Last modified on Jun 04, 2026 14:03

Contributors: B-Open and TCDF

Issued by: B-Open

Issued Date: 06 May 2025 

Ref: C3S3_430a – ECDE maintenance and development

Official reference number service contract: 2021/C3S2_430a_BOPEN

Acronyms

1. Introduction

1.1. Executive summary

This dataset provides a series of climate indicators derived from reanalysis and model simulations data hosted on the Copernicus Climate Data Store (CDS). These indicators describe how climate variability and change of essential climate variables can impact sectors such as health, agriculture, forestry, energy, tourism, or water and coastal management.

Those indicators are relevant for adaptation planning at the European and national level and their development was driven by the European Environment Agency (EEA) to address informational needs of climate change adaptation national initiatives across the EU and partner countries as expressed by user requirements and stakeholder consultation. The indicators cover the hazard categories introduced by the IPCC and the European Topic Centre on Climate Change Impacts, Vulnerability and Adaptation (ETC-CCA). 

The indicators have either been calculated through a specific workflow using CDS dataset in input or directly retrieved From the CDS when already available. In this way both the calculations and the resulting data are fully traceable. As they come from different datasets the underlying climate data differ in their technical specification (type and number of climate and impact models involved, bias-corrected or not, periods covered etc.). An effort was made in the dataset selection to limit the heterogeneity of the underlying dataset as ideally the indicators should come from the same dataset with identical specifications.

The indicators related to temperature, precipitation and wind (20 out of 30) were calculated from atmospheric variables in the same datasets: 'Climate and energy indicators for Europe from 2005 to 2100 derived from climate projections', and 'ERA5 hourly data on single levels from 1940 to present'. The other indicators are directly available from CDS datasets generated by specific theme projects.

1.2. Scope of documentation

This document provides a description of the indicators included in the European Climate Data Explorer and details of the dataset and the methodology used to produce them. First, the product requirements (section 2.1) against which this dataset was developed is presented, followed by the definition of the indicators (2.2 product overview) and the input datasets (2.3 Input datasets) used to calculate them. Thereafter, the methodology (2.4 Method) to produce the indicators forming this catalogue entry is described.

1.3. Version history 

This is the second version of the dataset. Section 3 Version changes provides the changes occurred for each indicator (if any) between v1.0 and v2.0.

2. Product description

2.1. Product requirements 

The development of those indicators was driven by the European Environment Agency (EEA) to address informational needs of climate change adaptation national initiatives across the EU and partner countries as expressed by user requirements and stakeholder consultations. They are relevant for adaptation planning at the European and national level and cover the hazard categories introduced by the IPCC and the European Topic Centre on Climate Change Impacts, Vulnerability and Adaptation (ETC-CCA, [1]). They are made available interactively through public visualisation apps on the European Climate Data Explorer hosted on EEA’s Climate-adapt site. 

2.2. Product overview 

2.2.1. Data Description

Table 1: Overview of key characteristics of the ECDE indicators.

2.2.2. Climate indicator definition

Table 2: Definition of the ECDE indicators.

2.2.3. Input datasets by indicator

The table below provides the input dataset used to derive each indicator. Each single dataset is described in more detail in section 2.3. For conciseness and clarity we have used a consistent set of short names for the datasets in the table below and in section 2.3.

Table 3: List of the ECDE indicators input datasets.

2.3. Input datasets

2.3.1. ERA5 and derived products. 

The ERA5 reanalysis is regarded as a good proxy for observed atmospheric conditions and currently covers 01/01/1940 to near real time and is regularly extended as ERA5 data become available. 

The historical values of the indicators were derived from hourly data from the ERA5 reanalysis (ERA5 single levels) whenever possible. This was the case for the indicators related to temperature, precipitation and wind (20 out of 30). For the remaining indicators, values over the historical period (either simulated or from reanalysis when available) were used directly.

The historical values of Days with high fire danger and Fire Weather indices are from the Fire danger indices historical data from the Copernicus Emergency Management Service dataset, that is based on the ERA5 reanalysis and updated in near real time. It is produced by the Copernicus Emergency Management Service (CEMS) for the Global ECMWF Fire Forecasting model (GEFF) and the European Forest Fire Information System (EFFIS). More technical specifications can be found in the dataset documentation.

2.3.2. SIS Energy

The simulated indicators related to temperature, precipitation and wind (20 out of 30) were calculated from daily atmospheric variables in the same climate projections dataset: Climate and energy indicators for Europe from 2005 to 2100 derived from climate projections. It is a set of bias-adjusted EURO-CORDEX projections composed of 9 GCM-RCM simulations at 0.25° x 0.25° spatial resolution, 3-hourly temporal resolution and cover emission scenarios RCP4.5 and RCP8.5. The 9 combinations of the 5 GCMs with the 5 RCMs is given in Table 1. More technical specifications can be found in the dataset documentation. 

