Mapping and Tracking Real-Time Atmospheric Carbon Emissions: Climate TRACE identifies the 70,000 highest emitting greenhouse gas sources
By Bronwyn Agrios
Today Climate TRACE launches the first detailed facility-level global inventory of greenhouse gas (GHG) emissions — identifying, tracking, and monitoring individual sources of emissions across twenty major sectors of the economy — and mapping the 70,000 highest emitting greenhouse gas sources down to their source.
No one, in the 30 years since governments began negotiating the reduction of greenhouse gas (GHG) emissions, has been able to identify exactly where all the pollution is coming from — until now. Climate TRACE data is based primarily on direct observation training AI algorithms to fuse data across multiple wavelengths and from more than 300 satellites and 11,100 air-, land-, and sea-based sensors, providing facility-level precision of individual emitters on a global scale.
Facility-level precision of individual emitters
The map and data browser were officially launched this morning as part of Vice President Al Gore’s speech at this year’s United Nations Climate Change Conference, COP27, in Sharm El Sheikh, Egypt.
“This level of granularity means that we finally have emissions data that enable us to act decisively. It also means we can prioritize efforts to achieve the deep cuts in greenhouse gas pollution we need to prevent the most catastrophic impacts of the climate crisis.” — Vice President Al Gore
This release makes the Climate TRACE data browser the largest open data set of global emissions inventory in history, providing valuable insights to help inform climate policy and improve our ability to take action on climate change, globally.
Data sources
Until today’s launch, it has not been possible to precisely map and track the specific sources of GHG pollution. While the Keeling Curve — a daily record of global atmospheric CO2 concentration — monitors total CO2 in the atmosphere, tracking the global average of humans increasing production of GHG from 313 parts per million (ppm) in March 1958 to 414.72 ppm of of 2021, there has never been a facility-level inventory of the largest known individual sources that are producing the 162 million tons of GHG into the troposphere every day.
Climate TRACE is a coalition of artificial intelligence (AI) specialists, data scientists, researchers, and nongovernmental organizations from around the world who together provide the most detailed facility-level global inventory of greenhouse gas emissions ever, mapping emissions data for 79,815 individual sources worldwide.
“The climate crisis can, at times, feel like an intractable challenge — in large part because we’ve had a limited understanding of precisely where emissions are coming from,” said Vice President Al Gore — the new machine learning models trained by the Climate TRACE coalition members create transparency. Sensors alone are not enough to monitor GHG — the atmosphere is so enriched with CO2 that the “signal-to-noise ratio” defies the ability to measure CO2 point-source emissions directly, leaving AI to spot sometimes subtle differences in satellite imagery to calculate emissions from individual sources the human eye and traditional monitoring methods have missed in the past.
Open Data
Climate TRACE emissions data is free and publicly available for download at https://climatetrace.org/downloads. Each download package includes annual country-level emissions by sector and by greenhouse gas from 2015–2021, the applicable inventory of facility-level emissions, and facility-level ownership data where available.
The Science
Accessing more detailed emissions data is especially critical given last week’s United Nations Emissions Gap Report indicating that governments are behind pace for reducing carbon dioxide (CO2) emissions by 2030, with none of the world’s biggest emitter countries reducing — their burning of fossil fuels, transportation, industry, conventional agriculture, and deforestation enough to meet the Paris Agreement goals.
Monitoring carbon dioxide (CO2) emissions are critical to the international community’s commitment to limit global temperature rise to 1.5°C.
AI observations v self reported data
Climate TRACE AI analysis captures emitting activities that are otherwise invisible or undetectable, making the launch of this data browser more comprehensive and more accurate, in addition to being more detailed than ever before.
Full global coverage: The Paris Agreement targets are currently informed by rough estimates principally derived from self-reported national inventories submitted intermittently to the United Nations (UN). Coverage has not been timely or comprehensive, as of the start of COP27 this week, no country submitted a complete accounting of their emissions for 2021, and more than 50 countries have not submitted any emissions inventories covering the past 10 years.
Accuracy: Traditional inventories that do exist often have large omissions and fail to provide the granular data needed to make decisions. The new Climate TRACE data finds emissions are as much as three times higher than self-reported data. For example, last year, Climate TRACE found the actual emissions from global oil and gas production collectively were around double what was self-reported to the UN in 2020. Even more oil and gas emissions went uncounted by countries that are not required to report these data. AI analysis of satellite-detected emissions from flaring and methane leakage in Russia, Turkmenistan, the U.S., and Middle East show consistent underestimates of methane emissions from both intentional flaring as well as leaks.
Built on a model of continual improvement, Climate TRACE will steadily become more comprehensive and timely as more and more data is collected. The Climate TRACE inventory was jointly created by 10 coalition members (Blue Sky Analytics, Carbon Yield, Earthrise Media, Former Vice President Al Gore, Hypervine.io, Johns Hopkins University Applied Physics Lab, OceanMind, RMI, TransitionZero, and WattTime) plus over 90 other contributing organizations and researchers, including: Carbon Mapper, CTREES, Descartes Labs, Global Energy Monitor, Google.org, Michigan State University, Minderoo Foundation/Global Plastic Watch, Planet Labs PBC, Synthetaic, Universiti Malaysia Terengganu, and others.
More data inputs = more precise AI and machine learning models, see Climate TRACE methodology documents on data inputs, modelling, and relevant peer-reviewed research. More more details around today’s launch watch Al Gore and Gavin McCormick presentation at COP27 https://unfccc.int/event/al-gore-and-the-climate-trace-coalition-release-detailed-inventory-of-the-sources-of-ghg-emissions