Point Cloud V2 Issue 2 (Sept 2016)

End of Summer Edition of the Point Cloud

 

The Point Cloud, AWARE’s Electronic Newsletter
Vol. 2, Issue 2. Date: 10/03/2016
aware.forestry.ubc.ca
Previous Issue

 

Welcome to the end-of-summer edition of the Point Cloud. In this edition:

 

 


 

Four new researchers joined AWARE this summer. Bastien Vandendaele, a PhD candidate, started in May and is working under Richard Fournier at Sherbrooke University on TLS and UAVs (Q24). Joseph Rakofsky, started his MSc with Nicholas Coops at UBC on growth and yield models (Q12). Chris Mulverhill also started his MSc with Nicholas Coops, studying stem class distributions (Q7).

Now that AWARE has been in operation for a year, we are seeing the results of our research. Karin van Ewijk presented her latest findings at CRSS in Winnipeg this June. In addition, three of our researchers, Kaysandra Waldron, Jean-Romain Roussel and Sean Lamb, are working on publishing their research findings in the near future. On a related note, you can read an interview with AWARE’s Benoit St-Onge in the July edition of GIM International. Finally, Shane Furze passed his PhD candidacy exam in August. Congratulations, Shane!

 

Feature Researcher

Karin van Ewijk – PhD

Karin is currently a post-doctoral fellow at Queen’s University with Dr. Paul Treitz working on Question 4 of Theme 2 within the AWARE research project. Karin studied wildlife management at Van Hall Larenstein, University of Applied Sciences in Groningen/Leeuwarden in the Netherlands where she got her first exposure to GIS and remote sensing. Working on the population dynamics of the river otter in relation to a potential reintroduction of this species in the Netherlands she realized how crucial integrating habitat information gleaned from remotely sensed data can be for a full understanding of species’ survival.

During her undergraduate studies Karin also had a chance to travel to Western Australia for eight months and work with endangered wildlife programs there. After her undergraduate degree, her interest in GIS and remote sensing resulted in obtaining a Masters in Geographic Information Science at the University of Minnesota (USA). After a brief period in Germany working as a GIS research assistant at the Johannes Gutenberg University in Mainz, she moved to Kingston, Canada where she started her PhD with Dr. Paul Treitz. Her research focus and interests during her PhD and the current post doc position are on the estimation of ecological and inventory variables (e.g. succession, species composition and diameter distribution) of forest ecosystems using remotely sensed data sources such as ALS (airborne laser scanning) and high spatial resolution optical imagery. Why and how remotely sensed data are related to forest variables is a main question in her current research project as well as the transferability of forest variable models based on remote sensing. When not working, Karin enjoys spending time with her family at the horse stable, going for a run, traveling the world, hiking and catching up with family and friends.

 

Research Snapshot

 

Identifying Tree Species Using LiDAR

Rachel Perron is a Master’s student under the supervision of Benoît St-Onge at the Université du Québec à Montréal. Her research project (question 22 of theme 4) centers on the identification of species at the single-tree level using liDAR and contextual information. The improvement of species recognition from remote sensing will contribute to producing optimal forest inventories and forecasting in growth and spatial distribution. It will also help to understand forest structure and dynamics. Rachel’s research uses LiDAR data gathered in 2012 and 2013 at J.D Irving’s Black Brook Forest in New Brunswick. The main objective is to identify the tree species over the entire territory. One sub-objective is to use this classification to locate fir regeneration within the spruce plantations. Her research uses an individual tree crown (ITC) approach, delineated from the canopy height model. Each tree has its own point cloud, from which 3D and intensity metrics are derived. Information such as plantation type, age of stand, are added as variables. A sample of trees is used to build a prediction model based on a random forest classifier, and is applied to the rest of the crowns. In the future, this method could be expanded to other areas. The identification process is summarized in figure 1.

 

Figure 1. Calibration and validation process used to build the classification model for single-tree species identification in Black Brook forest, 2016.

Preliminary results show a classification accuracy varying greatly between species (table 1). As lower accuracies seem linked to a low number of sample trees, the number used for calibration will be normalized. More sampling trees, targeting the least-represented species, will also be added to the calibration data by way of photo-interpretation.

Table 1. Confusion matrix of the classification of tree species by a random forest algorithm on tree samples from Black Brook forest, 2016.

Field Season Update

This year’s field season was very busy.  Due to last year’s late start, there was not enough time to collect field data in Alberta and Ontario.  To make up for last year’s lost field season, field teams in Ontario and Alberta put in a strong effort to collect data this summer.