Table 4: The 9 GCM/RCM combinations of the Climate and energy indicators for Europe from 2005 to 2100 derived from climate projection dataset

2.3.3. ERA5 Heat

The High UTCI Day index data are from Thermal comfort indices derived from ERA5 reanalysis. As indicated, it is based on surface variables from the ERA5 reanalysis (ERA5 single levels) and inherits the same spatial (0.25° x 0.25°) and temporal resolution (hourly). There are no climate projections of UTCI at the moment. More technical details can be found in the dataset documentation. 

2.3.4. SIS Operational Water Service

The Flood recurrence, Mean river discharge, Aridity actual, and Duration of Soil moisture Draughts index data are from the Hydrology-related climate impact indicators from 1970 to 2100 derived from bias adjusted European climate projections dataset. It is a set of 30-year statistics from two hydrological models forced by 8 bias-adjusted multi-model simulations from the EURO-CORDEX experiment. The hydrological models are from the Swedish Meteorological and Hydrological Institute (SMHI, E-HYPEgrid model) and Wageningen University (VIC-WIR Model). The hydrological simulations are either gridded (5km x 5km) or at catchment scale and cover scenarios RCP4.5 and RCP8.5. The 8 combinations of the 5 GCMs with the 5 RCMs is given in Table 1. More technical details about the hydrological models can be found in the dataset documentation and in the following Hydrological model specification. 

Table 7: The 8 GCM/RCM combinations of the water related indicators dataset.

2.3.5. SIS EU Tourism

The Days with high fire danger and Fire Weather index data are from the Fire danger indicators for Europe from 1970 to 2098 derived from climate projections dataset. It is a set of 6 bias-adjusted multi-model simulations from the EURO-CORDEX experiment. These simulations have a daily temporal resolution, a spatial resolution of 0.1° x 0.1° and cover scenarios RCP4.5 and RCP8.5. The 5 combinations of the 5 GCMs with 1 RCM is given in Table 1. More technical specifications can be found in the dataset documentation. 

Table 8: The 5 GCM/RCM combinations of the Fire danger indicators dataset

Note: The Fire Weather Index values were bias-corrected in this project to be compatible with the CEMS reanalysis FWI. See subsection 2.4.2 Indicator calculation.

The Snowfall amount indicator data (both historical and simulated) are from the Mountain tourism meteorological and snow indicators for Europe from 1950 to 2100 derived from reanalysis and climate projections dataset. The dataset is based on the UERRA reanalysis and a set of 9 bias-adjusted multi-model simulations from the EURO-CORDEX experiment. The indicator data simulations have an annual temporal resolution, a spatial resolution over NUTS3 regions, a vertical resolution of 100m and cover scenarios RCP4.5 and RCP8.5. More technical specifications can be found in the dataset documentation. 

Table 9: The 9 GCM/RCM combinations of the snow indicators dataset.

2.3.6. SIS European Storm Surges

The Relative sea level rise and the Extreme sea level data are from the Global sea level change time series from 1950 to 2050 derived from reanalysis and high resolution CMIP6 climate projections dataset. It is based on a set of simulations produced with the Global Tide and Surge Model (GTSM) of Deltares, a global 2D hydrodynamic model which incorporates tides, surges and mean sea-levels dynamically. Both the historical and future GTSM simulations include sea level rise data as input that are also available in the dataset. The Relative sea level rise field is annual and spatially-varying at 1° x 1° resolution and is relative to the 1986-2005 reference period based on RCP8.5 (not SSPs as for the GTSM simulations). The field is the median result of a probabilistic model that computes and combines processes affecting sea level. As such, it is model independent and only one field (the median) of SLR is provided in the dataset. More technical specifications can be found in the dataset documentation.

2.4. Method

2.4.1. Input dataset selection

The indicators are either retrieved from datasets when available or calculated through a specific workflow. In this way both the calculations and the resulting data are fully traceable. As they come from different datasets the underlying climate data differ in their technical specifications (type and number of climate and impact models involved, bias-corrected or not, periods covered etc.). An effort was made in the dataset selection to limit the heterogeneity of the underlying dataset as ideally the indicators should come from the same dataset with identical specifications.

2.4.2. Indicator calculation

The indicators were calculated according to the recommended definitions based on a technical paper from the European Topic Centre on Climate Change Impacts, Vulnerability and Adaptation (ETC/CCA). The indicators are either retrieved from datasets when available or calculated through a specific workflow. In this way both the calculations and the resulting data are fully traceable. 

