COS 242-3 - CANCELLED - Detecting small-scale forest disturbance with LiDAR derived ecosystem morphological traits

Background/Question/Methods

Ecosystem morphological traits (EMTs) are an ecosystem agnostic framework of three variables which can be derived from a range of remote sensing methods, and which can comprehensively describe structural characteristics of a diverse range of ecosystems. Many LiDAR-based methods for detecting forest disturbance currently use a host of statistically selected variables without biological links to the characteristics of the ecosystem. The literature indicates that many authors already use a proxy for one or more of vegetation height, vegetation cover, and vertical structural complexity when working LiDAR data, the EMTs. As such, we set out to identify the most suitable LiDAR-derived metrics to use as proxies for each of the EMTs and to demonstrate how each EMT can be used directly to identify small scale disturbance without the need for modelling.

Results/Conclusions

We show that, with a multitemporal dataset from a tropical forest, it is possible to use EMTs to not only identify disturbances caused by logging (p < 0.01) in the period between data collection but also identify regions of regrowth from prior logging (p < 0.01). This was done through observation of both the temporal dynamics of the metrics being assessed over 8 years but also through maps of change of metrics. Comparing these results, we were able to identify that LiDAR metrics that were most suitable as proxies for each EMT. For vegetation height, results showed that the top-of-canopy height, calculated from a canopy height model, was more sensitive to disturbance than high percentiles or the average of raw LiDAR return height distributions. For vegetation cover, it was found that fractional cover calculations with lower height thresholds were more sensitive to disturbance and the regeneration in the understory. For structural complexity in the vertical profile, the Gini coefficient was found to outperform foliage height diversity and standard deviations of LiDAR return heights at the detecting of disturbance and regeneration in forest stands.