Soil organic carbon concentration of Canada, at six soil depths, 250m spatial resolution

This dataset contains maps with the spatial distribution of soil carbon concentration (g/kg) in Canada at 6 soil depths: 0cm, 5cm, 15cm, 30cm, 60cm and 100cm. It is being made public to act as supplementary data for the publication 'Large scale mapping of soil organic carbon concentration with 3D machine learning and satellite observations', published in Geoderma: https://doi.org/10.1016/j.geoderma.2021.115402<br>The study is part of our project that aimed to produce the first wall-to-wall estimate of carbon stocks in plants and soils of Canada at 250 m spatial resolution using multisource satellite, climate and topographic data and a machine-learning algorithm. The soil organic carbon concentration maps are intermediate products to generate the soil organic carbon stock map in kg/m2.<br>To generate the soil carbon concentration maps, we used 6,533 ground soil samples, long-term climate data, multisource remote sensing data, topographic information, soil type, depth, and a 3D random forest regression model.<br>

本数据集包含加拿大范围内6个土壤深度（0cm、5cm、15cm、30cm、60cm、100cm）下的土壤碳浓度（soil carbon concentration）空间分布图谱，单位为g/kg。本数据集作为发表于《Geoderma》的论文《基于三维机器学习与卫星观测的土壤有机碳浓度（soil organic carbon concentration）大尺度制图》的补充数据公开，论文DOI：https://doi.org/10.1016/j.geoderma.2021.115402。
本研究为我团队某项目的组成部分，该项目旨在利用多源卫星、气候与地形数据及机器学习算法，以250米空间分辨率完成加拿大全境植物与土壤碳储量的首次全覆盖估算。本次公开的土壤有机碳浓度图谱为生成单位为kg/m²的土壤有机碳储量图谱的中间产物。
为生成上述土壤碳浓度图谱，本研究共使用了6533份野外土壤样本、长期气候数据、多源遥感数据、地形信息、土壤类型与深度数据，以及三维随机森林回归模型。