Canopy Height Monitoring Data and Model of Hainan Tropical Rainforest National Park

Biomass carbon sequestration and sink capacities of tropical rainforests are crucial for addressing climate change. However, accurate canopy height estimation is necessary to determine carbon sink potential and implement effective forest management. This study compares the performance of four advanced machine learning algorithms—Random Forest (RF), Gradient Boosting Decision Tree (GBDT), Convolutional Neural Network (CNN), and Backpropagation Neural Network (BP)—in predicting forest canopy height in Hainan Tropical Rainforest National Park. The study uses 140 field survey samples and 315 unmanned aerial vehicle photogrammetry samples, along with multi-modal remote sensing datasets, including GEDI and ICESat satellite LiDAR data, Landsat imagery, and environmental information.
The following is a description of the data package:
The data in the folder (Model) includes the trained models of the four machine learning algorithms (BP, CNN, GBDT, RF).
The data in the folder (Figures) includes the technology roadmap of the research, distribution map of the study area, and data point maps used in the study, along with scatter plots of RH80, RH85, RH90, and RH95 internal and external validation for the four machine learning algorithms (BP, CNN, GBDT, RF).
The data in the folder (Datasets) includes the filtered modeling data and external validation data from GEDI and ICESat.
The data in the folder (Code) includes the training code and prediction code for four machine learning algorithms (BP, CNN, GBDT, RF).
If you require raw high-resolution raster data from other periods, please contact Dr. Qiu Zixuan at [e-mail].