CH Estimation Model
The data fetched from ENTSO-E for CH is incomplete. For example, it is known that Switzerland has a lot of small run-of-river plants that are so small that they do not have to report live data, so that this data is not available. A first initiative was implemented to account for this missing data, as seen here. As a result, the missing production data was aggregated into the unknown production mode, whose emission factor was adapted to reflect the makeup of that unreported production. Members of the Swiss Bureau of Energy (SFOE) vetted this value.
While this approach provides accurate carbon intensity values for the electricity on the Swiss grid, it prevents Electricity Maps from providing accurate values for the share of electricity coming from renewable sources.
For example, looking at a snapshot of the mix:
We can see that the renewable percentage is vastly underestimated if we account that more than 70% of the unknown electricity is actually coming from renewable sources.
Hence the introduction of the Mode Breakdown model.
The goal of the model is to extract the part of the unknown production attributed to renewable sources and assign it to the respective production modes.
This affects solar, biomass and hydro production.
- The solar production is highly variable depending on meteorological conditions. Furthermore, the data source provides unreliable solar production data. For example 164 and 58 hours of solar production were missing in 2021 and between January and April 2022 respectively. We thus resulted to training a model of its own to replace entirely the solar production with estimates. See details for the calibration here
- For hydro and biomass, the part of unknown attributed to each renewable mode is directly subtracted from unknown and added to the respective mode.
For an overview of how the model generates a better production breakdown, see the following figure:
Finally, it is necessary to update the emission factor for the unknown Swiss production, as following these estimates, its makeup is modified significantly. See calculations here that lead to an average unknown emission factor of 383 gCO2eq/kWh (lifecycle)