2026-10-06T15:28:59Z https://api.figshare.com/v2/oai

oai:figshare.com:article/30686447 2026-09-30T11:23:02Z category_27580 item_type_3 month_year_09_2026

Code and Data Samples for Yan'an Carbon Storage Research Yeming Lao (22646096) Environmental geography carbon storage XAI <p dir="ltr">This repository provides a complete, fully reproducible workflow for all analyses, models, and figures presented in the manuscript.</p><p dir="ltr">All scripts use relative paths, require no manual code modification, and each section corresponds directly to the structure of the paper.</p><p dir="ltr">The reproduction workflow is organized into four major sections:</p><p dir="ltr">Section 1 — Remote Sensing Image Segmentation and Vegetation Mapping</p><p dir="ltr">This section trains an XGBoost-based classifier using Sentinel-2 data, performs batch prediction on tiled images, and produces vegetation distribution maps for 12 species. The workflow includes model training, tile prediction, and mosaicking into final vegetation maps.</p><p dir="ltr">Section 2 — Carbon Storage Calculation in Vegetation and LULC</p><p dir="ltr">Using the vegetation maps from Section 1 and species-level carbon density parameters, this section computes per-species carbon storage.</p><p dir="ltr">Additional LULC rasters are incorporated to estimate total carbon storage across the study area.</p><p dir="ltr">Section 3 — Factors Data Processing</p><p dir="ltr">Climatic, environmental, and land-surface variables are processed into a unified 100-m grid dataset.</p><p dir="ltr">Processes include NC-to-TIFF conversion, quarterly aggregation, carbon grid generation, factor extraction, and creation of a cleaned machine-learning–ready table.</p><p dir="ltr">Section 4 — Regression Modeling, SHAP Explainability, and Mapping</p><p dir="ltr">Random Forest, LightGBM, and XGBoost models are trained and evaluated.</p><p dir="ltr">SHAP values are computed to quantify feature contributions, transformed into spatial rasters, visualized, and analyzed to reveal spatial mechanisms and boundary effects.</p><p dir="ltr">Section 5 — Figures Drawing and Mapping</p><p dir="ltr">SHAP summary and dependence plots, SHAP maps and corresponding analysis figures are plotted and mapped in this section.</p><p dir="ltr"><br></p><p dir="ltr">Usage</p><p dir="ltr">Each section includes:</p><p dir="ltr">1. Required input data (provided in Data/ subfolders)</p><p dir="ltr">2. A corresponding Jupyter notebook (1–4)</p><p dir="ltr">3. Step-by-step procedures</p><p dir="ltr">4. Output directories for all intermediate and final results</p><p dir="ltr">Running the notebooks sequentially will reproduce all tables, figures, maps, and metrics presented in the manuscript.</p> 2026-09-30T11:23:02Z Dataset Dataset 10.6084/m9.figshare.30686447.v1 https://figshare.com/articles/dataset/Code_and_Data_Samples_for_Yan_an_Carbon_Storage_Research/30686447 CC BY 4.0