Estimating longleaf pine wood volume using remotely sensed data could be an inexpensive alternative to wood volume estimates based on direct measurements, and could be useful to quickly assess damage due to wind disturbance (e.g. tornado, hurricane). However, remotely sensed wood volume estimates may suffer greater inaccuracy because longleaf pine height and crown volume growth slow when trees reach maturity while diameter growth continues. Two common approaches to estimating wood volume from remotely sensed data fit models to field-based wood volume estimates using metrics derived from LiDAR point clouds. One approach uses metrics that describe stand characteristics: mean canopy height, canopy height variance; gap fraction; and coefficient of variation of leaf area density (hereafter: stand model). The other calculates a single metric: tree canopy volume (hereafter: canopy model). We used each equation to estimate pre- and post-storm wood volume in longleaf pine dominated forests that were affected by category 4 hurricane Michael. We sought to answer 1) Are wood volumes derived from LiDAR point clouds a viable alternative to better established, field-based methods? 2) Can remotely sensed wood volume estimates be used to assess damage due to wind disturbances such as hurricanes and tornados?
Results/Conclusions
The canopy model accounted for more of the variation in measured wood volume than the stand model, and predicted a larger decrease in wood volume following hurricane Michael. Wood volume estimates from the canopy model accounted for 58% of the variation in measured wood volume and predicted a 16.0% decrease from 191 m3 ha-1 in 2018 to 159 m3 ha-1 in 2019. Wood volume estimates from the stand model accounted for 24% of the variation in measured wood volume, and predicted a 6.7% decrease in wood volume from 284 m3 ha-1 in 2018 to 265 m3 ha-1 in 2019. The canopy model may provide more timely estimates of changes in wood volume than standard field based methods, and is reasonably accurate. The canopy model may offer more flexibility as well, because the canopy metric can be derived on a tree by tree basis which would allow for different model fits by tree species. The stand model may have suffered from non-uniform spatial distribution of trees in natural longleaf pine forests, and may perform better in more uniform planted stands. As remotely sensed data becomes increasingly available, methods to use those data to assess damage to natural and commercial forests will become more common.