We are exploring changes in forest and mangrove extent and condition in Caribbean ecosystems. We use a combination of the 2020 ESA 10 m WorldCover product, the Dynamic World Near Real Time (NRT) product by Brown et al. (2022) available in Google Earth Engine from 2015-2022, and the Global Mangrove Watch polygons for 1996, 2007-2010, and 2015-2020 to define changes in forest and mangrove extent. Within forest and mangrove areas, we are examining changes in condition that are a result of a loss of leaf area or other changes in canopy structure, without the complete loss of trees (i.e., more subtle changes). We use a combination of data sources to investigate this (high resolution optical data, multi-scale C-Band SAR from Sentinel-1 (5.405 GHz center frequency) and commercial SAR (i.e., Capella GEC X-band SAR (9.65 GHz center frequency), and spaceborne lidar (ICESat-2). Our results indicate direct correlations between various height metrics from ICESat-2 ATL08, including the mean and maximum canopy height, mid-canopy height (e.g., height at 17^th percentile) and top of canopy roughness with high resolution Capella GEC X-band SAR backscatter and Sentinel-1A backscatter in VV polarization. Correlations are higher for all variables with the high-resolution Capella GEC X-Band SAR (from 6-14% higher). For an example near L’Asile, Haiti, the Pearson correlation with maximum canopy height was -0.63 for Capella GEC and -0.51 for Sentinel 1A-VV. By combining lidar metrics from these locations with high-resolution Sentinel-1 and high-resolution commercial SAR (Capella), we can impute height and other structure variables where ICESat-2 is not available.

Climate changes and increasingly frequent forest disturbances cause large-scale changes in health condition, or lead to dieback of existing forest ecosystems. Such processes result in changes of horizontal and vertical structure of forest stands (canopy cover, defoliation, degradation of branches in tree crown). GEDI (Global Ecosystem Dynamics Investigation; NASA) data has been widely used to derive products focused mainly on forest height and biomass estimation, but it also offers robust metrics describing vertical structure of forest stands. We hypothesised that GEDI vertical structure parameters can be used to detect differences in health condition of Norway spruce (Picea abies H. Karst) forest stands.

Study area covered two national parks located in the Tatra Mountains: Polish - TPN, 211.97 km² and Slovakian - TANAP, 742.84 km². Used GEDI data comprised of L2A and L2B products acquired between 2019 and 2022 in growing season (01.06-30.09). ALS LiDAR point clouds acquired in 2019 (TPN) and 2018 (TANAP) were used as a reference for 3D structure data assessment. Sentinel-2 (ESA) and PlanetScope Dove (PlanetLabs) satellite imageries were used to derive forest mask and forest health condition classes.

Analyses conducted in 5 m classes of forest height showed that Plant Area Index (PAI), Cover, and Plant Area Volume Density (PAVD) values were significantly different (Kruskal-Wallis test, 0.05 significance level) for standing dead and healthy spruce stands. Values of these metrics were lower for dead compared to healthy forest stands (i.e. mean PAI: dead=1.61; healthy=2.31 in case of 20-25 m forest height class), indicating a degradation of tree crowns and canopy. These results confirm that GEDI laser beams can detect the differences in vertical structure of forests changing their health condition, which can be utilized in modelling and monitoring. We also plan to broaden the study to explore relationships between GEDI metrics and remote sensing vegetation indices.

Ash Dieback (ADB) has been present in the UK since 2012, and is expected to kill up to 80% of UK ash trees. Recently, genetic resistance has been found in some individuals. However, detecting quantifying the extent of ADB in individual tree crowns (ITCs), crucial to understanding resilience and resistance, relies on visual assessments, impractical over large scales. New remote sensing methods, using low-cost consumer drones and structure from motion photogrammetry (SfM) have the potential to quantify fine-scale structural and spectral metrics of ITCs, indicative of ADB over large scales.

In this study we extract high resolution 3D RGB point clouds of 47 canopy ash trees taken monthly throughout the growing season at Marden Park, Surrey, a site impacted by ADB. We semi-automatically segment ITCs, extract green chromatic coordinate (GCC), and test the relationship with visual assessments of crown health. We find measurements of GCC correlate with visual assessments of crown health, but measurements taken closer to the end of growing season, when the effects of ADB on foliage are likely to be more prevalent, have a stronger relationship. We then test the relationship between within crown GCC and path length in 123 ITCs to quantify spatial patterning of dieback, finding a negative relationship in infected trees, demonstrating that crowns dieback at their extremities.

In this study, we demonstrate a new method for identifying ADB across large scales using high resolution 3D RGB data, collected using accessible instrumentation, at a time when large scale monitoring of ADB is critical. We recommend the optimum time of year for data acquisition, an important factor for detecting ADB. This method determines the severity of dieback in ITCs, with the potential to provide new insights into disease dynamics and unprecedented detail of the fine-scale structural and spectral response to ADB.

Forest disturbances in Germany show an unprecedented frequency and extent in recent years. Observed disturbances are mainly a result of drought conditions and windthrow events, which in turn have also fostered a dramatic increase in bark beetle infestations. As a result, there is a growing need to map, analyse and assess forest disturbances. A multitude of projects is currently developing systems to monitor forest disturbances in Germany. Most of these systems use frequently available optical satellite imagery (e.g. Sentinel-2) and change detection analysis to identify forest disturbances. They aim at providing timely information about where forest was disturbed. To further enhance these systems, the integration of airborne laser scanning (ALS) data could be beneficial to gain further insights, as ALS complements satellite imagery by providing high-resolution and three-dimensional information on affected forest stands.
We present basic insights into forest disturbances and corresponding monitoring efforts and provide an overview on the status of ALS data availability in Germany. We investigate the potential of various ALS-based use cases to gather additional information for forest disturbance patches detected by satellite imagery. Post-disturbance ALS data could provide detailed information regarding the disturbed forest sites, such as remaining trees or coarse-woody debris. More relevant for appropriate decision-making would be applications based on ALS data collected prior to the disturbances, as it is often readily available. This data could be used to provide information about the affected growing stock, existing skid trails or geomorphology linked to reforestation efforts.
The integration of ALS data in existing forest monitoring systems could support informed decision-making for a sustainable management of disturbed forest stands. Additionally, it could be applied to analyse disturbance patterns retrospectively.

Tropical cyclones cause widespread damage and mortality in natural forest ecosystems. Previous studies in plantations have shown that critical wind speeds depend on forest type and structure. These trends are, however, less clear for natural forests on rugged landscapes since the actual wind load experienced by trees are challenging to estimate and large datasets on forest height changes are difficult to collect. Here we present preliminary findings on forest wind susceptibility by studying wet subtropical forests in Hong Kong affected by Typhoon Mangkhut in 2018. The region is currently in a >70-year restoration trajectory and contains a mosaic of natural rainforests and exotic plantations on a rugged landscape. Changes in forest structure caused by the Typhoon Mangkhut was captured in repeated LiDAR datasets collected in years 2010, 2017, and 2020. We used computational fluid dynamics (CFD) modelling to estimate the wind load experienced by trees and compared it with LiDAR-derived canopy metrics. In particular, we would like to study (1) how plantations and natural forests of similar stature respond to extreme wind events, (2) whether prior exposure to high wind speeds affect forest resilience against extreme windstorms, (3) how does structural metrics estimated from canopy height models correlate with wind damage. By answering these questions, we hope to better understand how future forest restoration projects might be affected by the increased frequency of extreme wind events under anthropogenic climate change.