Contributors:  A. Troccoli (ICS), M. Borga (ICS/UNIPD), M. Zaramella (ICS/UNIPD), G. Aldrigo (ICS), R. Bortolami (ICS), R. Ciceri (ICS), S. Cordeddu (ICS), L. Lusito (ICS), E. Restivo (ICS), S. Strada (ICS), C. Zanetti (ICS), Y-M. Saint-Drenan (ARMINES), R. Amaro e Silva (ARMINES), S. Parey (EDF), S. Claudel (EDF), H. Upton (EDF), P. Boorman (WEMC), K. Nielsen (WEMC)

History of modifications 

List of datasets covered by this document 

Acronyms and abbreviations

Introduction

This documentation describes the Global climate and energy indicators from 2015 to 2100 derived from CMIP6 projections dataset. The Coupled Model Intercomparison Project Phase 6 (CMIP6) is a global climate modelling initiative that provides state-of-the-art simulations of past and future climate conditions. This dataset has been co-designed within the Copernicus Climate Change Service (C3S) Energy service to support a wide range of climate and energy-related applications.

The dataset includes both gridded (NetCDF) and spatially aggregated (CSV) indicators, computed using consistent and reproducible methods. Climate indicators such as temperature, precipitation, wind speed, and solar radiation form the basis for estimating energy indicators related to wind power, solar photovoltaic, hydropower, and electricity demand.

This documentation describes the full data-processing workflow, from data retrieval and pre-processing to bias correction, indicator computation, and spatial and temporal aggregation. Additional technical details on the energy conversion models and climate-data processing tools are available in the following dedicated external pages, which have been linked along this document in the relevant sections:

• Tools for Climate Data Processing: explains the wind-profile scaling based on the power law, the computation of exclusion areas, and the spatial and temporal aggregation procedures.
• Energy Conversion Models: describes the energy data used to calibrate the models and provides detailed information on the methodologies used for wind power, solar photovoltaic power, hydropower, and electricity and energy-demand estimation.

Overview of the Workflow

The generation of climate and energy indicators follows a structured workflow that transforms raw climate projections data into processed, user-ready outputs. The processing chain includes the following main steps:

• Data retrieval and pre-processing: CMIP6 data (temperature, precipitation, wind speed, and solar radiation) are retrieved through the Earth System Grid Federation (ESFG) nodes. Pre-processing includes spatial and temporal interpolation procedures that are applied to all the variables to obtain datasets with the same resolutions of the ERA5 reanalyses (1-hour temporal resolution and 0.25° grid). This ensures homogeneity and coherence with the historical data stream. Wind components are combined to derive wind speed at 10 m height. See Section 2.1 and Section 2.2.

• Bias adjustment: All the climate variables are bias-adjusted using ERA5 reanalyses as reference, with different methodologies depending on the variable (see Section 2.3).

• Computation of the wind speed at 100 m: The wind power law is employed to derive the wind speed at 100 m height, starting from the wind at 10 m. See Section 2.4. 

• Computation of climate indicators: The climate variables are processed into indicators at 0.25°x0.25° grid resolution. These indicators are then spatially aggregated to national and sub-national levels (ADMIN 0 and ADMIN 1) and temporally aggregated to daily, monthly, seasonal, and annual scales. See Section 2.5.

• Computation of energy indicators: Climate indicators, both in gridded and aggregated form, are used as input to estimate potential generation and demand, together with input energy data employed for calibration and validation (see Section 3.1). Separate models are used for wind power, solar photovoltaic power, hydropower, and electricity demand (see Section 3.2). Some of these models vary by region depending on energy data availability. All the energy indicators are provided as aggregated files, while only some of them are also available in gridded form. For more details, please refer to Section 3.3.

A schematic representation of the overall workflow is shown in Figure 1.1.

Figure 1.1: Workflow for the projection data stream, illustrating the processing chain from CMIP6 climate data to gridded and aggregated climate and energy indicators.

Computation of the Climate Indicators

Projection Models and Scenarios Selection

A key step in the workflow is the selection of a suitable subset of climate projections for applications in the energy sector. The analysis focuses on global climate models participating in the sixth phase of the Coupled Model Intercomparison Project (CMIP6; Eyring et al., 2016). Compared to the Coordinated Regional Climate Model Experiment (CORDEX), CMIP6 provides a larger and more consistent set of global projections, based on the latest generation of climate models.

From more than 100 available CMIP6 model configurations, only a subset can be used operationally, due to computational constraints and differences in variable availability across the ESGF nodes or the Climate Data Store (CDS). A subset of CMIP6 models suitable for the energy service was therefore selected based on the following criteria:

• Each model provides the required climate variables at adequate temporal (sub-daily) and spatial resolution (at least 100 km), and includes far-future climate projections.
• The model subset spans the full range of Equilibrium Climate Sensitivity reported in the IPCC AR6, including low-likelihood/high-impact cases.
• Selected models are independent to avoid redundant climate behaviour.

