As summer 2023 field season came to a close, the Silva21 community got together for another round of update meetings to catch up on project updates, latest results and ask burning questions.
This time our meetings were divided into four themes:
Silvicultural strategies for climate resilience
Disturbance and resistant-focused adaptation
Innovative technologies and data-driven adaptation
Community engagement, collaboration and policy
HQPs were divided into different themes to provide updates on their projects, share current results, discuss the implications for adaptive silviculture and ask the Silva21 community questions. In each theme, we had a member from government or industry also present an update as well as discuss implications and ask questions to the community.
On Tuesday, October 10th the Silva21 community discussed silvicultural strategies for climate resilience, hearing from seven research projects and a partner from the Ontario Ministry of Natural Resources and Forestry. Click the research title and name below to read their update.
Fine-scale structural characterization of non-stand replacing (NSR) disturbances using bitemporal aerial laser scanning (ALS) data- Tommaso Trotto, PhD candidate, UBC
Natural disturbances have a significant impact on forest structure at varying spatial and temporal scales. Forests may experience abrupt changes caused by stand-replacing disturbances like severe wildfires or more subtle and pervasive modifications due to non-stand replacing (NSR) disturbances. NSR disturbances, such as moderate insect defoliations, may affect various aspects of forest structure, including crown height, shape and cover, spatial arrangement of tree patches, and growth dynamics. The complex response of forest structure to NSR disturbances poses challenges for developing approaches that can accurately capture temporal and spatial changes for informed forest management. However, the application of active remote sensing tools, such as aerial laser scanner (ALS), may offer valuable insights into forest structural changes resulting from NSR disturbances when repeated acquisitions are available.
This study aims to investigate forest structural changes at a fine scale using a bitemporal ALS dataset acquired in the southern portion of the Lac St-Jean region, Quebec, Canada in 2014 and 2020. Change detection on ALS data was performed via a raster differencing approach, which is widely adopted for change detection tasks given its robustness in handling data acquisition inconsistencies, such as varying ALS system specifications. This approach involves regularizing 3-dimensional ALS information onto a 2-dimensional grid.
The raster differencing was conducted on a set of 14 uncorrelated ALS-derived metrics, which were rasterized to 10 m for both ALS acquisitions and subtracted over time (t2 – t1, hereafter delta metrics). Metric selection included height metrics, canopy cover, light interception metrics (e.g., leaf area index), and shape metrics derived from covariance matrix eigendecomposition. To identify regions of pixels exhibiting similar response patterns to NSR disturbances (i.e. ecological regions), a two-stage clustering approach was applied to the delta metrics. This approach involved an initial kmeans pass, followed by multivariate agglomerative clustering, resulting in the identification of 8 ecological regions. The clusters were further assessed in terms of their ecological impact on crown decoloration, defoliation, and gaps through photointerpretation of 2 aerial imagery in 2012 and 2020. Lastly, a Random Forest model was fit on the delta metrics to identify the best ALS-derived structural attributes at separating the clusters, which we deemed sensitive to detecting NSR disturbances.
We observed increasing decoloration and defoliation levels as disturbance severity increased, which was associated with a decrease in lower height percentiles, canopy cover, and LAI. Eigendecomposition metrics also showed a consistent response as disturbance severity increased, with a greater vertical variance spread at moderate disturbance severity levels, possibly indicating a greater stand vertical structural complexity. In conclusion, we detected the occurrence of NSR disturbances based on the delta metrics with an overall accuracy of 67.5% across the region.
This study revealed the importance of considering cover and height metrics for the detection of NSR disturbances in the region, derived from the Random Forest model. Furthermore, it provided the foundation for the detection and characterization of NSR disturbances in a boreal forest context, leveraging the availability of bitemporal ALS data. This research aimed to support the use of ALSderived structural metrics for the detection, characterization, and implications of NSR disturbances on forest structure for ecosystem-based management.
Tommaso Trotto
PhD Candidate
University of British Columbia
Supervisor: Nicholas Coops
tommaso.trotto@ubc.ca
Can thinning mitigate the impact of natural disturbance? - Catherine Chagnon, RA, Laval
Tree growth response to current climate change in Ontario - Emmanuelle Baby-Bouchard, RA, ULaval
Climate sensitivity mortality modeling of Québec Tree Species - Christina Howard, PhD student, UBC
Integrations of climate drivers into tree-list growth modeling in the Acadian forest region - Jamie Ring, MSc student, UBC
If you are a member of the Silva21 team and would like to receive a copy of all slides, log in to our Members area.
Forgot the password? Email amy.wotherspoon@ubc.ca.
Our next update meetings will take place in Spring 2024! Stay tuned with all news Silva21 by subscribing to our newsletter at the bottom of this page.
Comments