Characterizing local forest structural complexity based on multi-platform and-sensor derived indicators

Global climate change, biodiversity decline, and increasing disturbances are challenging the health and resilience of forests. In this regard, forest managers have sought to promote enhanced structural complexity (ESC) which utilizes the positive correlation between structural complexity and biodiversity or resilience to inform management practices. In light of these concerns, we integrated remote sensing data from multiple platforms and sensors to test the potential for quantifying different levels of structural complexity in temperate forests. This analysis was conducted in the context of the BETA-FOR project, where silvicultural manipulations of forest structure replicate silvicultural or natural disturbances. BETA-FOR includes a wide- range of standardized treatments across representative Central European broad-leaved forests, which are sub-divided into aggregated (gap felling) and distributed treatments (selective thinning) in combination with varying deadwood structures. This study provides a novel analysis of ESC from complementary remote sensing perspectives in order to bridge scales among structural complexity indicators. Remotely sensed observations comprise in-situ measurements (mobile and terrestrial laser scanning), as well as spaceborne observations from various sensors (including Sentinel-1 radar, Sentinel-2 multispectral, and GEDI lidar). We found moderate to strong inter-platform correlations among structural complexity metrics (|r|>= 0.6) between mobile laser scanning (box dimension, canopy cover), terrestrial laser scanning (canopy openness index), Sentinel-1 (VH, cross-polarized backscatter), Sentinel-2 (NMDI, Normalized Multi-band Drought Index), and GEDI (total canopy cover). In addition, multivariate analyses revealed that ESC of gap aggregated treatments can be effectively delineated from control and distributed treatments across all considered remote sensing sensors/platforms. Therefore, the metrics from different platforms and sensors better characterize the changes in structural complexity through aggregated compared to distributed treatments. Furthermore, we identified the sensitivity of in-situ and spaceborne metrics towards the presence of standing deadwood structures. An unsupervised clustering analysis highlights distinct differences in structural complexity of aggregated treatments with snags and habitat trees compared with aggregated treatments without standing structures, as well as distributed and control treatments. Findings demonstrate the potential of various sensors and platforms for monitoring forest structural complexity. We recommend the spaceborne indicators Sentinel-1 VH cv, Sentinel-2 NMDI cv, and GEDI cover cv to monitor ESC at high spatio-temporal resolution as they show highest correlations to in-situ metrics, thus holding the potential to guide adaptive forest management.