Published June 30, 2026 | Version v2

Benchmarking machine learning and ensemble approaches for remote sensing–based forest aboveground carbon mapping

  • 1. ROR icon Yunnan Normal University

Description

This dataset supports the manuscript entitled “Benchmarking machine learning and ensemble approaches for remote sensing–based forest aboveground carbon mapping.”

The repository contains the processed datasets and source code used to develop and evaluate multiple machine learning models for estimating forest aboveground carbon (AGC) storage using multi-source remote sensing data.

Specifically, the repository includes:

  1. Training and validation samples
    Field-derived aboveground carbon measurements linked with corresponding remote sensing predictor variables.

  2. Remote sensing–derived feature variables
    Processed spectral indices, environmental variables, and auxiliary predictors used in model development.

  3. Model performance results
    Evaluation metrics for different machine learning models, feature selection strategies, hyperparameter optimization methods, and ensemble configurations.

  4. Predicted AGC outputs
    Model-based AGC estimation results used to generate spatial carbon maps.

  5. Code

Files

data.zip

Files (10.0 MB)

Name Size
md5:28f19615051f75cc04079286c160d45e
10.0 MB Preview Download
md5:9255a44f609883acdc27fa203ff2f10b
44.4 kB Download