Rising demand for individual carbon footprints
Individual initiative in the fight against climate change is possible in many areas and ranges from reduced meat consumption to the renunciation or reduced use of cars with combustion engines to sustainable financial investments. Whether it is environmentally conscious end consumers or companies that want to make their contribution by acting sustainably – in the private as well as in the commercial and industrial sectors, interest in one’s own carbon footprint is increasingly growing. The sustainability aspect now also plays a central role in the energy supply sector.
Data on greenhouse gas emissions
Dynamic Greenhouse Gas (GHG) emission factors in combination with electrical load profiles enable the calculation of individual GHG footprints. The individual amount of electric energy purchased from the grid multiplied by the emission factor for the respective region (e.g., a country) is a reasonable indicator for the emissions caused by power consumption. In this way, the individual GHG footprint of consumers can be determined ex post and, if necessary, settled via service providers.
If one wants to minimize GHG emissions already at the time of electricity consumption, it is possible to use the day-ahead forecast. The data can be integrated into smart home applications, for example, to control high-consumption appliances such as washing machines and dryers so that they are used when the GHG emissions in the electricity mix are lowest.
To enable and promote sustainable behavior, two datasets are provided here on the FfE opendata platform. The ex post values are calculated using actual energy production by energy source, while the day-ahead forecast is generated by utilizing (among others) data on day-ahead production of renewables. Figure 1 displays both ex post and ex ante emission factors for the past 14 days. Both datasets are automatically updated on a daily basis between 8 and 10 pm (CET/CEST).
The code associated with these datasets is open-source and publicly available on GitLab.
Emission factors (ex post) are calculated by offsetting the amount of actually produced electricity by energy source  against the specific emissions of that energy source . The specific emissions already account for the pre-chain, which is why wind energy, for example, also has a certain carbon footprint.
The forecast of GHG emissions in the electricity mix is performed using machine learning techniques. For this purpose a Random Forest Regressor has been trained using historical day-ahead forecasts of wind and photovoltaic generation as well as electricity prices as well as temporal features like the day of the week. Provided with up-to-date forecasts and additional features, the pre-trained model is able to predict GHG-emissions in the german electricity mix for the next day.
In order to account for the uncertainty of the prediction, two additional values are included besides the actual predicted value. The upper and lower bounds serve as an indicator on how certain the model is with the prediction.
The software used as an interface to the ENTSO-E Transparency Platform is based on freely available code from the project electricitymap.org.
Since the datasets are the result of an automatically performed calculation based on data from the ENTSO-E Transparency Platform, completeness and correctness cannot be guaranteed, since the input data may be incomplete or incorrect.
 The ecoinvent Database, Version 3.6: www.ecoinvent.org; Zürich: ecoinvent, 2019.
 European Network of Transmission System Operators for Electricity: Transparency Platform. [Online] https://transparency.entsoe.eu/, continuous updates since 2014.