Year 4 – Semester 4

Reporting  (Reporting, Publications and Presentations section)

Aware Year 4 – Semester 4

AWARE Winter Semester Summary

The NSERC Annual Report, which was submitted on time, reported great progress in each projects and increased in-kind contribution from the industrial partners.  The improved involvement of our industrial partners can potentially claw-back in part or in full, the NSERC withheld installment of $93K.

Project Reporting

Active Projects: Q1, Q2, Q3, Q5, Q7, Q9, Q14, Q15, Q17b, Q18, Q19, Q20, Q21, Q23, Q24, Q25

The AWARE research question Q20 will split into 3 – Q20a, Q20b and Q20c

Completed Projects  Q4,Q6,Q8,Q10,Q11, Q12, Q13, Q14, Q17a, Q22


Project Updates

Q1 –  Tristan Goodbody started working on this question beginning March 2019 as a postdoc.  He initialized inventory analysis on carbon modelling project with CFS Newfoundland. To determine the influence of individual tree metrics for improving area-based forest inventory estimates, Tristan initialized an area-based/individual tree modelling amalgamation.

Q2 – Kaysanda is currently on maternity leave with expected project return date of May 2019.

Q3 – Melodie’s work on Q3 is finished. She presented the results during the workshop held in Corner Brook in December 2018. The tools were provided to Corner Brook Pulp and paper and the province. Melody is currently polishing her thesis after being reviewed.

Q5 – Karin continued working on transferability and testing of FRI variable models between the eastern and western boreal forests.  She uses two statistical approaches (lm and rf) to establish a methodology using loocv validation to diminish the small sample size issue at Slave Lake. Initial results for LHT, BA, QMD and VOL datasets have been sent to all her co-authors. Karin had presented in the Emerging Research Seminar Series at Queen’s University in January 24, 2019. Her peer-reviewed paper was accepted in Forests.

Q7 – Chris completed data analysis from ground-based photogrammetric point clouds, the data he gathered from his fieldwork at the Slave Lake site. He began writing a paper regarding this analysis. His other paper, titled The utility of terrestrial photogrammetry for assessment of tree volume and taper in boreal mixedwood forests, is currently under review.

Q9 – Ayla is working on SDA analysis and interpolation of soil moisture regime. She completed an SDA sensitivity analysis and worked to calibrate the model based on both the Damman Guide Moisture Regime and the Gravimetric Soil Moisture of each plot.  Ayla also ran the SDA and Kriging based depth to water interpolation on 194 plots and had identified plots with potential interpolation errors

Q15a – Shawn’s work on evaluation of supervised random forests and support vector machine (SVM) classifiers for generating defoliation maps is currently underway. Shawn was unable to obtain a pan-sharpened EO-1 imagery from the Zhang lab and may have to remove it from his analysis and manuscript. Draft of his initial EO-1 manuscript is under review.

Q16 – Xavier completed classification of Stem Diameter Distribution (SDD) and selection of the classification technique. He is currently working on creating a predictive model that separate SDD by their general shape using canopy height model textural metrics and point cloud metrics. Xavier will also assess the importance and significance of the textural metrics, test the different Weibull fitting technics to extract the unimodal SDD shape parameters and select the best metric.

Q17a – Aurelie completed the preliminary version of her scientific paper based on this question. She is working on her MSc thesis that will be submitted for review in May 2019.

Q17b – Catherine selected relevant function indicators in actual scenarios by running tests to compare results between swat model outputs and Lidar FI.  She is also working on defining future scenarios for precipitation, harvesting, road building and best practices.

Q18 – Sam has completed the first half of data analysis for this question.  He also wrote the first draft of his paper and participated in UBC Forestry’s three-minute thesis competition.

Q19 – Van-Tho carried out several tests and fixed bugs to take L-Vox algorithm to operational level. The algorithm is now fully developed and being used for a scientific manuscript and in other projects.

Q20a – Charles has established sites to test methods which provided preliminary results. Several aerial photographs were acquired from these selected sites. Charles was also able to produce data output of canopy height model (CHM) using agisoft photoscan software, had analyzed errors associated with the outputs as well as performed required adjustments for CHM.  He wrote a discussion around these preliminary results with suggestions for ways to enhance model quality.

Q20b – Regeneration of stems after harvesting (TBH Fall 2019)

Q20c – The effects of wind on stems (TBH Fall 2019)

Q21 – To produce surfaces from point clouds over larges areas, Jean-Francois developed additional Python scripts. He adapted the phyton scripts for urban classification in Montreal, making them more robust for easier distribution.  He also performed preliminary classification comparisons by using new single photon lidar data at Petawawa and regular ALS.

Q23 –Parvez collaborated with JDI on tree species classification of the entire Black Brook forest. He is working on the manuscripts of the Black Brook forest species ID as well as feature standardization approaches in the three study areas – York regional forest, Petawawa research forest and Black Brook forest.

Q24 – Bastien developed a semi-automatic method to isolate tree crowns from stems, from UAV based lidar in deciduous forest to extract crown attributes. His manuscript on the use of deciduous dataset of the McCoy Brook site in New Brunswick is in progress. Bastien participated in an internship program at the University of Liege in Belgium from December 2018 to April 2019.

Q25 – Piotr continue to work on integrating forest growth models in Ontario (MIST) with the ALS data, which is a continuation of previous studies in BC and Alberta.  He has gathered all required growth functions and implemented the MIST growth model in R.  The MISTR package was used to generate growth curve database which is required for the curve matching approach.  The R package implementation is published and available in github


Student Forum

AWARE student forum no. 11 was held virtually by skype on April 23, 2019. Five HQPs presented their research results for questions 7, 18, 19, 20a and 23.