The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Tracking CO2 Emissions with Machine Learning with Laurence Watson - TWIML Talk #277
Jun 24, 2019
In this conversation, Laurence Watson, Co-Founder and CTO of Plentiful Energy and former data scientist at Carbon Tracker, dives into innovative methods for tracking CO2 emissions using machine learning and satellite imagery. He shares insights from Carbon Tracker's research on fossil fuel power plants and discusses the challenges of quantifying emissions accurately. The talk also highlights advancements in cloud solutions for data integration and how they enhance monitoring efforts, making environmental data more accessible and actionable.
AI Snips
Chapters
Transcript
Episode notes
Carbon Bubble
- Carbon Tracker aims to quantify climate risk for financial markets by demonstrating how fossil fuel valuations are tied to future sales.
- These valuations are threatened by climate change constraints and government regulations, creating a potential "carbon bubble".
Capacity Factor Importance
- Capacity factor, the proportion of time a plant runs compared to its maximum output, is crucial for financial evaluation.
- It balances revenues earned during operation with ongoing fixed costs like maintenance and lifetime extensions.
Initial Approach and Challenges
- Initially, Carbon Tracker aimed to use labeled data from US and EU plants for supervised learning to predict capacity factors.
- They encountered imbalanced training classes due to plants running more during daytime satellite imagery collection.