The final selection ensures a robust representation of future climate conditions and inter-model variability, while remaining compatible with the available computational resources, which, although substantial, are inherently limited. If additional resources become available in the future, further models could be incorporated to enhance the representation of climate variability.

To capture a broad range of possible future evolutions, four Shared Socio-economic Pathways (SSPs) were selected, covering low to high greenhouse-gas emissions: SSP1-2.6 (low), SSP2-4.5 (intermediate), SSP3-7.0 (medium-high), and SSP5-8.5 (high).

The final selection of six climate models and four scenarios is summarised in Table 2.1.

Table 2.1: List of global climate models used to generate the climate indicators, showing the contributing institutions, spatial and temporal resolutions, available future scenarios, and key model configurations.

[1] Lovato, T. and Peano, D. (2020). CMCC CMCC-CM2-SR5 model output prepared for CMIP6 ScenarioMIP. Earth System Grid Federation. https://doi.org/10.22033/ESGF/CMIP6.1365  http://esgf-node.llnl.gov/search/cmip6/?mip_era=CMIP6&activity_id=ScenarioMIP&institution_id=CMCC&source_id=CMCC-CM2-SR5

[2] EC-Earth Consortium (EC-Earth) (2023); Döscher, R., et al. (2022). IPCC DDC: EC-Earth-Consortium EC-Earth3 model output prepared for CMIP6 CMIP. WDCC at DKRZ. https://doi.org/10.26050/WDCC/AR6.C6CMEEE3

[3] Schupfner, M. et al., (2023). DKRZ MPI-ESM1.2-HR model output prepared for CMIP6 ScenarioMIP. World Data Center for Climate (WDCC) at DKRZ.

https://hdl.handle.net/21.14106/681c6d41caf44e1fad86ee06f71f579bdb207d7d

[4] Xiaoge, X. et al. (2023); Tongwen, W et al. (2019). BCC BCC-CSM2MR model output prepared for CMIP6 ScenarioMIP. WDCC at DKRZ. https://www.wdc-climate.de/ui/entry?acronym=C6_4119519

[5] Semmler, et al. (2023). AWI AWI-CM1.1MR model output prepared for CMIP6 ScenarioMIP. WDCC at DKRZ. https://hdl.handle.net/21.14106/e0534db9dd20c02b6e9b9d541f978648d97ef27a

[6] Yukimoto, et al. (2023). MRI MRI-ESM2.0 model output prepared for CMIP6 ScenarioMIP. WDCC at DKRZ. https://www.wdc-climate.de/ui/entry?acronym=C6_5254341 

Data Retrieval and Pre-processing Steps

Climate projection data are retrieved from the ESGF nodes at their native nominal resolution of approximately 100 km over the global domain. The data are provided on a rotated, irregular latitude–longitude grid, in which grid cells at high latitudes cover different latitude–longitude ranges than those near the Equator.

In particular, a rotated latitude–longitude system (or rotated pole system) is a coordinate system commonly used in climate models, where the Earth’s standard coordinate grid is tilted so that the model grid aligns more naturally with the region of interest. In practice:

• the “north pole” of the rotated grid is shifted to a different geographic location;

• grid points are defined in terms of rotated latitude (rlat) and rotated longitude (rlon);

• the resulting grid is more regular over the study region, reducing distortions during interpolation and numerical calculations.

Rotated coordinate systems are widely used in climate modelling because they offer:

• Improved regional representation: standard latitude–longitude grids have variable grid spacing, while rotated grids provide a more uniform resolution over the target region;

• Better numerical performance: grid alignment can improve numerical stability in model simulations;

• Simpler interpolation: data are easier to interpolate to a regular grid when the model grid is approximately rectangular.

The historical database uses a regular latitude–longitude grid with the following characteristics:

• Domain: global

• Spatial resolution: 0.25°

• Grid size: 721 latitude points × 1440 longitude points

• Bounds: latitude [90°, −90°], longitude [0°, 360°]

To ensure consistency, all climate projection data are interpolated from their native grids to this target grid using bilinear interpolation, implemented with the CDO operator remapbil.

A spatial mask is then applied to all variables to restrict the domain to the area defined by the Global Wind Atlas ² version 3 (GWA3). This domain includes all land areas and extends up to 300 km offshore, while excluding polar regions. The mask is further extended to include the grid points required to cover the ENTSO-E Pan-European Climate Database (PECD) ³ . This extension explains the rectangular shape visible over Europe in the final domain mask. The resulting spatial coverage is shown in Figure 2.1.

