Combining LiDAR data and airborne imagery of very high resolution to improve aboveground biomass estimates in tropical dry forests

Knowledge of the spatial distribution of aboveground biomass (AGB) is crucial to guide forest conservation and management to maintain carbon stocks. LiDAR has been highly successful for this purpose, but has limited availability. Very-high resolution (<1 m) orthophotos can also be used to estimate AGB because they allow a fine distinction of forest canopy grain. We evaluated the separate and joint performance of orthophotos and LiDAR data to estimate AGB in two types of tropical dry forests in the Yucatan Peninsula. Woody plants were surveyed in twenty 0.1 ha plots in a semideciduous forest at Kaxil Kiuic Biocultural Reserve (RBKK) and 28 plots in a semievergreen forest at Felipe Carrillo Puerto (FCP). We fitted three regression models: one based on LiDAR data, another based on orthophoto variables calculated for forest canopy and canopy opening fractions, and a third model that combined both sets of variables. Variation in AGB was decomposed into LiDAR, orthophotos and joint components using variation-partitioning analyses. In FCP, regression models using LiDAR data only showed higher fit (R-2 = 0.82) than orthophoto variables only (R-2 = 0.70). In contrast, orthophotos had a slightly higher fit (R-2 = 0.91) than LiDAR (R-2 = 0.88) in RBKK, because orthophoto variables characterize very well the horizontal structure of canopies on this site. The model that combined both data sets showed a better fit (R-2 = 0.85) only in FCP, which has a more complex forest structure. The largest percentage of AGB variation (88 per cent in RBKK and 67 per cent in FCP) was explained by the joint contribution of LiDAR and orthophotos. We conclude that both LiDAR and orthophotos provide accurate estimation of AGB, but their relative performance varies with forest type and structural complexity. Combining the two sets of variables can further improve the accuracy of AGB estimation, particularly in forests with complex vegetation structure.