Lievin Nkuba Kasanda
Master
Automated detection of the pre-existing road network in the boreal forest by remote sensing
Université du Québec en Abitibi-Témiscamingue
Director: Osvaldo Valeria
Co-director: Richard Fournier
For more information on my research, please see the following links...
E-mail :
E-mail :
Description of the current research project
Problem
Forests cover almost 4 billion hectares of the world's land surface, or around 31% of the land area (FAO, 2010; Torabzadeh et al., 2014). The resulting forest products represent more than $ 600 billion, or almost 1% of world GDP (Ivers, 2018). In Abitibi-Témiscamingue, thousands of users take advantage of the road network in place to harvest woody material (Dallaire, 2013). Roads are a key factor for sustainable forest management, in fact, more detailed information on the location and morphological characteristics of forest roads (currently fragmented) would allow managers to significantly improve forest planning. The data available on the roads presents inconsistencies due to the lack of detailed spatial information, the insufficiency of measurement variables reflecting the reality on the ground (Baltsavias et al., 2004) and the lack of appropriate methods for processing spatial data. from several different sources (Dallaire, 2013). This is the case in Quebec, where the Routard road database exists, but has significant limitations. Routard was developed in 2013 from a collection of scattered information collected from different forest stakeholders. Routard is produced from several different data structures, which implies that the spatial data of the territory's coverage are subject to inaccuracies (Dallaire, 2013). These limits restrict the use of current forest road network databases with, in particular, lack of data on geometric information and geolocation. However, geomatics tools with the rise of remote sensing, image processing, GPS and geographic information systems (GIS), offer new possibilities to better identify roads. To date, this potential remains relatively untapped. It is therefore essential to use new remote sensing technologies, which have demonstrated their performance in the classification of elements of the earth's surface.
Objective
The general objective of this project is to develop an automated or semi-automated detection method to identify the presence and characterize forest roads in the boreal forest with a special focus on abandoned roads. Abandoned roads are any road that has not been used for a very long time in forest activity and sometimes not completed with other road segments. This study specifically targets forest roads in the boreal environment in Quebec. It makes a significant contribution to the documentation of road network mapping in that the method used will make it possible to identify and characterize the forest road by its physical characteristics described above. In conclusion, we want to acquire an automatic means of mapping the forest road network in Quebec's boreal forest in order to characterize them in support of road capital management.
Research themes
- Forest management
- Machine learning
- Object-oriented classification
- Geomatics
- Airborne Lidar
- Remote sensing
- Digital image processing
Partners
Natural Sciences and Engineering Research Council of Canada (NSERC), the NSERC-UQAT-UQAM Industrial Chair in Sustainable Forest Management (AFD Chair), Produits forestiers Résolu, RYAM Gestion Forestière, RÉBEC and Barrette-Chapais.
Scholarships
- 2018 - NSERC/UQAT/UQAM Industrial Chair Scholarship in Sustainable Forest Management (AFD Chair
)
- 2015 - Scholarship of the Agence Universitaire de la Francophonie for Asia Pacific (AUF )
Academic Posters and Presentations
Nkuba-Kasanda, L.; Valeria, O. & Fournier R. Automated detection of pre-existing forest road networks in the boreal forest using airborne LiDAR and Sentinel-2. Colloquium presentation 13th colloquium of the Centre d'étude de la forêt, Université du Québec à Chicoutimi, Chicoutimi, Québec, May 1-3, 2019.
Nkuba-Kasanda, L.; Valeria, O. & Fournier R. Spatial tools for managing access to territories in a context of global changes in the boreal forest. Poster presentation Carrefour forêts 2019, Centre des congrès de Québec, Quebec City, Quebec, April 2-4, 2019.
Nkuba-Kasanda, L.; Valeria, O. & Fournier R. Automated detection of the pre-existing forest road network in the boreal forest using airborne LiDAR. Poster presentation 20th Colloquium of the AFD Chair. Université du Québec en Abitibi-Témiscamingue, Lorrainville, Québec, 2018.
Work experience
- 2018 - 2020: Master's Degree in Ecology at the Université du Québec en Abitibi-Témiscamingue, Rouyn-Noranda (Canada) under the supervision of Osvaldo Valeria and Richard Fournier, with the research topic: "Automated detection of the pre-existing road network in the boreal forest by remote sensing".
- 2015 - 2017: Master's degree in Computer Science in Intelligent Systems and Data Analysis at the National University of Vietnam \ Institut de la Francophonie pour l'Innovation (Hanoi Vietnam). Double graduation of a Master's degree in Computer Science in Digital Content Engineering in the Enterprise, at the University of La Rochelle (France). The research topic is "Data mining algorithms applied to digital soil mapping: implementation and testing of robust strategies for estimating prediction uncertainties", under the supervision of Philippe Lagacherie and Cécile Gomez.
- 2007 - 2012: Bachelor's degree in Computer Science in Computer Engineering at the University of Kinshasa (Democratic Republic of Congo).
- 2000 - 2007: Secondary education (college) at the General and Technical Secondary School of Kanshi / MIBA in Industrial Electronics (Democratic Republic of Congo).
Scientific publications
Professional training and certification
- First aid certificate in the workplace (2019-2022)
- Training for work in the natural environment
- Truck and ATV driving training
- Valid learner's permit (class 5)
- Training Certificate on Data Design and Administration, organized by Efrem MBAKI LUZAYISU, PhD in Engineering Sciences with the support of Louvain Interaction Laboratory (LILab) at the University of Kinshasa, Democratic Republic of Congo (July 2014)
- Certificates (1st and 2nd training workshops) on strategies and techniques for searching patent information in the PATENTSCOPE database, by the World Intellectual Property Organization (WIPO) in collaboration with the Ministry of Industry, Democratic Republic of Congo (April and August 2011)
Volunteering
- Technoscience Abitibi-Témiscamingue, a charitable organization working to promote science and technology among young people December 2019 to present
- University of Kinshasa, assistant professor in charge of practical work, supervision and follow-up of the completion of end-of-studies work with students from 2014 to 2016.
