AI for Sustainability

with Dr. Vik Pant

Dr. Vik Pant is the Chief Scientist and Chief Scientific Advisor for Natural Resources Canada (NRCan). He is responsible for providing strategic direction to build capacity within their scientific community and accelerating the creative application of innovative digital technologies including Artificial Intelligence.

Episode 16: AI for Sustainability

Public Sector Future

Episode summary

In this episode we speak with Dr. Vik Pant, the Chief Scientist and Chief Scientific Advisor for Natural Resources Canada (NRCan). Dr. Pant shares his priorities in his role and learnings from NR Canada’s digital accelerator as they work to implement new technologies. Discover how new artificial intelligence capabilities are being used to promote sustainability, from improved flood mapping and wildfire response to mining risk mitigation.

Download the episode transcript

Listen to this episode on any of these podcast platforms:

Spotify

Apple Podcasts

Google Podcasts

Deezer

Libsyn

Artificial intelligence for sustainability with Dr. Vik Pant

Dr. Vik Pant is the Chief Scientist and Chief Scientific Advisor for Natural Resources Canada (NRCan). Dr. Pant discusses the priorities in his role at NRCan and learnings from their digital accelerator as they work to implement new technologies to deliver on their objectives.

Discover how new artificial intelligence capabilities are being used to promote sustainability, from improved flood mapping and wildfire response to mining risk mitigation.

Natural Resources Canada

“Natural Resources Canada is a ministry of the federal government where I’m employed, whose focus is to improve the quality of life for all Canadians by ensuring that our country’s abundant natural resources are developed sustainably, competitively, and inclusively,” Dr. Pant explained.

Dr. Pant shared that approximately 17% of Canada’s nominal GDP can be directly or indirectly attributed to the natural resources sector, which is responsible for around 1.8 million jobs.

Approximately 5,000 people are employed by Natural ReCan. “Almost half of our expenditures and half of our workforce is in fact science and technology, research and development related,” Dr. Pant said.

The Canadian Forest Service and Geological Survey of Canada are both part of Natural Resources Canada, as well as the Canadian Center for Mapping and Earth Observation, which is Canada’s national cartographic agency responsible for space-based Earth observation and remote sensing.

“Science is at the heart of NR Canada, and that’s why we think of ourselves as a science-based economic department,” Dr. Pant shared.

Priorities as Chief Scientist of Natural Resources Canada

Dr. Pant has three main priorities and areas of focus in his role as the Chief Scientist of Natural Resources Canada.

1. Horizontal science

Ensure that there is collaborative learning among the scientific experts at NRCan and that there are opportunities and pathways for scientists to exchange knowledge so they can benefit from their complementarities.

2. Holistic science

Ensure that NRCan is looking at science not just from one intellectual tradition, which is Western science, but also from multiple intellectual traditions.

“We have relationships with communities that are knowledge holders, for instance, with indigenous knowledge and indigenous learning. It’s to ensure that our practices and our processes within the department are respectful of their intentions, their concerns, their goals and objectives as well, and also making sure that there are training programs and to ensure that there are knowledge-sharing programs available to our scientists,” Dr. Pant explained.

3. Digital acceleration of science

“Accelerate the creative application of innovative digital technologies to maximize the impact of software, of technology, in our department, and this means working with technology such as artificial intelligence, data science, blockchain, Internet of Things, 3D printing,” Dr. Pant said.

Elements to a successful technology launchpad in the public sector

Dr. Vik Pant shares five pillars of success to NRCan’s digital accelerator.

1. Talent – The first action is to hire talented and qualified individuals to build capacity.

“When I came into the department, I reached into my networks and connections and created value propositions that I knew would resonate with the types of folks that we were targeting,” Dr. Pant shared.

2. Co-creation & value realization – Once a team is assembled, the focus shifts to value realization.

“We wanted to make sure that the scientist and our data scientist, our digital solution architects, our software technologists came together, and they collectively worked to co-create value,” Dr. Pant explained.

3. Digital acumen – Nurture a digital first, data-aware culture.

“Not everybody wants to become a data scientist. The level of conversation in the department needs to be sophisticated where folks have a general understanding of the value of data, how to think about data as a resource with opportunities for value creation,” Dr. Pant described.

“And so, knowledge sharing and this notion of what we call the digital acumen, became very important for us as the accelerator– we wanted to make sure we created ample learning opportunities for colleagues from across our department to learn about data science,” Dr. Pant continued.

4. Strategic partnerships – Partner with like-minded organizations to build mutually beneficial relationships.

Go deeper than traditional working partnerships where you purchase something like software in exchange for payment.

“We looked around for like-minded organizations, primary case being Microsoft, where we wanted to focus on not just the topic of software, but we wanted to look at what else could Microsoft bring to the table, and in return, what we could bring to the table in a way that led to win/win, mutually-beneficial outcomes,” Dr. Pant explained.