Climate variables are then adjusted in time to ensure consistent temporal resolution. Air temperature (TA), wind components (U and V), and surface solar radiation (GHI) are interpolated from three-hourly to hourly resolution. A cubic spline interpolation with moving windows is used for TA, U, and V, while GHI is interpolated using a detrending approach based on top-of-atmosphere radiation. Total precipitation (TP) is aggregated from 3-hourly to daily resolution.

By convention, TP values at 00:00 UTC are assigned to the previous day, as each time step represents the accumulation over the preceding hours.

The cumulative GHI variable (surface solar radiation downwards) is converted to an instantaneous quantity by dividing by 3600 (the number of seconds in one hour), thereby converting units from J m⁻² to W m⁻². Wind speed at 10 m height 

ws_{10}

(WS10) is computed from the horizontal wind components 

u_{10}

and 

v_{10}

, using the following standard relation:

ws_{10} = \sqrt{u_{10}^2 + v_{10}^2} 

applied at each grid point and time step.

The full set of climate variables retrieved and processed includes:

• 2 m air temperature (TA)

• Total precipitation (TP)

• Surface solar radiation downwards (GHI)

• 10 m wind speed (WS10)

These variables provide the input for the subsequent computation of climate and energy indicators.

Figure 2.1: Spatial domain mask used for the climate and energy indicators dataset, based on the GWA3. 
The mask includes land areas and offshore regions up to 300 km, with an additional rectangular area covering the ENTSO-E PECD domain to ensure consistency with related datasets.

Bias Adjustment of the Climate Projection Data

Bias adjustment (BA)refers to the set of post-processing techniques used to reduce systematic biases in climate projection data, in this case CMIP6 projections. It is often a necessary step when climate model outputs are used as input for application models, especially when absolute values (rather than anomalies) are required, such as wind speed for wind power calculations.

In this workflow, bias adjustment is applied to climate variables rather than directly to energy indicators. This choice is motivated by the wider availability and higher temporal frequency of observational and reference datasets for climate variables, which makes the adjustment more robust.

Two bias-adjustment methodologies are applied to the CMIP6 projection datasets, specifically to the projection scenarios covering the period 2015–2100. The historical simulations of the CMIP6 models (1995–2014) are only used to derive the parameters required by the bias-adjustment methods. However, they are not themselves bias-adjusted and are therefore not part of the global climate and energy dataset. 

The two bias-adjustment methodologies are:

• Cumulative Distribution Function transform (CDFt; Michelangeli et al., 2009): a distribution-based approach that defines a transformation mapping the cumulative distribution of a CMIP6 variable to that of a reference dataset at each grid point.

• Delta adjustment (Navarro-Racines et al., 2020): a simpler approach that applies a constant correction factor based on the difference (or ratio) between the mean values of the model and the reference dataset at each grid point.

Both the CDFt method and the Delta adjustment are applied to gridded data, meaning that each grid-point is bias-adjusted individually for each climate variable. To derive the delta factors or the CDFt transformation functions, both bias-adjustment methodologies uses a calibration period that spans 1995–2014. Time series of calibration data are taken from ERA5 (which is bias-adjusted only for wind speed) and from the historical simulations of the CMIP6 models (which are not bias-adjusted).

The CDFt method is used for variables that exhibit a strong climate-change-related trend, such as 2 m air temperature (TA). To account for this trend, a 20-year time window is used to compute the CDFs, from which only the central 10-year period is retained as bias-adjusted output. The 20-year window is then shifted forward in time by five years, producing a new 20-year window that partially overlaps with the previous one. This approach is referred to as the overlapping moving time-window adjustment. For example, the 20-year window 2020-2039 was used to compute the bias-adjusted TA over the central 10-year period 2025-2034. Exceptions to this centered approach are the first (2015-2034) and last (2080-2100) 20-year windows, which are used to computed the bias-adjusted TA over the non-central periods 2015-2024 and 2085-2100 respectively.

Instead for 10 m wind speed and precipitation (WS10 and TP), the CDFt is applied using non-overlapping moving time-window adjustment, rather than overlapping windows, as no explicit trend correction is required. For WS10 and TP, computation and output time-windows match and span 20-year periods. For example, to compute bias-adjusted WS10 or TP over the first bias-adjustment window (2015-2034), the same 20-year period is used to compute the CDFs. The last bias-adjustment window is exception to this approach, covering five years (2095-2100).

The Delta adjustment is computationally much less demanding and is applied to variables that do not show a strong climate-change-related trend, such as solar radiation (GHI). Although wind speed and precipitation (WS10 and TP) also do not exhibit strong trends, they are nonetheless corrected using the CDFt method. This choice avoids the risk of generating negative, and therefore unphysical, values that could arise from mean-based delta corrections. 

