Initiative overview
In July 2025, the Enforcement Branch (EB) of Environment and Climate Change Canada (ECCC) tasked ECCC’s Artificial Intelligence and Emerging Technology Centre of Excellence (AI&ET CoE) to investigate the potential use of space-based remote sensing technology in detecting illegal discharges and pollution events in freshwater bodies. Leveraging AI models, the goal is to automate detection and prioritise potential incidents for human validation and inspection referrals.
This research supports EB’s mandate to enforce federal environmental laws across Canada’s vast and diverse ecosystems, including remote and protected regions. Given the challenges of monitoring such expansive areas, particularly in remote locations, EB has traditionally relied on company-reported data and toxicity levels described in literature to guide its enforcement activities. These limitations emphasise the need for innovative approaches to maintain compliance.
The initial focus of the research is on the mining industry, which falls under the Canadian Environmental Protection Act and the Fisheries Act. Specifically, under the Fisheries Act, EB conducts inspections to ensure compliance with the Metal and Diamond Effluent Regulations (MDMER). MDMER allows the controlled discharge of effluent from metal and diamond mines into fish-bearing waters, provided strict effluent quality standards are met. These regulations aim to protect Canada’s lakes and rivers from harm caused by mining operations.
Currently, inspections under MDMER are conducted biennially but involve substantial costs—up to $25,000 CAD for remote sites—with limited revisit frequencies and challenges in detecting non-compliance. The primary aim of the initiative is to transition from an episodic, infrequent, and reactive inspection model to a system of continuous and proactive monitoring of industrial sites. Implementing AI-based monitoring can significantly increase the likelihood of conducting targeted inspections aligned with non-compliance events, which may otherwise go unnoticed.
Additionally, the monitoring system—called PolluSense—provides valuable insights into facility operational patterns and risk profiles, helping prevent large-scale environmental damage and reducing cleanup and health costs. The AI&ET CoE has now advanced PolluSense to the prototype phase, which is undergoing experimentation and testing with EB. The initial objective is to develop scalable and proactive monitoring capabilities that enhance EB’s cost-efficiency and strengthen environmental protection efforts.
Looking forward, the AI&ET CoE plans to expand geographic training data beyond Ontario and Quebec to cover additional Canadian regions and explore applying the technology to other industries, such as agriculture, oil sands, and pulp and paper.
