Study Finds Climate TRACE Understates U.S. City Vehicle CO₂ by 70 Percent

Context A team at Northern Arizona University examined the Climate TRACE database, a global greenhouse‑gas inventory that uses artificial‑intelligence (AI) models to infer emissions. The researchers compared the AI‑derived vehicle emissions for 260 U.S. cities with data from Vulcan, an on‑road emissions dataset built from official traffic counts and fuel‑use statistics. Vulcan, developed by the same university, carries an uncertainty of roughly 14 %.

Key Facts - Across the sample, Climate TRACE’s vehicle CO₂ numbers were, on average, 70 % lower than Vulcan’s. - In Indianapolis and Nashville, the AI‑based estimates fell more than 90 % short of the Vulcan values. - The study, published in *Environmental Research Letters*, notes that similar under‑reporting was found in a prior analysis of Climate TRACE’s power‑plant emissions. - Researchers matched each city’s reported emissions to the corresponding Vulcan figure, then calculated the percentage difference. The large gaps persisted despite Vulcan’s modest uncertainty, indicating a systematic bias rather than random error.

What It Means If policy makers rely on Climate TRACE for urban emissions inventories, they may underestimate the scale of vehicle‑related carbon output and misallocate mitigation resources. Accurate emissions data are essential for setting realistic reduction targets, allocating funding, and tracking progress toward climate goals. The authors stress that while AI offers promising speed and coverage, rigorous validation, transparency, and expert review remain crucial to maintain trust.

Kevin Gurney, lead author, emphasized that data shared with decision‑makers must be unbiased and meet the highest scientific standards, even if perfect accuracy is unattainable. The team recommends that Climate TRACE incorporate independent benchmarks like Vulcan, disclose model assumptions, and subject outputs to regular peer review.

Looking Ahead Future work will test whether the underestimation extends to international cities and explore adjustments to the AI algorithms. Monitoring updates to Climate TRACE’s methodology will be key for anyone tracking urban emissions data.