Wind Profile Scaling for High Heights 

Unlike ERA5, many datasets—such as climate projections and seasonal forecasts—do not provide wind speed at a height of 100 m. This is the case for the CMIP6 climate projection models used for the computation of this dataset. In such situations, wind speed at 100 m is reconstructed using a power-law formulation, which requires a location-dependent wind shear exponent (α).

The α coefficient represents the vertical wind shear between 10 m and 100 m and it is derived using ERA5 wind speeds over the historical period. It is computed for each grid point, hour of the day, and month of the year, in order to capture both diurnal and seasonal variability. The resulting α fields are provided as a dedicated NetCDF dataset available through the Climate Data Store (CDS).

Wind speed at 100 m (WS100) for each climate model and scenario is then derived from the corresponding 10 m wind speed (WS10) using the wind power law:

v_{2} = v_{1} \left( \frac{h_2}{h_1} \right)^{\alpha}

where:

• v₁​ is the wind speed at height h₁ (10 m)

• v₂​ is the wind speed at height h₂​ (100 m)

• α is the dimensionless wind shear exponent.

Further details on the wind profile scaling methodology and on the computation of the α coefficient are provided on the page Power Law for Wind Profile Scaling.

Gridded and Aggregated Climate Indicators

Spatial and Temporal Aggregation Procedures 

Some crucial steps of the processing workflow involve aggregating the gridded climate indicators derived from CMIP6 into more accessible formats, both spatially and temporally. These processes are the spatial and temporal aggregations.

Spatial Aggregation

• Input: Gridded (NetCDF) indicators.

• Output: Aggregated indicators at ADM0 (ADMIN0, national level) and ADM1 (ADMIN1, first sub-national level) regions

  https://www.naturalearthdata.com

  .

The output is delivered in CSV format, where each column corresponds to a region and each row to a time step. Spatial aggregation is computed using a weighted average over land grid cells, taking into account land-sea masks and the fraction of each cell lying within the administrative boundaries. The temporal resolution of the output files is the same as the input.

Temporal Aggregation

• Input: Hourly gridded climate indicators in NetCDF format.

• Output: Daily, monthly, seasonal, and annual averages or totals, depending on the variable. Aggregated values are computed independently for each grid cell and stored in NetCDF format for the gridded version, and in CSV format for the spatially aggregated (ADM0/ADM1) version.

More detailed descriptions of both procedures can be found at the following pages:

• Spatial Aggregation Procedure

• Temporal Aggregation Procedure 

Overview of Climate Indicators

The dataset provides a suite of climate indicators, delivered in both gridded NetCDF format and regionally aggregated CSV format. These indicators serve as the foundation for the derivation of energy indicators and support a broad range of climate-related analyses and applications.

All indicators are consistently derived from CMIP6 data, covering the period from 2015 to 2100. Table 2.2 below summarises the full set of climate indicators included in the dataset, specifying their units, source variables, aggregation and output formats.

To illustrate the behaviour of the main climate indicators and the transition from historical reanalysis data to climate projections, Figure 2.2 shows example time series of yearly mean values for selected countries and variables.

Table 2.2: Gridded and aggregated climate indicators provided in the historical data stream.

Figure 2.2: Example of time-series of annual aggregations for selected climate indicators at country level (ADM0). Climate indicators include: 2m temperature (TA, annual mean), total daily precipitation (TP, annual sum), global horizontal irradiance (GHI, annual mean), and 10m wind speed (WS10, annual sum). Each plot shows data for a specific country: United States (US) for TA, Italy (IT) for TP, India (IN) for GHI, and South Africa (ZA) for WS10. Historical values are based on the ERA5 reanalysis (blue solid line; time period: 1995-2025). For climate projections, depending on the climate indicator, each plot displays either multiple scenarios for a single model or multiple models for a single scenario. The line colours are specified in the legend of each plot. When a single model or scenario is selected, its name is indicated in the plot title.             

 Quality control

A quality control (QC) procedure is applied to all climate indicators described in the previous section to ensure data integrity, physical consistency, and methodological robustness. The QC combines automated checks on the gridded outputs with targeted diagnostic analyses of the bias-adjustment methods.

General quality checks

All gridded files are systematically verified to ensure that they:

• are not corrupted;

• contain the expected number of valid and excluded grid points for the processed domain;

• exhibit indicator values within physically reasonable ranges.

These checks allow early detection of technical issues and unphysical values before dissemination.

