A researcher at the Centre for Sustainable Communications (CESC) based at the KTH Royal Institute of Technology in Stockholm, Sweden, has developed an application to calculate the carbon footprint of individual websites.
It’s called Greenalytics and matches web statistics with environmental data to show the emissions generated by the infrastructure behind a website. It’s effectively a mash-up, compiling information from various sources to calculate a result in real time. It uses statistics from Google Analytics to measure how many visitors a site has and environmental data on electricity consumption from servers, routers and from users, from research carried out by KTH.
In fact the application is running live at http://www.greenalytics.org/sites, showing the total emissions for a number of Swedish web sites for the year to date. Apparently KTH’s own site, which is among the top 100 in Sweden, produced seven tonnes of carbon dioxide last year, corresponding to a climate impact of 40,000 kilometres by car or 70 hours of air travel. So far this year the figure stands at 58.0 kg of CO2 (this is clearly a year-to-date figure – just three days - rather than the last 12 months as it says on the site).
Jorge Zapico, the researcher who produced Greenalytics, had some tips on how to reduce a website’s climate emissions; ”The electricity source for the server is the most important factor that can easily be influenced. Choose a server located in a country with a climate-friendly mix of electricity sources such as Sweden. If the server is powered by electricity from renewable sources, it is even better. It is also important to build light websites that do not need to load heavy content”.
An interesting project because of the complexity it needs to deal with. On the user side, the Google data provides the number, location and duration of visit and the CO2 is calculated by aggregating the impact per country - the total time visitors from that country spent on the site multiplied by how much electricity their computer consumes multiplied by the electricity factor (how much carbon dioxide is emitted per electricity unit) of that country. But there are also built-in assumptions and generalisations, such as the mix of devices accessing the site and the use of annual, national emissions factors for energy used. On the infrastructure side, emissions are calculated based on the energy used per unit of data transmitted, which itself includes a number of other assumptions and is in any case a generic figure.
It does show how complex carbon footprinting can be, particularly in a world of cloud computing and on-line business. But if organisations are to be required to take responsibility for their emissions, it’s an issue that has to be faced.