Scholarlink

<p>Forests are essential for biodiversity conservation, climate change, natural education, scientific research, and carbon sequestration. This study uses machine learning-based Random Forest (RF) regression to estimate the Above Ground Biomass (AGB) of the Ozark and Ouachita forests at a 10-meter resolution by combining data from Sentinel-2, Sentinel-1, and GEDI (Global Ecosystem Dynamics Investigation) on Google Earth Engine. The RF model included 34 out of 154 variables representing topographical, spectral, and textural factors demonstrating strong correlations with measured biomass. The RF model showed strong performance with R-squared and RMSE values of 0.95 and 18.46 for the training dataset and 0.75 and 34.52 for the validation dataset, respectively. The primary predictors were elevation, vegetation indices such as Leaf Area Index (LAI), Normalized Vegetation Index (NDVI), and Enhanced Vegetation Index (EVI), as well as forest height measurements like RH100, RH98, and RH95. The study also extrapolates the historical biomass from 2015 to 2023 using Landsat 8 data using the image normalization technique. The model effectively identified spatial patterns in biomass, with mean extrapolated values ranging from approximately 100 Mg/ha to 200 Mg/ha from 2015 to 2023, underscoring the value of open-source cloud platforms and integrating topographical, spectral, and textural data for accurate regional AGB estimation. The results support precise estimation of fire-related emissions and strategic planning to improve forest health and sustainability, contributing significantly to biodiversity conservation and carbon sequestration efforts.</p> Graphical Abstract <p>This study explores an innovative workflow to estimate the Above Ground Biomass (AGB) of forests, focusing on the Ozark and Ouachita regions. This research utilizes open-source data and platforms like Google Earth Engine and integrates satellite-based radar, optical data, and GEDI lidar measurements to capture the diverse characteristics of forests. The process involves satellite data fusion and employing machine learning techniques to analyze various factors such as elevation, vegetation indices, and forest height metrics to capture the different dynamics of the forest to estimate AGB accurately. The study also applies an interpolation technique to reveal historical trends, offering insights into the spatial patterns of biomass distribution over time. The study provides a transferable framework that can be adapted to estimate biomass in other regions around the globe, empowering efforts toward sustainable forestry practices and environmental health. This approach combines advanced remote sensing technology with machine learning to support biodiversity, sustainable forest management, and carbon sequestration.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning and Multisensor Data Fusion for Forest above Ground Biomass Estimation in Arkansas

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月11日

查看全文
  • Abdullah Al Saim,
  • Mohamed H. Aly

摘要

Forests are essential for biodiversity conservation, climate change, natural education, scientific research, and carbon sequestration. This study uses machine learning-based Random Forest (RF) regression to estimate the Above Ground Biomass (AGB) of the Ozark and Ouachita forests at a 10-meter resolution by combining data from Sentinel-2, Sentinel-1, and GEDI (Global Ecosystem Dynamics Investigation) on Google Earth Engine. The RF model included 34 out of 154 variables representing topographical, spectral, and textural factors demonstrating strong correlations with measured biomass. The RF model showed strong performance with R-squared and RMSE values of 0.95 and 18.46 for the training dataset and 0.75 and 34.52 for the validation dataset, respectively. The primary predictors were elevation, vegetation indices such as Leaf Area Index (LAI), Normalized Vegetation Index (NDVI), and Enhanced Vegetation Index (EVI), as well as forest height measurements like RH100, RH98, and RH95. The study also extrapolates the historical biomass from 2015 to 2023 using Landsat 8 data using the image normalization technique. The model effectively identified spatial patterns in biomass, with mean extrapolated values ranging from approximately 100 Mg/ha to 200 Mg/ha from 2015 to 2023, underscoring the value of open-source cloud platforms and integrating topographical, spectral, and textural data for accurate regional AGB estimation. The results support precise estimation of fire-related emissions and strategic planning to improve forest health and sustainability, contributing significantly to biodiversity conservation and carbon sequestration efforts.

Graphical Abstract

This study explores an innovative workflow to estimate the Above Ground Biomass (AGB) of forests, focusing on the Ozark and Ouachita regions. This research utilizes open-source data and platforms like Google Earth Engine and integrates satellite-based radar, optical data, and GEDI lidar measurements to capture the diverse characteristics of forests. The process involves satellite data fusion and employing machine learning techniques to analyze various factors such as elevation, vegetation indices, and forest height metrics to capture the different dynamics of the forest to estimate AGB accurately. The study also applies an interpolation technique to reveal historical trends, offering insights into the spatial patterns of biomass distribution over time. The study provides a transferable framework that can be adapted to estimate biomass in other regions around the globe, empowering efforts toward sustainable forestry practices and environmental health. This approach combines advanced remote sensing technology with machine learning to support biodiversity, sustainable forest management, and carbon sequestration.