Assessment of bias-adjustment performance

For indicators corrected using the CDFt method (i.e., 2 m temperature, total precipitation, and 10 m wind speed), the performance of the bias adjustment is evaluated by analyzing Cumulative Distribution Functions (CDFs) at grid points and time steps identified through quality checks as having values outside expected ranges. These quality checks have been performed over the entire domain and time period, while CDFs have been produced only for selected cases flagged by the quality checks.

At each selected grid point, four distributions are compared:

1. the source dataset (CMIP6 historical data),

2. the target dataset (ERA5 reanalysis),

3. the raw projection dataset (CMIP6 original climate projections),

4. the bias-adjusted projection dataset (CMIP6 bias-adjusted climate projections).

Some examples for 2 m temperature demonstrate how the moving-window CDF-based approach captures non-stationarity in future projections by progressively adjusting high-end extremes across overlapping time windows (Figure 2.3). Similar diagnostics are shown for total daily precipitation (Figure 2.4) and 10 m wind speed (Figure 2.5).

For Global Horizontal Irradiance (GHI), which is corrected using a delta adjustment, Probability Density Functions (PDFs) are analysed instead. These diagnostics confirm that the correction appropriately shifts the distributions in line with the ERA5 climatology (Figure 2.6).

Figure 2.3: Example of assessment of the CDFt bias-adjustment method applied to the 2 m temperature of the CMIP6 BCCS climate projection model for scenario SSP2-4.5 at a grid-point over Australia (24.75°S, 138.5°E), shown by the purple marker on the map (plot b). The plots (a, c-f) show the four Cumulative Distribution Functions of: (1) the source (historical, orange line), (2) target (ERA5, blue line), (3) raw projection (projection, green line), and (4) bias-adjusted projection data (red line). All CDFs refer to a February time step at 06:00 UTC across different bias-adjustment (BA) output windows: (a) third BA output window (2035-2044); (c, e) fourth and fifth BA output windows (2045-2054 and 2055-2064, respectively); and (d, f) zoomed-in views of the highest values from plots (c) and (e).

Figure 2.4: Example of assessment of the CDFt bias-adjustment method (CDF curves on the left plot) to the total daily precipitation of the CMIP6 AWCM climate projection model (future scenario: SSP3-7.0) at a grid-point located over the Amazon Basin region (8.25°S, 56°W; purple marker on the map on the right) for time-steps occurring in October over the third bias-adjustment output window (2055-2074).

Figure 2.5: Example of assessment of the CDFt bias-adjustment method to the 10m wind speed of the CMIP6 MEHR climate projection model (future scenario: SSP5-8.5) at a grid-point located over Patagonia (49.75°S, 73°W; purple marker on the map on the right) for time-steps occurring in April at 18:00 UTC over the fourth bias-adjustment output window (2075-2094).

Figure 2.6: Example of assessment of the Delta adjustment method to bias adjust Global Horizontal Irradiance (GHI). The left panel shows the Probability Density Functions (PDFs) for the CMIP6 CMR5 climate projection model (scenario SSP1-2.6): (1) raw original projection (green line and histogram), and (2) bias-adjusted projection (red line and histogram). The blue and orange vertical solid lines represent the climatology of ERA5 and CMIP6 historical data, respectively, over the period 1995–2014. All data refer to a grid-point located over India (25.5°N, 80.75°E; orange marker on the map on the right) for time-steps occurring in May at 12:00 UTC over the last (fourth) bias-adjustment output window (2075-2100).

Extreme precipitation over sensitive regions

Quality control analyses reveal limitations in the bias adjustment of total daily precipitation over regions where observations are sparse and precipitation is rare or highly irregular, such as arid and semi-arid areas and small islands. In these regions, both observational datasets and reanalyses used as bias-adjustment targets are affected by large uncertainties (Figure SPM.3, IPCC, 2021), which propagate into the corrected climate projections.

An inspection of bias-adjusted gridded data highlights grid points where total daily precipitation exceeds 2 m day⁻¹. This threshold is based on the highest observed daily precipitation on record (1.825 m; Holland, 1993). Spatial diagnostics show that these extreme values occur primarily over tropical and subtropical regions, remote islands, and desert areas, although isolated occurrences are also found elsewhere. Model-specific analyses indicate that the spatial distribution of these extremes varies across climate models. An example of comparison between the models AWCM, BCCS and ECE3 for the same climate scenario (SSP 5-8.5) is reported in Figure 2.7.

Detailed CDF analyses over sensitive regions illustrate the origin of these anomalies. When the target reanalysis climatology is wetter than the historical climate model simulation, the application of the CDF-based bias adjustment can overcorrect already wet projection data, producing unrealistically large precipitation totals. This behaviour reflects a known limitation of distribution-based bias-adjustment methods when wet-day frequencies differ substantially between source and target datasets.

