the case of Bokypyin township, Myanmar Aung Khant Paing creator Pichdara, Lonn Chairperson Tsusaka, Takuji W. Examination Committee Ekbordin Winijkul Examination Committee Deutscher Akademischer Austauschdienst (DAAD), Germany Scholarship Donor text Pathum Thani Asian Institute of Technology 2026 continuing

print

107 leaves : ill.+e1 online resource Estimating Aboveground Carbon (AGC) in the rehabilitated mangrove forest is challenging due to high structural heterogeneity of forest. Although field-based sample plots measurement provides accurate species-level data, introduce sampling bias and limit spatial coverage. In contrast, remote sensing (RS) data offer continuous spatial coverage but lack of species-level detail ground truth, highlighting the need for integrated approaches for reliable AGC estimation. In Myanmar, severe mangrove degradation has led to the rehabilitation of mangrove forests resulting in high structural heterogeneity due to silvicultural operations. Therefore, this study evaluated AGC in rehabilitated mangrove forest of Bokpyin Township in Myanmar by the integration of field-based inventories, RS derived predictors and, develop a reliable machine learning (ML) model.The research adopted two-stage approach. Field data was collected from (42) sample plots and integrated with (14) remote sensing predictors extracted from Sentinel (1), Sentinel (2) and Shuttle Radar Topography Mission (SRTM). A Random Forest (RF) and Gradient Tree Boosting (GBT) machine learning algorithm was trained to these integrated datasets and model performance was tested by cross validation by Leave One-Out-Cross-Validation (LOOCV). The trained models were then used to generate spatially continuous AGC maps.This study produced a mean Shannon Index of (0.95± 0.42) where R. apiculata showed the highest Important Value Index of (64.8%). Additionally, the presence of B. hainesii, IUCN red list species, showed IVI of (5.0%) highlighting the high conservation value of the study area. For AGC estimation, Traditional field-based estimate revealed (10.05 ± 5.93) and RF and GBT models showed (9.27 ± 4.05) and (9.71 ± 5.32) MgCha-1 respectively. The strong alignment between these approaches demonstrates that while field-based method supports a reliable baseline, the ML models enable continuous estimation of spatial AGC. These findings will support cost-effective monitoring of large-scale carbon sequestration provide for the development of national Blue Carbon inventories for Myanmar and beyond. Future research should focus on increasing sample plot coverage and incorporating additional remote sensing data to improve prediction accuracy, while policymakers can use these methods to strengthen carbon accounting and climate mitigation strategies. A thesis submitted in patial fulfillment of the requirements for the degree of Master of Science in Natural Resources Management Thesis (M. Sc.) - Asian Institute of Technology, 2026 Carbon sequestration Myanmar Coastal forests Myanmar Mangrove ecology Myanmar Asian Institute of Technology. http://203.159.5.9/ait-thesis/detail.php?q=B24138 http://203.159.5.9/ait-thesis/detail.php?q=B24138 20260817163512.0