One exception is the Fire Weather Index (FWI) data of the SIS Tourism dataset. A bias was found between the original values and the values based on reanalysis (CAMS) and it was decided to apply a bias-correction procedure to correct the mean and the variance of the simulated FWI with the mean and variance of the reanalysis based FWI and taking the 1991-2020 period as reference (see Appendix 1).

2.4.3. Regional aggregation

Where relevant the indicators have been aggregated over standard administrative bounderies used by european institution for reporting. The aggregation was performed using the eartjkit aggregation tools. The administrative boundaries available are:

• the NUTS classification (Nomenclature of territorial units for statistics) maintained by Eurostat (see Eurostat (2021) website). NUTS is a hierarchical system for dividing up the economic territory of the EU and the UK. The indicators uses NUTS regions ranging from NUTS0 (country) to NUTS2 (sub-country) and NUTS3 in the specific case of Snowfall amount. 
• Europe zones is the number of countries considered for the European domain (27 EU countries, 32 EEA member countries and 38 EEA member and cooperating countries).
• Transnational regions regions involve cooperating regions from several countries of the EU forming bigger areas (e.g. danube, alpine, ionian etc.) to promote better cooperation and regional development within the Union

Appendix 2 lists all the available regional layers.

2.4.4. Limitations

The EEA ECDE initiative gathers in a single dataset a number of indicators that are relevant for adaptation planning at the European and national level. Ideally the indicators should come from the same dataset with identical specifications, as this was not possible considering the available datasets hosted on the CDS an effort was made in the dataset selection to limit the heterogeneity of the underlying dataset. Out of the 30 ECDE indicators in this catalogue entry, 19 of them use the same input dataset (SIS Energy and ERA5 Single level, see Table 3) and provide a consistent core of indicators. The other indicators required the use of different datasets because they require more sector specific data (fire, hydrology, tourism) and are therefore not directly comparable. In particular, differences in future behaviour (trends, variability, etc..) between indicators using different input datasets can be, at least partially, attributed to the underlying climate projection included in each dataset. Besides the possible heterogeneity between indicators it is advised to consult the Product User Guide of individual indicators to understand their own limitations. 

2.4.5. Validation and quality assurance

The strategy behind the calculation of the ECDE indicators is to use Quality Assured and validated datasets distributed through the CDS and use fully traceable and repeatable workflows to perform the computation. All the datasets used in input and listed in 2.2.3. Input Datasets by Index have gone through the CDS quality assurance and are under C3S governance and scrutiny should any error be found in the future. This strategy eliminates the uncertainties or errors due to the input data or tools used.

The workflows to calculate the indicators and their documentation could be subject to errors and to minimise them the quality assurance procedure described below has been followed. This process included several reviews by different responsible at various steps during the computation and publication of the indicators.

Figure 1: ECDE internal quality assurance and review process

Regarding the validation of specific indicators the validation has mainly been performed by comparing the output with either the already existing indicators available on the ECDE or other publication from the EEA where the same indicators were presented and computed different input data. Even though it was not necessarily possible to check against a numerically identical dataset it was possible to check that the indicators are consistent with other publications.

3. Version changes

The main updates of each catalogue version are listed in the table below.

Table 10: Catalogue entry overall change log

The table below provides all the indicator specific changes between v1.0 and v2.0.

Table 11: List of the ECDE indicators and their changes

Known issue:

In the original v1.0 release the variables Flood Recurrence and Mean River Discharge were distributed in their original grid (Lambert azimuthal equal area grid). At publication of v2.0 the regular grid version of the variables was preferred as it is more user friendly and consistent with all the other variables. So the current v1.0 data for variables Flood Recurrence and Mean River Discharge does not correspond to the original v1.0 which was published in its non regular grid. Users wishing to download the data in the original grid version can do from the CDS catalogue entry for hydrology-related climate impact indicators from which the indicators are derived.

4. Concluding remarks

The ECDE indicators dataset provides a series of climate indicators derived from reanalysis and model simulations data hosted on the CDS. It gathers data both in a gridded format and as regional averages that are used to make the indicators available through interactive visualisation apps on the European Climate Data Explorer hosted on EEA’s Climate-adapt site. It is provided as a dataset to support further analysis by climate change adaptation practitioners that wish to have access to the underlying data of the ECDE. 

5. Appendix 1: Fire weather index bias correction

Several ECDE plots of the indicators include the comparison of the simulated and reanalysis based values. In the case of the Fire Weather Index (FWI), the simulated values over the historical period show a systematic bias against reanalysis based values making the comparison non-realistic to the user. To be able to facilitate the comparison, the simulated values were adjusted with a simple approach consisting in correcting the mean and scaling the variance of the simulated time-series as a the implementation of a state-of the art bias-adjustment method goes beyond the scope of this project. This document describes the specifications of the simple bias-correction method that was implemented.          

