Comparison of ensemble ML methods for mangrove AGB prediction
Description
This dataset supports a comparative analysis of machine-learning models for predicting mangrove aboveground biomass (AGB) across a large area of interest covering major mangrove ecosystems along the southern coastline of Lampung Province, Indonesia. It integrates field-measured AGB with Sentinel-2 Level-2A surface reflectance imagery to develop predictive models for mangrove ecosystems. A harmonized modelling framework was applied to evaluate ensemble and non-ensemble regression algorithms using identical input variables and parameter settings, including Random Forest (RF), Gradient Boosting Regressor (GBR), XGBoost (XGB), LightGBM (LGBM), Support Vector Regression (SVR), and a multilayer perceptron neural network (MLP). Multi-resolution inputs are provided to assess scale-dependent model performance and trade-offs between prediction accuracy and model complexity.