[U-04] Geospatial Applications for Societal Benefits

[U04-P05] Monitoring Mangrove Change and Estimating Aboveground Biomass in Sungai Pulai, Malaysia: A Remote Sensing and Machine Learning Approach

*Mohammed Arkan Majeed¹, Siti Nur Aliaa Roslan¹, Siti Khairunniza Bejo¹, Mahirah Jahari¹, Helmi Zulhaidi Bin MohdShafri¹, Norizah Kamarudin¹, Hamdan Omar², Laili Nordin³, Bambang Trisasongko⁴, Huang Kaijin¹, Zhou Han¹, Zhao Heng Yang¹, Muhamad Afizzul Misman², Abdul Rashid Bin Mohamed Shariff¹ (1.Universiti Putra Malaysia UPM, 2.Forest Research Institute Malaysia, 3.Geoprecision Tech Sdn Bhd, 4.Bogor Agricultural University)

Keywords:Aboveground biomass (AGB) estimation, Mangrove Forests, Remote sensing, Machine learning, Alos-2 PALSAR, Landsat

Mangrove forests, such as the Sungai Pulai Mangrove Forest Reserve in Johor, Malaysia, play a crucial role in coastal protection, carbon sequestration, and biodiversity conservation. However, deforestation for coal production poses significant threats to these ecosystems. Therefore, quantifying deforestation from 1990 to 2023 using Landsat time series data and estimating aboveground biomass (AGB) using ground truth data and Alos-2 PALSAR data in 2023 using remote sensing and machine learning techniques is crucial importance for making informed decisions about management and conservation. HV polarization was chosen for machine learning training due to R² performance. Four machine learning algorithms, namely Linear Regression, Random Forest, XGBoost, and SVM, were compared based on their Root Mean Square Error (RMSE) and coefficient of determination (R²). The linear regression model showed slight superiority over other models with the lowest RMSE of 83.298 Mg ha^-1 and the highest R² of 0.481, indicating better accuracy and performance in estimating AGB compared to other ML used. The study contributes valuable insights into vegetation change over the past three decades and the application of machine learning in estimating AGB. Despite the challenges posed by the spatial resolution of free SAR data and the limited number of samples, the study successfully estimated AGB in mangroves using the remote sensing and machine learning approach. Further research is recommended to improve the accuracy of the model and explore the integration of additional data sources, thereby enhancing the sustainable management and conservation of the Sungai Pulai Mangrove Forest Reserve.