Advanced technology to tackle Greenhouse Gas Emissions
Google’s AI Impact Challenge grantee WattTime is pioneering a new approach to tackling emissions, with help from a team of Google.org Fellows.
Excitement to work together on a pressing Global program
It’s not every day that you get 50 inbound calls from climate leaders around the globe but shortly after WattTime was announced as a Google AI Impact Challenge grantee, WattTime Executive Director Gavin McCormick’s phone was ringing off the hook. Much like our Google AI Impact Challenge reviewers, the callers took interest in WattTime’s innovative proposal to create a global tool for tracking power plant emissions.
Global power generation accounts for an enormous share of our planet’s greenhouse gas (GHG) emissions. While there is good data about emissions in the US and parts of Europe, there is no global, independent, consistent source of data. This lack of data limits the ability for power providers, researchers, and consumers to take steps to lessen the environmental impact of emissions.
Tackling the problem of Global Emissions
The goal of the Google AI Impact Challenge was to support organizations applying AI for positive social impact, including a number of organizations working on advanced solutions across sustainability, climate, and environmental protection. WattTime received a $1.7M grant through the Challenge to work with CarbonTracker and the World Resources Institute to create a model for using satellite imagery and verified emissions data from the US and Europe to estimate emissions from thermal power plants anywhere in the world.
Standardizing and making power plant emissions data broadly available unlocks benefits for people across the sector. For the first time, power plants and electrical grid operators can track how they are doing against their peers. Climate scientists can better analyze pollution sources and global warming. We have a way to track major climate commitments, like the Paris Agreement, and understand what we need to change in order to meet them. WattTime is pairing the data with additional machine learning to add new insights to their Automated Emissions Reduction (AER) software. AER helps any end user — from large scale power consumers to individuals — effortlessly reduce emissions from energy in real-time. It works by helping devices with flexible electricity needs, like heating and air conditioning systems or electric cars, pull more energy from the grid when solar and wind power are present, and less when the grid is using dirtier power sources.
The Google.org Fellowship: Putting ideas into practice
As with all of our Google.org grants, we like to do more than just give money. To help WattTime build their AI model, we paired our grant with a Google.org Fellowship which enables Googlers to spend up to six months doing full-time pro bono work with nonprofits and civic entities. The program has provided more than 100,000 hours of support to organizations worldwide since 2019 when it launched. Over the course of six months, seven Google.org Fellows helped WattTime use AI to estimate carbon-fueled power plant emissions based on a combination of satellite imagery and other data sources collected by WattTime.
How do you measure emissions using machine learning? Machine learning requires a training dataset to learn from so that it can start to make predictions. In this case, the team started by matching time-stamped images of power plants in the US with existing time-stamped emissions data. By labeling that data and understanding the relationship between inputs like the size and heat profile of the smokestack plume, you can generate projected emissions data using just a satellite picture.
And Google.org Fellows played a big role in all of this — helping to identify multiple publicly available satellite image collections and matching the images with emissions data from three different government agencies. The Fellows also applied Google’s machine learning expertise to set up reusable tooling that made it easy for WattTime to experiment with their machine learning models, try different input data, understand the accuracy of their models, and apply ML fairness principles to make sure that models performed equally well across geographies.
Takeaways from the Google.org Fellows
“Climate change like any other problem can only be solved if we agree on the facts. This project helps bring clarity to emissions and build clear tools that are needed for scoping and addressing climate change. I really enjoyed how the vision wasn’t limited to doing something purely in the western world, but having a holistic global perspective, and not just limited to the energy sector but building a collaboration with nonprofits and working with policy makers from all countries.” — Alok Talekar, Software Engineer (Google.org Fellow)
“One of the learnings from my research was that while science appears objective and is driven by facts, it is still consumed by humans who are themselves products of their own cultures and own biases. Climate TRACE became a global, not just western, scientific collaboration. All of us had to consider different viewpoints and practices in order to bring this undertaking to life and keep it alive and moving forward. It is truly an international effort that shows clear benefits to a collective mindset when dealing with global challenges.” — Kathi Kitner, UX Researcher (Google.org Fellow)
“One of my biggest personal takeaways is that there is reason for optimism: not unbridled and naïve, but measured and realistic. The scope of the problem is huge, and success depends on carefully laying out the path from our work to the impact on people. Every step of the way, though, we found people willing to help and join the collaboration to solve real problems. As our momentum increases and the challenge begins to appear surmountable, I am filled with cautious, excited hope.”- Daniel Tyrrell, Software Engineer (Google.org Fellow)
From one project to a global consortium
Based on the flurry of phone calls after the Google AI Impact Challenge, WattTime saw there was energy across the sector to create a platform with an even bigger vision — to do what WattTime was doing for power plants, for all major industries: shipping, forestry, steel, and more. Essentially, to build a consortium of organizations to collectively measure all GHG emissions globally, in real time. As WattTime, Carbon Tracker, and the Google.org Fellows dug in and helped develop the plans, identify potential coalition members, and organize the initial meeting that helped to make it possible.
5 short months after the initial round of calls, Climate TRACE was born, and it was unveiled this July with 10 members. The hard work of the collective is building a system that can consistently and accurately combine the estimates that each coalition member generates into one cohesive dataset of GHG emissions. This data will allow users — like you and I — to explore and understand emissions anywhere around the globe, and to compare the estimates to published data, reports, and goals.
And just like that wave of calls to WattTime, the Climate TRACE launch sparked some powerful momentum with over 60 organizations interested in joining. Thanks to Climate TRACE’s egalitarian structure, all parties see their contributions as critical to the vision of a universal, independent, transparent GHG emissions database. They hope the database will be ready to launch next year alongside COP26, dramatically improving the ability of governments and industry to progress toward the Paris accords.
What’s next for Google.org and Climate Action?
The world is facing a rapidly accelerating climate crisis and we are committed to continuing to do more to address climate change. Last week, as part of Google’s ambitious plan for a third decade of climate action, we announced a new Google.org Impact Challenge on Climate. We’ll be making €10M in grants to fund organizations with big, bold ideas that use technology to accelerate Europe’s progress toward a greener, more resilient future. Like WattTime, our Google.org Impact Challenge Climate grantees will receive mentorship and technical support from Googlers via the Google for Startups Accelerator. Applications are due November 6th — please help us spread the word, and follow along as we share what we learn.
Google.org brings the best of Google to help solve some of humanity’s biggest challenges and provide opportunity for everyone. We’re proud to support WattTime as they pioneer powerful solutions to monitor and address global emissions.