5.1. Simple bias correction:  mean and variance adjustment 

The computing steps of the simple adjustment approach can be divided in three main categories:    

5.1.1. Pre-processing: converting FWI to DSR

Because the FWI is not suitable for averaging over space or time [2], a nonlinear transformation into the Daily Severity Rating (DSR) is necessary (DSR = 0.0272 FWI1.77). The DSR is intended to be directly proportional to the expected effort required for fire suppression and control and is suitable for space or time averaging. As a consequence, the FWI daily values are first converted to DSR to which the simple bias-correction procedure is applied and then DRS values are converted back to FWI.   

• Step 1: FWI to DSR conversion 

5.1.2. Mean and variance calculation

The second series of steps consists in calculation the necessary parameters for each time-series over the reference period for all simulations. The reference period was set to 1981-2010. 

• Step 2: Compute the average of each calendar month (climatology) of the model time-series over the reference period.
• Step 3: Compute the standard deviation of each month of the model time-series over the reference period.
• Step 4: Compute average of each calendar month (climatology) of the reanalysis time-series over the reference period.
• Step 5: Compute the standard deviation of each calendar month the reanalysis time-series over the reference period.

5.1.3. Processing

The third series of steps is to perform the adjustment with the previously calculated variables at each grid point. The processing is applied for each calendar month separately. 

• Step 5:  Compute the daily 30 year running mean. 
• Step 6: Detrend the time series by subtracting the 30 year daily running mean from the daily projected values.
• Step 7: Adjust the variability of the previously detrended time-series by the standard deviation ratio (reanalysis value divided by the model value). For a model standard deviation below 0.1 (corresponding to an FWI close to 0) the ratio is set to 1 to avoid diverging values.
• Step 8: Retrend the time-series by adding the 30 year running mean to the variance adjusted values.
• Step 9: Compute the anomaly between the re-trended values and the projections monthly climatology.
• Step 10: Sum the anomalies to the reanalysis climatology to obtain the bias corrected values

In the ECDE the bias corrected DSR is saved as an ancillary variable after step 10.

5.1.4. Post-Processing

The the mathematical operations performed for the detrending and retrending can introduce negative DSR values. These possible negative values are invalid as the DSR must be positive, in particular to be converted to FWI. The negative values are simply set to zero to remain in the validity range of the DSR. To prevent loosing the information on where negative values are present, the correction is applied downstream in the ECDE workflows where the bias-corrected DSR is used.

• Step 11: Set the negative values to 0. 

Finally convert the adjusted DSR values back to FWI values

• Step12:  DSR to FWI conversion

5.2. Limitations

The simple bias-correction approach, although it reduces the systematic bias, has inherent limitation du to the simplicity of the approach.  Figure 1 illustrates a favorable case where the simple approach reduces the bias and brings the corrected time series in better agreement with the reanalysis. Figure 2 shows a case where the simple approach is not capable to correct the model values that are relatively high (0-60) compared to the reanalysis values that are relative low (0-20) with significant differences in interannual variability. It also illustrates a case where this simple adjustment method introduces negative values that can't be left unchanged and are set to zero when used in further calculations.  

Figure 2: Time series illustrating the bias-reduction of the between model simulations and observations (reanalysis). "Raw DSR reanalysis" are the observed values, "Raw DSR projections" are the original model values, DSR-bc is the corrected model values, and DSR-bc-zero is the later with negative values set to zero.   

Figure 3: Same as figure 2 but for an unfavourable case where the bias-corrected values are not efficiently corrected. Note the presence of significant negative values in the DSR-bc values. 

6. Appendix 2: Regional layers

The table below lists all the regional layers available and the corresponding code to use in the file naming conventions.

7. References

[1] Crespi A., Terzi S., Cocuccioni S., Zebisch M., Berckmans J., Füssel H-M (2020) “Climate-related hazard indices for Europe”. European Topic Centre on Climate Change impacts, Vulnerability and Adaptation (ETC/CCA) Technical Paper 2020/1. DOI: https://doi.org/10.25424/cmcc/climate_related_hazard_indices_europe_2020

[2] Van Wagner, C.E., 1987. Development and structure of the Canadian Forest Fire Weather Index System, Forestry Technical Report (CFS - Ottawa). Canadian Forestry Service, Headquarters, Ottawa. http://cfs.nrcan.gc.ca/publications?id=19927. 

8. Related articles

• Page:
  ECDE indicators catalogue entry v2.0: Product User Guide
• Page:
  European Climate Data Explorer (ECDE) indicators