The city of Los Angeles has been collecting data on its carbon pollution footprint for over a decade as a part of its efforts to fight global climate change. In 2017, the city reported that it had reduced greenhouse gas emissions by 40% from 2008 levels, through various changes such as improving electricity usage in traffic lights, energy efficiency gains in buildings and airport design modifications. It turns out that the city’s emissions measurements were off, though – by more than 50%.

And it wasn’t just L.A.: The city of Cleveland, Ohio, undercounted its emissions by 90%, while Torrance, California, and Blacksburg, Virginia, both miscalculated their cities’ emissions by more than 100% in their respective public reports, according to a study published in the journal Nature Communications in February. Researchers from the Vulcan Project found that U.S. cities overall under-reported their greenhouse gas emissions by nearly 20%. The Vulcan Project is a multi-year effort funded by NASA and the U.S. Department of Energy  to analyze data on carbon emissions over the United States. The major discrepancies Vulcan Project researchers found between their own data  and many cities’ self-reported numbers is now  cities to rethink the way they conduct their own emissions inventories.

U.S. cities typically report their own carbon pollution loads from everyday activities — traffic congestion, electricity usage, waste treatment, as a few examples — and these inventories help direct cities’ efforts at reducing greenhouse gas emissions, whether by planting trees to bolster the tree canopy or reducing food waste. This work, often carried out by a city sustainability office, is an increasingly important aspect of urban planning, as the majority of global carbon emissions come from urban areas. Among the 48 cities included in the study, seven under-counted their numbers by more than 50%. As Bloomberg previously reported, if these results were extrapolated across the country, the under-counted emissions would amount to more than 474 million tons of CO₂ — almost 25% more than California’s total emissions for 2015.

The findings have sparked a discussion among urban emissions experts about the methodologies cities use to measure their carbon pollution, and how to collect data that best supports actionable policy.

It’s not that cities are under-reporting as a way to game the system — there’s no official system to game. Since there is no national mandate for U.S. cities to collect carbon emissions data — and no standard methodology for doing so — cities do this work voluntarily. As a result, there is considerable variability in what each city measures and how, which may help explain why their carbon pollution records don’t match the Vulcan Project’s.

Having both accurate data and actionable policy is essential for cities in the work of understanding and reducing greenhouse gas emissions. The study’s authors say cities need to be working from a more comprehensive dataset to achieve these goals.