Dataset for Marine Plankton Species Distribution Model Analysis

This repository contains datasets and scripts for bootstrap analysis of species distribution models (SDMs) using various algorithms, including GLM, GAM, RF, BRT, and ANN. The datasets include environmental and plankton biomass data for training and testing models under present and future scenarios. The scripts allow for model performance evaluation with multiple bootstrap samples. Key components include: Sampled Darwin model data based on the mapping data of the compiled datasets of global scope (env_tax_global), Tara Ocean datasets (env_tax_tara), AMT projects (env_tax_amt). In the random datasets, each data point's location and sampling time were randomly selected. The locations were uniformly sampled across the entire ocean, while the sampling times were randomly chosen within the time span corresponding to the original global and Tara datasets. The physical model used in the Darwin simulation is the MIT General Circulation Model (MITgcm), accessible at http://mitgcm.org. The generic ecosystem code is available at https://gitlab.com/jahn/gud, and detailed equations and documentation can be found at https://darwin3.readthedocs.io/en/latest/phys_pkgs/darwin.html.本仓库包含用于基于多种算法开展物种分布模型（Species Distribution Models, SDMs）自助法分析的数据集与脚本，所涉算法包括广义线性模型（Generalized Linear Model, GLM）、广义可加模型（Generalized Additive Model, GAM）、随机森林（Random Forest, RF）、提升回归树（Boosted Regression Tree, BRT）以及人工神经网络（Artificial Neural Network, ANN）。该数据集涵盖当前及未来情景下用于模型训练与测试的环境数据与浮游生物生物量数据。配套脚本可基于多组自助采样样本开展模型性能评估。 核心组成部分如下： 采样得到的达尔文模型数据基于全球范围整合数据集（env_tax_global）、塔拉海洋（Tara Ocean）数据集（env_tax_tara）以及AMT项目（env_tax_amt）的映射数据生成。在随机数据集当中，每个数据点的位置与采样时间均为随机选取：采样点位均匀覆盖全球海洋，采样时间则在原始全球数据集与塔拉海洋数据集对应的时间跨度内随机选取。 达尔文模拟所采用的物理模型为麻省理工学院通用环流模型（MIT General Circulation Model, MITgcm），其访问地址为http://mitgcm.org。通用生态系统代码的获取地址为https://gitlab.com/jahn/gud，详细方程与文档可参阅https://darwin3.readthedocs.io/en/latest/phys_pkgs/darwin.html。