I was looking for information on the Australian electricity markets when I stumbled upon www.carma.org . This website will provide you with very useful data on how much electricity is generated and how much carbon is emitted for over 50,000 power plants across the world. It also comes handy that you can download the data in a .csv format and then manipulate it in Excel or other spreadsheet software that you might like.

CARMA, which stands for carbon monitoring for action, is the product of the efforts of the Washington DC based think tank Centre for Global Development. A very good effort indeed.

However the blog post would not be complete without pointing out some shortcomings of the data, and suggestions on how to improve it. The most striking piece of data that is missing is the type of power plant. While you could sometimes guess from the name of the power plant or the company running it, it is not always easy to say and it is difficult to manipulate the data using this criteria.

Another important aspect to point out is that some of this data is estimated based on models. To quote their website

“For non-reporting plants, CARMA estimates emissions using a statistical model that has been fitted to data for thousands of reporting plants in the U.S., Canada, the EU, and India. The model utilizes detailed data on plant-level engineering and fuel specifications.”

Does this not mean that they do have the data on what type of plant it is and which fuel it uses, not to mention the detailed data on plant-level engineering? It would be nice if we can also see this data. The more data the better as long as we can easily process it. We should also take the estimated data with a pinch of salt. There is an underlying assumption here that all power plants in the world are the same and operated in the same manners. The model has already decided that US, Canada, EU and India are representative of the whole world. Using the data to try and find abnormalities in a non-reporting country is therefore futile as the data has been estimated assuming normality. We would go in circles.

Overall the data comes in very handy and is potentially very useful, however it would be nice to see less modelling and more hard data. The efforts that went into this must have been enormous, and they are appreciated.