Point Cloud V5 Issue 2 (January 2020)

The Point Cloud, AWARE’s Electronic Newsletter
Vol. 5, Issue 2. Date: 01/07/2020
Previous Issue
Not displaying correctly?

View this email in your browser

Feature Researcher

Jean-François Prieur is a PhD candidate at Université de Sherbrooke under Richard Fournier (Applied Geomatics) and Daniel Kneeshaw (Biology, UQAM). He completed his undergraduate B. Comm. (accounting) degree at Concordia University, Montréal, in 1993. After working 15 years as an IT consultant in the fields of investment banking and ecommerce infrastructure, he changed career paths and completed a Graduate Diploma (GIS) and a M.Sc. Geography (GIS) at Université du Québec à Montréal under Benoît St-Onge. His M.Sc. research dealt with correcting and improving SRTM digital elevation models in order to better measure canopy heights across large areas of forest. Jean-François is working on Question 21 of the cross cutting theme for AWARE which deals with species identification at the individual tree level.

The forest is an important global strategic resource from both an environmental and economic view point. It provides forestry products (and the jobs related to them) as well as contributing to natural environment biodiversity. More importantly, the forest is the biggest terrestrial carbon sink on the planet: 80% of above-ground biomass and 40% of below-ground biomass is contained within its boundaries. Remote sensing has become ubiquitous in the generation of forest inventories. Within the large range of technologies available to practitioners, airborne laser scanning (ALS) is particularly well adapted to precision forestry as it provides detailed structural information (given the laser pulses capacity to penetrate closed canopy) as well as spectral information related to the target (intensity).

However, forest structural information cannot be fully exploited if species information is not known. Precise species identification is an important parameter for forest inventory, for the quantification and surveillance of biodiversity as well as the study of forest ecosystems and habitats. Precise species information is the information that is the most requested by the forestry industry and government organizations in the elaboration of forest inventories.

He has been working for the past several years on operational deployment of these species’ classification methods in order to accurately survey the strengths and limitations of current approaches. His thesis will include a comparison of these current methods with three different sensor types, the contribution of deep learning in species identification with sparse training data as well as the potential to apply models to more than one site.

Outside of his PhD project, JF and his wife Annie enjoy travelling (especially to Japan for the last few years) as well as gardening (hard in the Québec winter!) and taking care of our home.

Research Snapshot

Q17b: What landscape level ecosystem goods and service indicators can be developed with airborne LiDAR data?

Catherine Frizzle is a PhD candidate under Richard Fournier (Applied Geomatics) and Mélanie Trudel (Hydrology) at University of Sherbrooke. Her research project is focusing on LiDAR derived ecological function indicators useful to quantify water related ecosystem services at the landscape level. Her work is using framework developed in Q17a by Aurélie Schmidt.

The Soil and water assessment tool (SWAT) hydrological model was calibrated for Harry River Watershed, Newfoundland. A reference scenario with the SWAT model was produced with a land use without harvest. In this case, harvest fraction of hydrologic response units (HRU) in each sub watershed was set to 0 and assigned to forest status. Then, the sensitivity analysis of the SWAT outputs to harvest was tested by changing fraction of corresponding HRU from forest to harvest according to harvest scenario provided by Corner Brook Pulp & Paper Ltd. The contribution in sediment yield (in t/ha/yr) were calibrated with hydrometric values to identify SWAT parameter suited for this watershed. Differences in sediment yield between the reference scenario and the one with harvesting increased up to a factor of 7. The variables of importance to explain sediment yield variability from the SWAT model were determined using partial least square regression with morphometric and land use pattern. Many of those variables can be produced from LiDAR data. What we have learned so far is that landscape metrics, such as fraction of harvest area in each sub-watershed, stand out as a strong predictor. This variable can be precisely monitored with airborne LiDAR data and used as a landscape indicator. Fig 1. uses the airborne LiDAR data acquired in 2016 to show different canopy height profile according to the harvest year: 25 regions of 11,28 m radius (400 m2) were selected for each year demonstrating the gradual regeneration of the stand. The plot on the right side of Fig. 1 gives the 95 percentile (p95) of LiDAR returns calculated on raster patches (20 x 20 m). The p95 can be used as a landscape indicator that allows separating forest patches from harvested patches at various years with a set threshold value.

LiDAR data provide useful spatial information specifically on topographic variables and forests attributes. Digital terrain model allows preparation of topographic variables useful in hydrological studies, namely: slope, channel length, watershed delineation, and topographic wetness index. From LiDAR point clouds, forests metrics can also inform us on forest cover and height, which informs on forest capacity to intercept precipitation and limit runoff and erosion. Metrics extracted from LiDAR data are particularly well suited to monitor stand- and patch-level attributes, however, there is a need to adapt these metrics for landscape-level indicators required to quantify water related ecosystem services.

Figure 1: George lake located in the Harry River Watershed (NL) provisioning cabin owners. Harvest activities are visible in the foreground.

Figure 2: (a) Canopy height profile and (b) 95 percentile (p95) of height in meters for different harvest year.

e-Lecture Series

The AWARE e-lecture series was a great success. The e-lectures ran Wednesdays beginning Sept. 25 to Oct. 30, 2019 at the Canadian Institute of Forestry (CIF) website. The series, generally titled An Update on Canadian Forestry Applications of LIDAR and Digital Photogrammetry, had the following weekly topics:

• The AWARE Project: Results and Outcomes

• Digital Soil Mapping in New Brunswick

• New LIDAR Technologies on the Horizon – SPL and Multi-Spectral LIDAR

• Assessing Non-Timber Values Using LIDAR and Advanced Remote Sensing Data

• Digital Photogrammetric Applications to Enhanced Forest Inventory

Below is the chart of the participants as plotted by CIF:

Downloadable copies of slides and audio recording of the lectures are available at the CIF website as well as in the AWARE website.

Copyright © 2016 University of British Columbia, All rights reserved.





This email was sent to *|EMAIL|*

why did I get this?    unsubscribe from this list    update subscription preferences