Search CARMA by country, state, province, county, metro area, city, power company,
power plant, or zip code.

Frequently Asked Questions

Who produces CARMA? CARMA is a product of the Confronting Climate Change Initiative at the Center for Global Development. Located in Washington DC, The Center for Global Development is an independent, non-profit think tank that works to reduce global poverty and inequality by encouraging policy change in the U.S. and other rich countries. CARMA is currently maintained on a part-time basis by Kevin Ummel.

Where do the data come from? The data in CARMA are compiled from numerous sources. Emissions data for thousands of power plants in the U.S., Canada, the European Union, India, and South Africa come from official reports and databases. Other data are derived from information provided by proprietary commercial databases, the International Energy Agency, the International Atomic Energy Agency, the U.S. Department of Energy, the Environmental Protection Agency, and the open-source GeoNames database in conjunction with CARMA’s internal statistical models. See the CARMA working paper for more details.

How current are the data? The current release is v3.0, released in July 2012. It contains data (reported and estimated) for 2004, 2009, and the future. Information regarding the future is accurate as of March, 2012.

How accurate are CARMA’s estimates? When reported (i.e. disclosed) data are not available, CARMA relies upon statistical models that predict plant electricity generation and CO2 emissions using information about the size, age, fuel type, and engineering specifications of individual generating units. A technical paper providing documentation for the CARMA v3.0 database includes an extensive test of the models’ predictive skill.

I want to get all of the data for research purposes. How can I do that? There is a CARMA Power User .zip package available for researchers who wish to analyze the full database offline. It includes documentation and variable descriptions. To get access, please contact us with a short description of your project and intended use of the data.

Why does CARMA treat dams and nukes the same as solar and wind power? CARMA focuses on one particular dimension that is important to the environment: carbon emissions from power production. For this reason, we treat hydroelectric and nuclear power the same as other forms of carbon-free energy like solar and wind. This does not mean that dams and nukes are free from serious environmental problems. CARMA does not endorse or favor any particular technology.

Why isn’t the fuel type of individual plants made available? Proprietary licensing agreements with some of our data suppliers prevent us from revealing the fuel sources (coal, gas, nuclear, etc.) of individual plants. For more information, see our page on Plant-Specific Information.

What do terms like Intensity, MWh Produced, and Future mean? The glossary provides complete definitions of the terms used on CARMA. Also, every plant, company, and region page includes a section (beneath Power Trends) with links to explanations and additional information.

How can I get my hands on the data? allows users to search and browse their way through the database. For those who are interested in downloading data for further analysis, many pages include the option to download a .csv file of the displayed data. If you are interested in using CARMA’s data on your own website, we have an Application Programming Interface (API) to facilitate the process. Those interested in obtaining the full datasets for academic or professional research should Contact Us. Any data used in print or web publication should cite CARMA accordingly. Users can also download the full database or large subsets using the Dig Deeper tool and clicking “Download dataset”.

I’ve entered a city or state/province in the search box without success. What’s wrong? CARMA’s geocoding of power plants includes more than 13,000 geographic entities. The search field is not case-sensitive, but it does require correct spelling. In some cases, the database may contain the spelling dictated by the local language. For example, the city of Ghent in the Flanders region of Belgium is entered in the database under its Dutch spelling: Gent.