license	cc-by-4.0
language	en fr
size_categories	1K<n<10K
tags	canada environment climate emissions ghg npri esg sustainability rag llm-training
pretty_name	PollutionData — Canadian GHG & Pollutant Releases (Sample)
configs	config_name data_files default split path train pollutiondata_sample.jsonl

PollutionData — Canadian GHG & Pollutant Releases (Sample)

Sample of 5,000 Canadian facility-level environmental records across 3,005 facilities. Mix of pollutant releases (NPRI) and greenhouse gas emissions (GHGRP). Last 2 years of data from the full PollutionData API.

Want the full dataset? 1.08M+ records, every NPRI/GHGRP-reporting facility in Canada since 1993, queryable by facility / pollutant / industry / region. Get an API key →

What's in here

Field	Type	Description
source_id	string	Originating system (npri or ghgrp)
entity_type	string	release (NPRI pollutant release) or emission (GHGRP greenhouse gas report)
record_id	string	Stable per-record identifier from the source
municipality	string	Facility location
event_date	string (ISO date, year-precision)	Reporting year (YYYY-01-01)
category	string	Pollutant / GHG type
facility_id	string	Stable facility identifier (NPRI/GHGRP)
total_co2e	float	Total CO2-equivalent emissions in tonnes (where applicable)
data	object	Full raw record from the source (JSON) — quantities, NAICS, NPRI codes

Composition (sample)

entity_type	rows
release (NPRI pollutants)	4,000
emission (GHGRP GHGs)	1,000

3,005 unique facilities represented.

Use cases

• ESG due diligence — counterparty and supply-chain emissions screening
• Carbon accounting — Scope 3 input data for upstream supplier emissions
• Climate research — emissions trends, polluter benchmarking, regional analysis
• RAG / agents — ground LLM responses in actual Canadian industrial emissions data
• Compliance & benchmarking — compare facility performance against industry peers
• Journalism / NGO reporting — identify largest polluters in a region or sector
• Schema exploration before committing to API integration

Limitations of this sample

• Snapshot only — frozen 2026-05-25. New reports added annually as facilities file.
• 5K rows — full dataset is 1.08M+. Sample is enough to test schema, prompts, prototypes; not enough for production analytics.
• No querying — can't filter by pollutant, NAICS, facility, or year. Use the API for that.
• Year-precision dates — NPRI/GHGRP are annual reports, not transactional records.

Production access

• Docs: https://pollutiondata.ca/docs
• Pricing: https://pollutiondata.ca/pricing — free tier 25 req/day, Pro $49/mo 1K req/day
• Quickstart:

  curl "https://api.pollutiondata.ca/emission?facility_id=XXX&limit=10" \
    -H "x-api-key: YOUR_KEY"

License & attribution

Data sourced from Canadian federal environmental disclosure programs: NPRI (National Pollutant Release Inventory) and GHGRP (Greenhouse Gas Reporting Program). All source data is under Open Government Licence — Canada. Sample redistribution is permitted under CC-BY-4.0.

Contact

• API and enterprise inquiries: nicolas@builddata.ca
• Bug reports: https://github.com/NicolasPrimeau/Nimbus (or via the contact link on pollutiondata.ca)

Mirror locations

The same sample is published in three places — pick whichever you prefer:

• Hugging Face: https://huggingface.co/datasets/nimbusdata/pollutiondata-canadian-emissions-sample
• Kaggle: https://www.kaggle.com/datasets/niexon/pollutiondata-canadian-emissions-sample
• GitHub: https://github.com/nimbusdata/pollutiondata-sample