Contributors: A. Troccoli (WEMC), L. Sanger (WEMC), C. Goodess (WEMC), J. Ogonji (WEMC), L. Dubus (WEMC), R. Vautard F Pons (CEA), X. Jin (CEA), G. Levavasseur (CEA), R. Legrand (MF), L. Grigis (MF), S. Martinoni-Lapierre (MF), C. Viel (MF), S. Parey (EDF), B. Oueslati (EDF), Y-M. Saint-Drenan (ASSOCIATION POUR LA RECHERCHE ET LE DEVELOPPEMENT DES METHODES ET PROCESSUS INDUSTRIELS, ARMINES, FRANCE), J. Mendes (MO), J. Osborne (MO), G.Guentchev (MO)
1. General
1.1. Description
Electricity demand is computed via a statistical model using 3 climate variables as predictors: 2 m air temperature (TA), solar radiation (GHI) and Wind Speed at 10 m (WS10).
EDM defined as the daily energy (units=MWh) consumed on the grid. Electricity demand is also provided as daily mean power (units=MW). Daily mean power = daily energy / 24 (hours)
They are also provided at monthly, seasonal and annual time scale. For NRG, the aggregation is a sum, whereas for PWR, it is an average.
Individual models were built for each European country, using Generalized Additive Models (GAMs), and ERA5 data as climate inputs, as NUTS0 averages. The models are built using recent years ENSTO- E data, varying from 2006 and 2018 included depending on the country. In each case, the models were set up and validated on independent periods. The trend is estimated on the longest possible period, then fixed, so that the final product, which is the EDM reconstructed on 1979-2019, has no long-term trend, and a mean level corresponding to that of the training period. In a later stage, we may provide demand data has anomalies with respect to a reference which needs to be specified.
In any case, based on the definition above, users can define their own reference and calculate their own anomaly dataset, then any future scenario for projections.
One of the predictors of energy demand is a smoothed temperature over a period which varies depending on countries. For seasonal forecasts, this means that temperature data is necessary for the days before the start of the seasonal forecast model run. The choice has been made to use ERA5 temperature data. As ERA5-T is available with a few days delay, this means that the seasonal forecasts for demand can be run only a few days after the seasonal climate variables are available.
EDM is available for three streams:
Historical: This dataset uses climate predictors from the ERA5 reanalysis and retrieved from the CDS and averaged at NUTS0 (country). No bias adjustment has been applied. |
Seasonal Forecasts: This dataset uses climate predictors from ECMWF System 5, Météo-France ARPEGE System 7 and Met Office GloSEA5 System 14. Seasonal Forecasts are retrieved from the CDS and are subsequently bias adjusted via quantile mapping. The forecast datasets are updated monthly when the raw forecast of each system is available and averaged for NUTS0 (country). |
Projections: This dataset uses climate predictors 11 regional climate model (RCM)/global climate model (GCM) combinations for the EURO-CORDEX project (Table 1, Appendix) and for two Representative Concentration Pathways (RCP4.5 and RCP8.5), with RCP2.6 also added for two of the RCM/GCM combinations. Historical forcing is used from 1979-2005, and RCP forcing from 2006-2100. The original EURO-CORDEX simulations are retrieved from the producers (some were already available on the Earth System Grid Federation (ESGF) portal) and are subsequently bias adjusted by applying the Cumulative Distribution Function transform (CDFt) method. The gridded |
1.2. Units
Electricity demand (energy): MWh, labelled as NRG in the output files. Electricity demand (mean power): MW, labelled as PWR in the output files.
1.3. Links
Data will eventually be available via the C3S Energy/CDS demonstrator.
1.4. Data format
NUTS0 are available as CSV files.
1.5. Keywords
Electricity Demand (EDM) Load, Power, Energy, ENTSO-E, Generalized Additive Models.
1.6. Contact
Please raise a ticket through the ECMWF Support Portal (ECMWF login required).