Several mitigation strategies were tested, including clipping bias-adjusted precipitation in dry regions using thresholds derived from the target climatology. While some local improvements were achieved, residual inconsistencies remained across grid points and projection periods. As these approaches were not considered sufficiently robust, no additional correction beyond removing the most extreme outliers (> 2 m day⁻¹) is applied. Users are therefore advised to interpret precipitation projections with caution in sensitive regions and to consult uncorrected climate projection data when investigating local extremes.

Figure 2.7: Grid-points where at least one time-step exhibits a total precipitation value exceeding 2 m (orange markers). The size of the marker depends on the number of occurrences (i.e. time steps) of this exceedance at that location considering the CMIP6 AWCM, BCCS and ECE3 climate projection models and the SSP5-8.5 scenario over the whole projection period, 2015-2100.

High-end values of wind speed and irradiance

For 10 m wind speed, QC identified a small number of grid points, mainly at high latitudes, and time steps where values exceed 100 m s⁻¹. This threshold is based on the highest observed wind speed on record (113 m s^-1; Courtney et al., 2012). These high wind speed values result from the bias-adjustment process rather than technical errors and are within the range of historically observed extremes. Consequently, they are retained (Figure 2.8).

For GHI, following delta adjustment, some locations show values exceeding 2500 W m⁻². This intentionally high threshold is used to test the robustness of the correction over a wide range of conditions. Since these irradiance values stem from large but valid correction factors, they are also retained (Figure 2.9).

Figure 2.8: Example of assessment of the CDFt bias-adjustment method at a location where the 10m wind speed exceeds 100 m s^-1, which is the upper limit of the reasonable values for wind speed (black solid vertical line on the left plot). All curves refer to the CMIP6 MRM2 climate projection model (climate scenario: SSP2-4.5) at a grid-point located over eastern Russia (53.75°N, 134.75°E) for time-steps occurring in January at 16:00 UTC over the second bias-adjustment window (2035-2054).

Figure 2.9: Example of assessment of the Delta adjustment method at a location where Global Horizontal Irradiance (GHI) exceeds 2500 W m⁻², the upper limit of reasonable values for GHI (black dashed vertical line, left panel). The left panel shows the Probability Density Functions (PDFs) for the CMIP6 CMR5 climate projection model (scenario SSP1-2.6): (1) raw original projection (green line and histogram), and (2) bias-adjusted projection (red line and histogram). The blue and orange vertical solid lines represent the climatology of ERA5 and CMIP6 historical data, respectively, over the period 1995–2014. All data refer to a grid-point located over Peru (13.75°S, 71.25°W; orange marker on the map, right panel) for time steps occurring in November at 17:00 UTC during the first bias-adjustment window (2015–2034).

Consistency between historical and projection wind speed at 100 m

An additional QC step evaluates the consistency between historical and projection data for 100 m wind speed by comparing country-level averages over the overlapping period 2015–2024. As described in the Global climate and energy indicators from 1950 to present derived from reanalysis: Product User Guide (PUG) (Section 2.4, Wind Profile Scaling for High Heights), the historical stream provides two alternative version of the ERA5 100 m wind speed: 1) the bias-adjusted WS100, which is obtained by bias-adjusting the native ERA5 100 m wind speed (WS100) using the Global Wind Atlas version 2 (GWA2), and 2) the Alpha-based WS100, which is computed using the power law, based on the bias-adjusted historical ERA5 10 m wind speed (WS10). The Alpha-based WS100 has been produced to improve consistency between historical and projection data, particularly in climate projections for which the native 100 m wind speed is not available. To assess the offset in the transition from the historical to the projection stream, both versions of the ERA5 WS100 (bias-adjusted and Alpha-based) were compared to the climate projections.

Using aggregated files at the country-level (ADM0), the offset was evaluated by comparing ERA5 and CMIP6-projection averages over the overlapping period 2015–2024. For each CMIP6 model, the four SSP averages were first used to define a model-specific envelope and evaluate whether the ERA5 value falls inside or outside this range. When ERA5 fall outside the envelope, the offset was computed as the difference between the ERA5 average and the CMIP6 multi-scenario average over the same period. This offset was then compared to a predefined threshold of 0.2 m s^-1, chosen to exceed typical model noise while remaining physically meaningful. Results show that offsets are generally small and that the alternative formulation improves alignment with projections, reducing both the number of affected countries and the magnitude of the largest discrepancies (Table 2.2).    

Table 2.2: Evaluation of the offset between ERA5 (historical stream) and CMIP6 (projection stream) data for 100 m wind speed (WS100) over the period 2015–2024. For each CMIP6 model, the offset has been assessed at the country-level aggregation (ADM0) for the two versions of ERA5 WS100 (bias-adjusted and Alpha-based). The table reports: the number and percentage of countries showing an offset larger than the predefined threshold (0.2 m s^-1), together with the country code (ISO 3166-1 alpha2) and value (in m/s) of the largest absolute offset.

