Sign in Subscribe
Machine learning

AI Researchers Develop Innovative Framework for Estimating Tree Carbon Sequestration

Strategic Insights

— 1 min read

Researchers at Hong Kong Polytechnic University have proposed an innovative framework for estimating tree above-ground biomass, a crucial step towards achieving carbon neutrality. The existing global allometric models face challenges in accurately modeling local trees' biomass. The new framework involves collecting local tree felling data, implementing Light Detection and Ranging (LiDAR), and developing a machine learning-based allometric model. The researchers used a LiDAR backpack to collect point-cloud models of 100 felled trees in Hong Kong, encompassing 31 tree species and 17 tree families. The study's results showed that the best-performing allometric model, developed by XGBoost, achieved an accuracy of R2 = 0.82, mean absolute percentage error (MAPE) = 40.70%, and mean absolute error (MAE) = 214.37 kg.

Key Takeaways:

  • The innovative framework proposed by Hong Kong Polytechnic University researchers involves collecting local tree felling data, implementing LiDAR, and developing a machine learning-based allometric model.
  • The researchers collected point-cloud models of 100 felled trees in Hong Kong, encompassing 31 tree species and 17 tree families.
  • The LiDAR backpack used in the study collected point-cloud data from the felled trees.
  • The study's results showed that the best-performing allometric model, developed by XGBoost, achieved an accuracy of R2 = 0.82.
  • The mean absolute percentage error (MAPE) was 40.70%, and the mean absolute error (MAE) was 214.37 kg.

Statistics:

  • 100 trees were felled in Hong Kong from March 2023 to April 2024.
  • 31 tree species and 17 tree families were included in the study.
  • The LiDAR backpack captured point-cloud data from each of the 100 felled trees.
  • The best-performing allometric model achieved an accuracy of R2 = 0.82.
  • The mean absolute percentage error (MAPE) was 40.70%, and the mean absolute error (MAE) was 214.37 kg.

Sources:

  • Individual tree above-ground biomass estimation by integrating LiDAR and machine learning. Trees, Forests and People, 2025, 21():100955.
  • A free version of this journal article is available at https://doi-org.sdpl.idm.oclc.org/10.1016/j.tfp.2025.100955.

Read more