“Now, for instance, in the relationships we have with Microsoft, we focus on goals not just in terms of software but in terms of higher order objectives, sustainability, net zero, climate change mitigation, adaptation, remediation,” Dr. Pant elaborated.

5. Governance – Go from pilot to production while providing opportunities for experimentation and abiding by government regulations.

“You want to give data scientists and all the stakeholders that are involved in that design and development process a lot of latitude. But at the same time, at a macro level, from a portfolio perspective, we are the government,” Dr. Pant shared.

“We still have to make sure that we’re following all the rules, we’re abiding by all the regulations,” Dr. Pant continued.

Utilizing artificial intelligence technology to analyze satellite imagery

Natural Resources Canada’s digital accelerator partnered with the Canadian Center for Mapping and Earth Observation, which has access to satellite imagery through the Canadian Space Agency. Historically, analyzing the satellite data was a manual exercise, using microscopes or other magnifying equipment to analyze and understand the features in the imagery.

The Canadian Center for Mapping and Earth Observation had been training an ensemble of artificial neural networks with the help of human input experts to annotate satellite images down to the pixel level. With a sufficiently large dataset of expert annotated satellite images, they could train a machine learning model.

Now, it is possible to “take satellite imagery as it comes in off of the satellites and then use advanced artificial intelligence techniques and technologies to analyze those images with a view to understanding, down to the pixel level, what that pixel represents,” Dr. Pant shared.

“If you think about a pixel as representing the top of a tree versus the top of a road versus the top of a building… you can imagine that now when you have raw pixels overlaid with artificial intelligence derived semantic layer… then you could really make all types of very interesting decisions from a planning perspective, from a policy perspective, from a forecasting perspective,” Dr. Pant continued.

Flood mapping and forest fire response improved with artificial intelligence

Natural Resources Canada is responsible for preparing flood maps to understand the destructive nature of floods, from where they can destroy property and harm lives and livestock.

The goal is to continually improve flood mapping to be better at flood predictions and response. This new technology generates a superior flood map with augmentation from subject matter experts and artificial intelligence bolstering the set of insights, which can be leveraged to make decisions. NR Canada must also monitor wildland and forest fires, which harm the economy, ecology, and environment.

As a result of AI satellite technology, there is an improved understanding of the topology, contours, and covers on land that is at risk of wildland and forest fires. This knowledge is combined with existing models from the Canadian Forest Service.

“Now you can have a good understanding of how to respond to wildland fires and forest fires, and even perhaps predict them,” Dr. Pant explained.

Mining risk mitigation in partnership with Microsoft

Another project NRCan is working on, in partnership with Microsoft, is mining risk mitigation.

Dr. Pant explained that Canada has about 10,000 mine sites. It’s a full-time function to ensure that ecological hazards and environmental harms are mitigated at active mines. Abandoned mines also must be monitored to ensure they don’t create destruction and harm to the environment or the ecology. Visiting all the mine sites is an expensive enterprise not only financially but also factoring in the time, risk, and labor involved.

Satellite imagery and machine learning modelling help on an ongoing basis to ensure the conditions of the abandoned mine sites are cataloged and inventoried. Over time they can do difference analysis and see how things are changing, for example if a rock pile is degrading or an emissions pond is leeching.

“We can make some good approximations about the quality of the mine site and also then make decisions downstream about what type of actions to take to remediate it,” Dr. Pant said.

“This is an ongoing project. We’ve been teaming very successfully, using our datasets, Microsoft’s dataset, our expertise and knowledge in subject matter, domain knowledge, as well as Microsoft’s expertise in the domain, but also in terms of software, and really just bringing our ideas together, and trying and testing different approaches to see what really works,” Dr. Pant shared.

Geo-deep-learning available for public use

Geo-deep-learning is the Canadian Center for Mapping and Earth Observation artificial neural network ensemble, which is used for assigning layers and features, from land cover pixels down to the land cover classes down to the pixel level. This software code is available on GitHub, which is a publicly accessible code repository.

“We know that there are lots of public and private sector, and academic researchers who are using that code base to not only use it for their purposes but also then to recommend changes or improvements, or submit a code request,” Dr. Pant explained.

“As much as possible, we are focused on open innovation and open science,” Dr. Pant shared.

Advice for successful strategic partnerships

Dr. Pant shares his advice for creating and sustaining strategic partnerships.

• Transparency at all levels – Let leaders know what you are trying to do to build trust. Avoid surprises.
• Allocentric (outside-in) thinking – Know what the motivations are of the other parties and then help to align them with your own motivations as well.

To find out more:

Learn more about Natural Resources Canada

Learn more about Geo-Deep-Learning

Follow Vik Pant on Twitter

Research: Digital Approaches for Sustainability