Computation of the Energy Indicators

The energy indicators provided in this dataset are derived from the climate indicators described in the previous sections, using dedicated models that convert meteorological variables into estimates of electricity generation or demand. These models are designed to reflect the physical and operational characteristics of different energy technologies. The energy indicators include:

• Wind power energy indicators

• Solar photovoltaic energy indicator

• Hydropower energy indicators

• Electricity and energy demand indicators

In addition to climate inputs, the models require reference energy data—such as installed capacity, historical electricity generation, and system configurations—for calibration and validation. These reference datasets ensure that the resulting indicators are regionally representative and consistent with observed energy system behaviour.

The following sections summarise the input and output data associated with each energy conversion model implemented in the projection stream, along with the reference energy datasets used in their development. For full details on the methodologies and assumptions behind each model, please refer to the dedicated page: Energy Conversion Models.

Reference Energy Datasets

To accurately simulate electricity generation and demand from climate indicators, the energy conversion models rely not only on meteorological inputs but also on high-quality reference energy datasets. These datasets include observed or reported information on:

• Installed capacity (e.g., wind, solar, hydroelectric)

• Actual electricity generation by technology and country

• Load profiles and total electricity demand

• Energy system characteristics (e.g., wind turbine characteristics, hydropower reservoir capacities, seasonal patterns)

The reference energy data serve two primary purposes:

1. Calibration – ensuring that model outputs reflect realistic levels of energy production and demand.

2. Validation – comparing simulated indicators against historical records to assess model performance and regional representativeness.

These datasets are collected from a mix of publicly available sources, commercial providers, international statistics, national energy agencies, and, in some cases, directly from Transmission System Operators (TSOs). When necessary, data gaps are addressed using proxy information or scaling techniques.

A full description of the datasets and sources is available on the dedicated page: Reference Energy Datasets.

Energy Indicators Modelling

The conversion of climate indicators into energy indicators is performed using a set of energy conversion models, each designed to represent the physical and operational characteristics of a specific energy system. These models take climate variables—such as wind speed, solar radiation, precipitation, and temperature—as input and output energy-related variables like electricity generation or demand.

This section provides a summary of the input and output data used in each model as implemented in the projections stream. Full methodological details, including calibration procedures and underlying assumptions, are described in the dedicated external page: Energy Conversion Models.

Wind Power Energy Indicator

Input data:

• Wind speed at 10 and 100  m (as described in Table 2.1) 

• Wind turbine characteristics (e.g. power curve, hub height, rotor diameter)

• Wind farm location and installed capacity (from The Wind Power, Global Energy Monitor, and other sources)

Model overview:
The wind power model estimates capacity factor by applying turbine-specific power curves to wind speeds extrapolated to the turbine hub height, using either 10 m or 100 m wind speed as the reference, depending on the turbine configuration.

Output data:

• Gridded wind power capacity factor at hourly resolution (NetCDF, on the entire domain described in Figure 2.1)

• Spatially aggregated indicators at ADM0 and ADM1 levels (CSV format) with hourly, daily, monthly, seasonal, and annual resolution

• Both gridded and aggregated data are provided at different hub heights and turbine types.

See Wind Power Conversion Model for more details.

Figure 3.1 below shows examples of time series of annual means for on-shore wind power (WON).

The availability of two versions of ERA5 100 m wind speed (bias-adjusted and Alpha-based WS100) results in two corresponding versions of ERA wind power indicators. In Section 2.6.5, the offset between historical and projection WS100 has been assessed. Using the same methodology, the offset has been evaluated for the onshore wind power indicator at 135 m hub height, which requires the largest wind-speed extrapolation using the power law. Table 2.3 summarizes the offset evaluation for WON at 135 m across the six CMIP6 climate projections.

Table 2.3: Evaluation of the offset between ERA5 (historical stream) and CMIP6 (projection stream) data for WON at 135 m wind speed (WS100) over the period 2015–2024. For each CMIP6 model, the offset has been assessed at the country-level aggregation (ADM0) for the two versions of ERA5 WON at 135 m produced with either the bias-adjusted or the Alpha-based WS100. The table reports: the number and percentage of countries showing an offset larger than the predefined threshold (0.01 Mw/Mw), together with the country code (ISO 3166-1 alpha2) and value (in Mw/Mw) of the largest absolute offset.