Our Ontario data collection was conducted at the Petawawa Research Forest, located 180 km NW of Ottawa.  Having been studied for almost a century, the Petawawa Research Forest has a vast wealth of PSPs and data to draw from.  Our six person field team spent most of July collecting data from PSPs, sampling 57 plots in total. Read the full story

 

The PRF team: Dr Paul Treitz (Queen’s), Joe Rakofsky (McGill), Karin van Ewijk (Queen’s), Stacey Leson (UBC), Rachel Perron (UQAM) and Jean-François Prieur (Sherbrooke U./UQAM) (left to right). (Photo: Rachel Perron)


We also sent a field team of six people to Alberta to collect data at West Fraser’s FRM, located near Slave Lake.  Despite the difficult access and a particularly bad year for biting insects, we collected data from 59 sample plots, without any incidents with bears. Read the full story


We also sent a field team of six people to Alberta to collect data at West Fraser’s FRM, located near Slave Lake.  Despite the difficult access and a particularly bad year for biting insects, we collected data from 59 sample plots, without any incidents with bears. Read the full story

The Slave Lake team (left to right) – Piotr Tompalski, Yu Chen, Yuhao Lu, Chris Mulverhill, Ignacio San Miguel, Brandon Bung (Photo: Piotr Tompalski)


Because many of the projects in New Brunswick and Newfoundland are nearing completion, field data collection was relatively quiet in both these core sites this summer.   Bastien Vandendaele spent a month in Corner Brook gathering TLS data and followed that with a few weeks in Edmundston gathering data and working on a 19 ha UAV LiDAR scan.

 

LiDAR Update

Last year’s survey in Newfoundland was only 30% complete before leaf fall and bad weather forced us to postpone the rest of the survey to this summer.  To date, another 30% has been covered, with the rest of the survey expected to complete before the end of September.

In Ontario, we flew a multi-spectral LiDAR survey over Petawawa.  This was completed at the end of July, with the data delivered in August.  This is one of the first uses of multi-spectral LiDAR in Canada and we are very excited to see what the richer dataset can give us. Read the full story

Y1 AGM

AWARE’s first AGM was held in Corner Brook, NL from May 24-26th. Over the three days, there were updates on AWARE core sites, research presentations for all our active projects, guest speakers from CBPP, CFS, FP Innovations, and a key note speech from Ross Nelson, one of the pioneers of remote sensing.

We also managed to squeeze in two workshops on TLS and computational tools, a field trip to HRW, and everyone’s favourite activity, a lobster boil at the CFS field station in Pasadena. This AGM was well attended, including a contingent from CFS, government representatives from Newfoundland and New Brunswick, and industrial representatives from our partners at Tembec, FP Innovations, JD Irving and CBPP. Next year’s AGM will be held in New Brunswick. Hope to see you all there!

AWARE Post-doctoral Research Opportunity

METHODS DEVELOPMENT FOR TERRESTRIAL LIDAR APPLIED TO FOREST INVENTORY WITH EMPHASIS ON INTEGRATING A TREE ARCHITECTURAL MODEL

Location: Applied Geomatics Department, Université de Sherbrooke, Sherbrooke, Quebec, Canada
Timeframe: Up to two years starting between April 2016 and September 2016

The Terrestrial LiDAR research group at the Applied Geomatics Department of the Université de Sherbrooke is seeking to hire a post-doctoral candidate. The research will be focussed on the development of methods to estimate tree and stand structural attributes from point cloud data acquired with terrestrial scanners in forested environments. These new methods will take advantage of architectural models to describe detailed branching structure and foliage distribution in trees. The candidate should have a strong background in programming and LiDAR. The ideal candidate will:

1. be experienced with developing algorithms to deal with point clouds and able to implement them in C++ on an open source platform for use and distribution (e.g. Computree)

2. have demonstrated practical experience to process terrestrial LiDAR data

3. have knowledge of architectural models to simulate detailed tree structure

The post-doctoral candidate will work on methods to improve the current processing of terrestrial LiDAR data to simulate tree structure with the objective to publish these methods in peer-reviewed scientific journals. The post-doctoral candidate will be integrated in a team of 3 PhD and 3 MSc. students working on research projects involving the use of terrestrial LiDAR in forestry. The department offers a stimulating scientific environment for learning, with approximately 60 active graduate students and post-doctoral candidates. The research project proposed for this post-doctoral position is part of a large Canadian research initiative called AWARE (aware.forestry.ubc.ca) encompassing 25 research projects devoted to the development of methods for using LiDAR in forestry. This project is associated with a wide range of partners including the Canadian Forest Service, the Office National des Forêts in France, several forest companies in Canada and a network of university researchers worldwide, offering excellent opportunities to collaborate with leading researchers both in Canada and around the world. The Applied Geomatics Department is the largest research centre dedicated to remote sensing in Canada. It is located at the Université de Sherbrooke, a French speaking university with about 15,000 students. Knowledge of French is helpful but not mandatory for this position. Sherbrooke, a city of about 160,000 people, is a dynamic regional pole offering a rich city life and surrounded by a vast choice of outdoor activities for all seasons. The city offers a bilingual experience, as it is home to both French and English universities.

Review of applications will begin on April. 1, 2016 and continue until the position is filled. Interested parties should send their CV and two letters of reference to:

Dr. Richard Fournier
Department of Applied Geomatics (FLSH)
Université de Sherbrooke
2500 boul de l’Université, Sherbrooke (Quebec) Canada J1K 2R1
E-mail: Richard.Fournier@USherbrooke.ca
Tel.: 1-819-821-8000 ext 63209

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

You are receiving The Point Cloud because your e-mail was added to the AWARE mailing list.
You can update your preferences or unsubscribe from this list