A new study from Northern Arizona University reveals a significant gap in urban carbon accounting. Research shows that Climate TRACE, an AI-driven emissions database, underestimates vehicle CO2 emissions in 260 U.S. cities by an average of 70 percent, with some cities missing up to 90 percent of their actual output. This discrepancy highlights the risks of relying solely on satellite imagery and machine learning for climate data. While national totals may appear accurate, the lack of granular detail in urban areas can lead to flawed city climate plans and misallocated electric vehicle subsidies.