Figure 3.1: Example of time series of annual means for on-shore wind power (WON) at three different turbine heights (84 m, 100 m, and 135 m), aggregated at country level (ADM0) over South Africa (ZA). Historical values cover the time period 1995-2025 and provide two alternative version of ERA5 reanalysis for turbine heights at 100 m and 135 m: using bias-adjusted WS100 (ERA5-BA, blue solid line) or Alpha-based WS100 (ERA5-Alpha, black solid line) to extrapolate wind speed at the turbine height. For turbine heights at 84 m, wind speed at 10 m (WS10) is internally extrapolated to the hub height using the power law. Climate projection values are based on the CMIP6 MEHR climate projection model, showing four future scenarios (period: 2015-2100). Line colours are specified in each plot legend. Note that the y-axis range differs between the three plots.

Solar Photovoltaic Energy Indicator

Input data:

• Surface solar radiation downwards (GHI) and 2 m temperature (TA) (as described in Table 2.1)

• Tilt and azimuth configuration (derived from private PV plant metadata and global rules)

Model overview:
This model estimates photovoltaic electricity production using a physics-based conversion model that considers solar irradiance and air temperature to estimate capacity factors which consider the installed DC capacity as the normalisation factor. It incorporates representative module tilt and azimuth for utility-scale fixed installations (tilt and azimuth), and system losses, and is applied spatially across the entire domain.

Output data:

• Gridded solar photovoltaic capacity factor at hourly resolution (NetCDF, on the entire domain described in Figure 2.1)

• Spatially aggregated indicators (CSV format at ADM0 and ADM1 levels) at hourly, daily, monthly, seasonal, and annual temporal resolutions.

See Solar Photovoltaic Energy Conversion Model for more details.

Figure 3.2 below shows an example of time series of annual means for solar PV capacity factor (SPV).

Figure 3.2: Example of time series of annual means for solar PV capacity factor (SPV), aggregated at country level (ADM0) over Spain (ES). Historical values are based on the ERA5 reanalysis (blue solid line; time period: 1995-2025), while climate projections are based on the CMIP6 CMR5 climate projection model, showing the four future scenarios (period: 2015-2100). Line colours are specified in each plot legend.

Hydropower Energy Indicators

Random Forest (RF) Regression Model (European Domain)

Input data:

• Total precipitation and 2 m temperature, aggregated to weekly resolution and at ADM0 level (as described in Table 2.1)

• Hydropower generation, installed capacity, and reservoir filling rates from the ENTSO-E Transparency Platform

Model overview:

This statistical machine learning model estimates hydropower generation in Europe using a Random Forest (RF) regression approach trained on observed high-resolution generation data. The model is developed separately for each country and is used to compute the following indicators:

• Generation from reservoirs (HRG)
• Generation from run-of-river and pondage (HRO)
• Inflows to reservoirs (HRI)

The model uses lagged weekly climate variables (precipitation and temperature) to account for delayed hydrological responses and is validated using historical records through a Leave-One-Year-Out cross-validation strategy. Only countries with at least two years of reliable data are included. Model performance is evaluated using metrics such as the Nash-Sutcliffe Efficiency (NSE).

Output data:

Time series of aggregated hydropower indicators (HRG, HRO, and HRI) as CSV files at ADM0 level and at weekly, monthly, seasonal and annual temporal resolution.

See Hydropower Conversion Models, Section 1 "Random Forest Regression Model (European Domain)" for more details.

Figure 3.3 below shows examples of time series of annual sums for the three hydropower indicators.

Figure 3.3: Example of time series of annual sums for hydro power (HP) for three indicators (generation from reservoirs, HRG; inflows to reservoirs, HRI; and generation from run-of-river and pondage, HRO), aggregated at country level (ADM0) over France (FR). Historical values are based on the ERA5 reanalysis (blue solid line; time period: 1995-2025), while climate projections are based on the CMIP6 AWCM climate projection model, showing the four future scenarios (period: 2016-2100). For HP, the time series starts in 2016 since the Random Forest model does not have data of temperature and precipitation for the previous six months to compute HP. Line colours are specified in each plot legend. 

Installed Capacity Weighted Precipitation (IWP) Proxy (Global Domain)

Input data:

• Total precipitation (as described in Table 2.1) aggregated at monthly resolution and at sub-national level (NUTS-2 for European countries and ADM1 for the rest of the globe)

• Hydropower plant locations and installed capacities from the Global Energy Monitor (see Figure 3.4).

Model overview:

A proxy approach that computes hydropower potential as a weighted sum of n-month cumulative precipitation over regions hosting hydropower plants. Weights are based on installed capacity.

Output data:

Aggregated IWP time series as CSV files at ADM0 level at monthly, seasonal and annual temporal resolution.

See Hydropower Conversion Models, Section 2 "Installed Capacity Weighted Precipitation (IWP) