Scholarlink

<p>This research focuses on devising an effective approach for estimating aboveground biomass (AGB) within tropical forests, specifically targeting the Daintree Rainforest in Australia. The study utilizes an array of datasets, including GEDI L4A Raster Aboveground Biomass Density, Sentinel-2 Surface Reflectance, and Copernicus GLO-30 DEM, to analyze the topographic and spectral features of the region. These features were crucial for the AGB estimation process. The research was carried out in the Google Earth Engine (GEE) platform using coding-based processing, where datasets were integrated, resampled, and filtered to ensure consistency and quality. Several machine learning models, including Random Forest (RF), Classification and Regression Trees (CART), Gradient Boosting Trees (GBT), and Minimum Distance (MD), were tested within the GEE environment. The models were trained and assessed using multiple statistical metrics such as RMSE, R², and MAE. Among the models tested, the CART model demonstrated superior performance, with an RMSE of 33.6522, R² of 0.928, and MAE of 21.744 indicating its strong potential within the broader machine learning framework for AGB estimation in this environment. Overall, this study underscores the utility of machine learning techniques implemented in GEE for enhancing the accuracy of biomass assessments in tropical forests, thereby contributing valuable insights for forest ecosystem monitoring and management.</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

Tropical forest biomass assessment using LIDAR, DEM, and Sentinel-2 in the Daintree forest

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月23日

查看全文
  • Nehir Uyar

摘要

This research focuses on devising an effective approach for estimating aboveground biomass (AGB) within tropical forests, specifically targeting the Daintree Rainforest in Australia. The study utilizes an array of datasets, including GEDI L4A Raster Aboveground Biomass Density, Sentinel-2 Surface Reflectance, and Copernicus GLO-30 DEM, to analyze the topographic and spectral features of the region. These features were crucial for the AGB estimation process. The research was carried out in the Google Earth Engine (GEE) platform using coding-based processing, where datasets were integrated, resampled, and filtered to ensure consistency and quality. Several machine learning models, including Random Forest (RF), Classification and Regression Trees (CART), Gradient Boosting Trees (GBT), and Minimum Distance (MD), were tested within the GEE environment. The models were trained and assessed using multiple statistical metrics such as RMSE, R², and MAE. Among the models tested, the CART model demonstrated superior performance, with an RMSE of 33.6522, R² of 0.928, and MAE of 21.744 indicating its strong potential within the broader machine learning framework for AGB estimation in this environment. Overall, this study underscores the utility of machine learning techniques implemented in GEE for enhancing the accuracy of biomass assessments in tropical forests, thereby contributing valuable insights for forest ecosystem monitoring and management.