Internship : Using AI to evaluate the impact from cascading hazards – fires and floods

Canadian Center for Mapping and Earth Observation (CCMEO)

Number of internship offered : 1

Internship duration : to be discussed (12-16 weeks)

Presentation of the host organization :

As the national mapping agency, Natural Resources Canada’s Canadian Centre for Mapping and Earth Observation (CCMEO) is responsible for, among other things, the creation, management, publication and maintenance of a wide range of geospatial data covering the Canadian landmass. Through the GeoBase division, CCMEO is responsible for the maintenance of fundamental geospatial data, such as the national hydrographic network. This data is of critical importance to the country as it provides the basis for the creation of a multitude of value-added geospatial data (for example, the country’s statistical data is referenced from the fundamental data). To be relevant, this data must be continually updated. Technological developments in recent years have enabled the emergence of new approaches, such as deep learning (DL), that provide near-term solutions for updating geospatial data quickly and efficiently. For about three years, the GeoBase Division has established a research and development team to develop artificial intelligence (AI) based tools for extracting map objects from remote sensing data (optical imagery, radar, and elevation data).  

Project description :

Costly weather events are becoming commonplace in Canada. In the past ~10years alone, there have been significant storms, floods, and fires which have devastated communities and resulted in billions of dollars of damage. We know that after a fire the there is an increased risk of flooding, specifically flash flooding, and debris flows.

For this project, we are looking for someone to explore the cascading effects from fires to flood using AI. We intend to use Fort McMurray as a case study site as a major wildfire swept through the community in 2016.

Impact :

We hope the results of this work can help with the National Flood Hazard Identification and Mapping Program. (FHIMP), provide information to make better investment decisions, and a greater understanding of the impact from cascading events.

Evaluation :

The evaluation of the internship will be based on the quality of the literature review, the quality of the implementation of the method and the rigor of the tests performed.  

Project feasability :

The objectives are very achievable. The intern will also have access to various experts in the field, who can help him/her in his/her tasks. The defined objectives can be revised during the internship, depending on the progress.  

Data :

The data includes nationally available datasets from platform, including spatial and temporal record of fires and floods, climate data from The student with work with existing training sites, explore the possibility of additional training data. 

Computing resources :

The intern will have access to a laptop computer to access software and resources. For more intensive testing, the intern will have access to a high-performance infrastructure and a cloud infrastructure, both with GPUs. Everything will be accessible remotely. 

Intern’s main responsability :

The objective of the internship is 1) to conduct a short literature review on similar methods and their potential and 2) to implement and test one of these methods in the R programming language, using the caret library.  

More specifically, the tasks will be as follows: 

  • Become familiar with the available data inventory; 
  • Conduct a literature review on fires and the impacts to floodings; 
  • Develop training/test data;
  • Prepare a test plan; 
  • Perform the tests on the high performance computing environment; 
  • Analyze the results (quantitatively and qualitatively); 
  • Prepare and present a report on the results obtained and recommendations for future work.  

Required skills :

Familiarity with machine learning algorithms, specifically Random Forest, and transfer learning and deep learning. Experience with the R or Python programming language. 

Additional skills :

Knowledge in geomatics and image processing is an asset. 

Language :

The language used by the intern must be English. The intern’s work team will be composed of French and English speakers. 

Requirements :

Canadian citizenship or permanent residency

Contact :

If you are interested in this internhip, please contact by email coordonnateur du programme DOTS