Evaluating and comparing the performance of ICESat-2 and GEDI data for estimating aboveground biomass in mangroves

The launch of the ICESat-2 and GEDI spaceborne LiDAR missions has provided new opportunities for large-scale estimation of mangrove aboveground biomass density (AGBD); however, their footprint-scale performance remains unclear. This study evaluated and compared the performance of AGBD estimation models based on ICESat-2 and GEDI data from mangrove ecosystems. First, we examine the effects of different feature categories on model accuracy. Next, we compared the performance of Random Forest, eXtreme Gradient Boosting (XGBoost), and Multilayer Perceptron in constructing the AGBD estimation frameworks. Finally, we assessed the estimation accuracies of the ICESat-2 and GEDI under different observation scenarios, including day/night and strong/weak beams. The results indicate that the combination of height, canopy coverage, and topographic features improves the ICESat-2 AGBD estimates, whereas height and topographic features enhance the GEDI estimates. Among all algorithms, XGBoost consistently achieved the highest accuracy (ICESat-2: R² = 0.71; GEDI: R² = 0.73). Under nighttime strong-beam conditions, the ICESat-2 and GEDI both achieved high accuracies (ICESat-2: R² = 0.80; GEDI: R² = 0.76). Considering the denser sampling of the GEDI, large-scale AGBD mapping is recommended. Overall, this study highlighted the potential of the ICESat-2and GEDI for mangrove biomass monitoring.