Aboveground biomass estimation in a subtropical forest using airborne hyperspectral data

Accurate estimation of forest biomass is critical in the study of global carbon balance and climate change. This research was undertaken in the Yushan forest, in southeast china. We used metrics extracted from hyperspectral data as predictor variables to establish three types of biomass prediction models, and then the models are verified by cross-validation method. Overall, all of the three types of models were well predicted, the broad-leaved forest model was relatively high (Adj-R
2
=0.68), followed by coniferous forest (Adj-R
2
=0.65), and the mixed forest was relatively low (Adj-R
2
=0.63). It was also confirming that the vegetation indices (i.e., NDVI, NDVI_RE and GNDVI) were important in predicting biomass in subtropical forests. This method can be used to estimate the forest biomass in subtropical forest effectively.