们首先对以下土层深度分别进行质地分类：0-5cm、5-15cm、15-30cm、30-60cm、60-100cm；随后采用焦点统计的多数投票法，从上述5个图层中提取每个像元的优势质地类别，并按如下规则加权：0-5cm层权重为1，5-15cm层为2，15-30cm层为3，30-60cm层为6，60-100cm层为7（原计划权重为8，为避免平局将权重减1，考虑到所有土层中95-100cm深度的土壤对整体质地的影响相对最低）。 本数据集通过栅格计算器中的以下条件语句，构建栅格函数以完成砂粒、粉粒、粘粒的质地分类： 砂土：Con(((粉粒 + 1.5×粘粒) < 150), 1, 0) 壤质砂土：Con(((粉粒 + 1.5×粘粒) ≥ 150) 且 ((粉粒 + 2×粘粒) < 300), 2, 0) 砂质壤土：Con((((粘粒 ≥70) 且 (粘粒<200) 且 (砂粒>520) 且 ((粉粒 + 2×粘粒) ≥300)) 或 ((粘粒<70) 且 (粉粒<500) 且 ((粉粒 + 2×粘粒) ≥300))), 4, 0) 壤土：Con(((粘粒≥70) 且 (粘粒<270) 且 (粉粒≥280) 且 (粉粒<500) 且 (砂粒≤520)), 8, 0) 粉砂壤土：Con((((粉粒≥500) 且 (粘粒≥120) 且 (粘粒<270)) 或 ((粉粒≥500) 且 (粉粒<800) 且 (粘粒<120))), 16, 0) 粉砂土：Con(((粉粒≥800) 且 (粘粒<120)), 32, 0) 砂质粘壤土：Con(((粘粒≥200) 且 (粘粒<350) 且 (粉粒<280) 且 (砂粒>450)), 64, 0) 粘壤土：Con(((粘粒≥270) 且 (粘粒<400) 且 (砂粒>200) 且 (砂粒≤450)), 128, 0) 粉砂粘壤土：Con(((粘粒≥270) 且 (粘粒<400) 且 (砂粒≤200)), 256, 0) 砂质粘土：Con(((粘粒≥350) 且 (砂粒>450)), 512, 0) 粉砂粘土：Con(((粘粒≥400) 且 (粉粒≥400)), 1024, 0) 粘土：Con(((粘粒≥400) 且 (砂粒≤450) 且 (粉粒<400)), 2048, 0) 上述条件函数分别应用于各土层深度（地表下0-5cm、5-15cm、15-30cm、30-60cm、60-100cm）对应的soilgrids.org砂粒、粉粒、粘粒平均栅格数据；将各条件函数的计算结果求和，即可生成各深度的校验栅格。所有分析均在ArcGIS Pro中基于soilgrids.org自带的古德Homolosine（Goode's Homolosine）（陆地）投影完成，数据通过ArcGIS Online的ArcGIS Image服务以相同投影发布。 最初，各质地类别被赋予1、2、4、8、16、32等数值后求和，以便排查类别间的重叠问题。我们持续调试上述分类规则，直至无类别重叠且0值占比尽可能低。修正重叠与缺失问题后，将数据集重分类为1至13的数值范围，并构建属性表以支持弹窗显示与图例生成。更多详情可参阅官方博客：Dsicoveries and Revelations: Exploring the World Soils 250m Layers、Getting Started with Multidimensional Soil Layers in the ArcGIS Living Atlas of the World