Sign in Subscribe
Machine learning

High-Resolution Mapping and Impact Assessment of Forest Aboveground Carbon Stock in the Pinglu Canal Basin

Strategic Insights

— 2 min read

Accurate estimation of forest aboveground carbon stock (AGC) is critical for climate change mitigation and ecological management. Investigators from Guangxi University have developed a high-resolution AGC estimation workflow that integrates Sentinel-2, Sentinel-1, ALOS PALSAR, and SRTM data with field survey measurements for the Pinglu Canal basin. The research found that feature selection via Recursive Feature Elimination and modeling with a Random Forest algorithm optimized through hyperparameter tuning yielded high predictive accuracy under the ALL data combination, enabling the generation of a 10 m-resolution AGC map.

Key Takeaways:

  • The study developed a high-resolution AGC estimation workflow that integrates multiple remote sensing data sources for the Pinglu Canal basin, achieving an R2 value of 0.818 and RMSE of 11.126 tC/ha.
  • The total AGC in 2024 was estimated at 2.26 x 10^6 tC, highlighting the effectiveness of targeted ecological interventions in mitigating carbon loss and promoting forest recovery.
  • The research demonstrated a cost-effective, scalable method for AGC mapping using freely accessible remote sensing data and machine learning, providing insights into balancing large-scale infrastructure development with ecosystem conservation.
  • The study found that afforestation and vegetation restoration during canal construction led to higher AGC values than projected under natural conditions, highlighting the importance of targeted ecological interventions.
  • The research highlighted the effectiveness of machine learning algorithms in optimizing hyperparameters for high predictive accuracy in AGC mapping.
  • The study established a baseline scenario based on historical AGC trends (2002-2021) and climate data to evaluate human-induced changes in AGC values.
  • The research emphasized the importance of considering ecosystem conservation when developing large-scale infrastructure projects.
  • The study provided insights into the effectiveness of Random Forest algorithms in AGC mapping, facilitated by hyperparameter tuning.
  • The research demonstrated the potential of high-resolution AGC mapping for informing forest management and conservation decisions.
  • The study concluded that the developed workflow is scalable and can be applied to other regions, providing a cost-effective method for AGC mapping using machine learning and remote sensing data.

Statistics:

  • R2 value: 0.818
  • RMSE: 11.126 tC/ha
  • Total AGC in 2024: 2.26 x 10^6 tC
  • Historical AGC trends (2002-2021) data used for baseline scenario establishment

Sources:

  • High-resolution Mapping and Impact Assessment of Forest Aboveground Carbon Stock In the Pinglu Canal Basin: a Multi-sensor and Multi-model Machine Learning Approach. Forests, 2025;16(7):1130.
  • NewsRx. New Machine Learning Study Findings Have Been Reported by Investigators at Guangxi University (High-resolution Mapping and Impact Assessment of Forest Aboveground Carbon Stock In the Pinglu Canal Basin: a Multi-sensor and Multi-model Machine ...). Ecology, Environment & Conservation. August 22, 2025; p 466.

Read more