2. Dataset coverage
2.1. Geographic area
European countries within the C3S Energy domain is: 26.5° N to 72.5° N by 22.0° W to 45.5° E. Specifically, there 32 European countries have been considered, based on the availability of daily load data at the time of models training): AT, BA, BE, BG, CH, CZ, DE, DK, EE, EL, ES, FI, FR, HR, HU, IE, IT, LT, LU, LV, ME, MK, NL, NO, PL, PT, RO, RS, SE, SI, SK, UK. Depending on ENTSO-E and other input data availability, four additional countries might also be considered in the future: Albania (AL), Cyprus (CY), Iceland (IS) and Turkey (TR).
2.2. Temporal resolution
Daily, monthly, seasonal (DJF, MAM, JJA, SON) and annual mean power (PWR, in MW) and cumulated energy consumption (NRG, in MWh).
2.3. Time period
Historical: 1979 to present (ca. 1 month behind real time); updated every month based on ERA5T data availability. | |||
Seasonal Forecasts: The hindcasts cover the 1993-2016 period for the three models used by C3S Energy: ECMWF Sys5, Météo-France Sys7 and UK MetOffice Sys14. ECMWF has all the month, MTFR SY07 currently covers October to April with the rest of the months to be completed by mid- May 2020 as this is a new system. METO SY14 ranges from June to April (notice that METO SY14 is a rolling release model, hindcast have to be computed each months). The forecasts have been computed every month from the first month of availability in the current version on the CDS: | |||
Forecast availability | |||
ECMWF Sys5 | From January 2017 | ||
MTFR Sys7 | From October 2019 | ||
METO Sys14 | From June 2019 | ||
Projections: 1971-2100 | |||
2.4. Spatial resolution
Data is available at NUTS0 level for the 32 countries listed above.
3. Lineage statement
3.1. Original data source
Energy data: ENTSO-E Power Statistics
Historical: ERA5 |
Seasonal Forecasts:
|
Projections: EURO-CORDEX Climate indicators (air temperature, wind speed and solar radiation) available via the Earth System Grid Federation (ESGF) portal. |
3.2. Tools used in the production of indicators
R Studio Version 1.1.463 – © 2009-2018 RStudio, Inc. R version 3.6.1 (2019-07-05)
R packages (latest versions): chron, corrr, easyNCDF, fields, glue, lubridate, maptools, ncdf4, raster, readxl, rgdal, rgeos, seas, sf, shapefiles, sp, tidyverse
NCO netCDF Operators version 4.7.5: ncks
4. Data Quality
The EDM data reconstruction quality has been assessed for each individual country over the historical period, using ENTSO-E data.
5. Appendix
Table 1: List of the 11 EURO-CORDEX simulations and the institutes that provided the data.
Short | Driving | RCM | Contact Institute | Period |
HIIC | ECEARTH | HIRAM | Danish Meteorological Institute (DMI) | 1951 - 2100 |
RAIC | ECEARTH | RACMO | Royal Netherlands National Meteorological Institute (KNMI) | 1950 - 2100 |
RCIC | ECEARTH | RCA | Swedish Meteorological and Hydrological Institute (SMHI) | 1970 - 2100 |
RAMO | HADGEM | RACMO | KNMI | 1951 - 2098 |
RCMO | HADGEM | RCA | SMHI | 1970 - 2098 |
ReMO | HADGEM | REGCM | International Centre for Theoretical Physics (ICTP) | 1971 - 2099 |
WRIP | IPSL | WRF 381P | Institute Pierre Simon Laplace (IPSL) | 1951 - 2100 |
RCMP | MPI | RCA | SMHI | 1970 - 2100 |
CCMP | MPI | CCLM | Climate Limited-area Modelling Community | 1950 - 2100 |
HINC | NORESM | HIRAM | Danish Meteorological Institute (DMI) | 1951 - 2100 |
ALCN | CNRM | ALADIN63 | CNRM (Centre National de Recherches Meteorologiques, Toulouse 31057, France) | 1952 - 2100 |
