LiDAR data transects: a sampling strategy to estimate aboveground biomass in forest areas

The estimation and mapping of aboveground biomass over large areas can be done using the remote sensing tools. The objective of this study was to estimate the aboveground biomass of two types of tropical forest: semi-evergreen (SETF) and semi-deciduous tropical forest (SDTF) in the Yucatan Peninsula, Mexico, using metrics obtained from LiDAR (Light Detection and Ranging) data. Data from 365 plots of the National Forest and Soils Inventory of Mexico were used to calibrate aboveground biomass models using multiple linear regression and Random Forest. These models were used for mapping aboveground biomass along LiDAR strips. The transformed regression model explained the variance by 62% (RMSE = 41.44 Mg ha(-1) for SETF & 36.60 Mg ha(-1) for SDTF) for both types of vegetation. The models of Random Forest explained the variance by 57% (RMSE = 40.73 Mg ha(-1)) for SETF and only 52% (RMSE = 35.10 Mg ha(-1)) by SDTF. The mismatch between the field data and LiDAR data, as well as the error in the precision of the coordinates of the inventory plots, were recognized as factors that influenced on the results. Despite the above, the estimates obtained could serve as a basis to estimate the complete biomass inventory in the study area by incorporating spectral data derived from a remote sensor that covers the entire area.