Press "Enter" to skip to content

ClimateTRACE: Using Satellites and Machine Learning to Pinpoint Global Emissions

Data on the greenhouse gas emissions from power plants, factories, cargo ships, controlled burns and every other human source on Earth could soon be part of the public domain, according to a consortium of technology companies and climate change nonprofits. 

These are the goals of ClimateTRACE, a group that’s won the backing of Google.org and former U.S. Vice President Al Gore to use advanced computing techniques to convert satellite images and sensor data into a publicly available source of real-time data on carbon dioxide emissions and other air pollution. 

ClimateTRACE has been in the works since last year, when Google.org awarded a $1.7 million grant to a group of nonprofits including U.K.-based Carbon Tracker and U.S.-based WattTime for a proposal to track global power plant emissions from space. The idea is to combine satellite imaging of smoke being emitted from power plants, factories and other sources with infrared heat imaging and nitrogen oxide (NO2) sensor data, and analyze the results via machine learning to yield emissions data. 

This approach could yield far more detailed and up-to-date information than the self-reporting methods commonly available to governments and researchers today. Some governments regulate and monitor GHG emissions on a source-by-source basis, as with the U.S. Environmental Protection Agency (EPA) and the European Union’s Climate Monitoring Mechanism. Others rely on self-reporting regimes that lack transparency, don’t identify individual sources, or can be subject to deliberate under-reporting, according to Wednesday’s press release.

Even the most accurate and transparent data is collected periodically, often on an annual basis, yielding data that can be years out of date, WattTime executive director Gavin McCormick said. Trying to combat climate change with such data “is like asking a doctor to fix a serious disease with no more information than a list of symptoms the patient had years ago.” 

The ClimateTRACE coalition, which also includes nonprofits CarbonPlan, Earthrise Alliance, Hudson Carbon, OceanMind, Rocky Mountain Institute, alongside tech companies Blue Sky Analytics and Hypervine, hopes to have the first iteration of its full tool and first full global data report released by the summer of 2021. “Like many AI projects, the tool will continuously improve as the team adds more data and works out more sophisticated algorithms,” the press release noted.

In an email exchange, McCormick highlighted that ClimateTRACE is in its beginning stages of developing the technology to yield accurate measurements across the emissions sources it’s tracking. “It’s just a prototype — we’re not at our desired accuracy in any area yet.” 

“But in the electricity sector, we’re farthest along with coal-fired power plants with once-through cooling systems,” he said. “In other sectors, I’ve been particularly impressed with our various partners’ results on forestry, steel plants, fires, and large ships.” 

As its accuracy improves, ClimateTRACE could become a valuable tool for multiple use cases, from detecting illegal polluters to providing data to verify compliance in carbon cap-and-trade markets and international climate change agreements like the Paris climate agreement, he said. 

Building a global database on marginal carbon emissions

WattTime got its start in 2014 by using historical and real-time power plant emissions data to discover the marginal carbon emissions intensity of grid electricity from hour to hour across the United States. Its first product, a free app to guide utility customers on when their energy consumption was more or less carbon intensive, led to partnerships with universities, electric vehicle chargers, and scientists at the Energy Department’s Pacific Northwest National Laboratory. 

WattTime, now a subsidiary of the Rocky Mountain Institute, combines historical power plant emissions data from the EPA’s Continuous Emission Monitoring System ICEMS) with real-time data from the Open Access Same-Time Information System (OASTIS) used by federally regulated grid operators such as PJM, ERCOT and CAISO. The result is a real-time signal of electricity carbon intensity across different utility and grid operator regions in the U.S. and Canada, as this snapshot of WattTime’s online map, tracking carbon intensity during a mid-July afternoon, indicates: 

WattTime’s calculations are now being used by California to guide energy storage projects complying with the carbon emissions regulations for its Self-Generation Incentive Program (SGIP) — one of the first real-world implementations of such a fine-grained marginal GHG intensity measure for a major utility program. Its Automated Emissions Reduction (AER) technology, first piloted in 2017, provides the same marginal GHG intensity data for distributed energy device vendors, including smart thermostats from Energate and electric vehicle chargers from Enel X’s eMotorWerks. 

The price of a marginal watt of grid power doesn’t always align with its carbon intensity, which means that devices trying to optimize both at once may have to decide which to prioritize from hour to hour, McCormick noted in a recent interview. “We want to see this marginal emissions signal embedded in everything, where it doesn’t make people’s lives harder,” to give electricity consumers the data to make the choice on their own.  

Data coming from the ClimateTRACE project could enhance the accuracy and granularity of such marginal GHG calculations, as well as extend its usefulness to countries that lack rigorous GHG emission reporting regulations, he said. “AER works by knowing marginal emissions data […] In some countries like the U.S. we are lucky that our government already collects such data. But most countries in the world don’t.” 

It could also give companies trying to reduce or eliminate their carbon footprint more data to guide their decisions, he noted. For example, the World Resources Institute’s Greenhouse Gas Protocols, an accounting and reporting standard used to guide many corporate carbon emission reduction efforts, was developed in 2014. Back then, there wasn’t technology to inform companies about whether shifting power use from one hour to another could directly impact whether utilities and grid operators would be forced to call on natural gas-fired peaker plants to start up to meet peak grid demands — “but now, it is possible.” 

Source: Greentech Media