Data for: Mapping the global distribution of lead and its isotopes in seawater with explainable machine learning

This repository contains the data for the article 'Mapping the global distribution of lead and its isotopes in seawater with explainable machine learning' by Olivelli et al., submitted to Earth System Science Data (https://doi.org/10.5194/essd-2025-17). The files included are as follows: Pb_climatology.nc and Pb_climatology.csv include the climatologies of Pb concentration, 206Pb/207Pb and 208Pb/207Pb in netcdf and csv format. The climatologies are obtained from the models developed and carefully described in the article and are gridded according to the World Ocean Database (WOD) 2018 grid. WOD_Pb_dataset-cleanedSOPbconc.csv is the dataset used for the development of the Pb concentration, 206Pb/207Pb and 208Pb/207Pb models. Ensemble_Pb-conc.csv, Ensemble_67Pb.csv and Ensemble_87Pb.csv contain all the predictions of the 100 model ensemble members for the Pb concentration, 206Pb/207Pb and 208Pb/207Pb models, respectively. They also include the coefficient of variations of each cell in the WOD grid calculated from the 100 individual ensemble predictions. Global-prediction_masked-df_no-coords.csv is the prediction dataset and is used to predict all values of Pb concentration, 206Pb/207Pb and 208Pb/207Pb. The associated code repository can be found at: https://github.com/OlivelliAri/Pb-ML_GEOTRACES. --- Version v4 includes smoothed fields of Pb concentration, 206Pb/207Pb, and 208Pb/207Pb in the Pb_climatology.nc file. Smoothing is applied using a 3 × 3 window (lat × lon). Version v3 includes the 'Global-prediction_masked-df_no-coords.csv' file. Version v2 includes changes in the structure of the 'Pb_climatology.nc' file to facilitate data visualization and processing.本仓库包含Olivelli等人提交至《地球系统科学数据》（Earth System Science Data）的论文《利用可解释机器学习绘制海水中铅及其同位素全球分布》（原文标题：Mapping the global distribution of lead and its isotopes in seawater with explainable machine learning）的配套数据，DOI：10.5194/essd-2025-17。 包含的文件如下： 1. Pb_climatology.nc 与 Pb_climatology.csv 分别以NetCDF（Network Common Data Form）格式和CSV（Comma-Separated Values）格式存储了铅浓度、206Pb/207Pb及208Pb/207Pb的气候态分布数据。该气候态数据由论文中开发并详细阐述的模型生成，且按照世界海洋数据库2018（World Ocean Database 2018，WOD 2018）网格完成网格化处理。 2. WOD_Pb_dataset-cleanedSOPbconc.csv 为用于构建铅浓度、206Pb/207Pb及208Pb/207Pb预测模型的数据集。 3. Ensemble_Pb-conc.csv、Ensemble_67Pb.csv 与 Ensemble_87Pb.csv 分别存储了100个模型集成成员针对铅浓度、206Pb/207Pb及208Pb/207Pb模型的全部预测结果，同时包含了基于100组独立集成预测结果计算得到的WOD网格中每个网格单元的变异系数。 4. Global-prediction_masked-df_no-coords.csv 为预测数据集，用于生成铅浓度、206Pb/207Pb及208Pb/207Pb的全量预测值。 配套代码仓库地址为：https://github.com/OlivelliAri/Pb-ML_GEOTRACES。 --- 版本更新说明： - 版本v4：在Pb_climatology.nc文件中新增了铅浓度、206Pb/207Pb及208Pb/207Pb的平滑场数据，平滑处理采用3×3（纬度×经度）窗口实现。 - 版本v3：新增了Global-prediction_masked-df_no-coords.csv文件。 - 版本v2：调整了Pb_climatology.nc文件的结构，以优化数据可视化与处理流程。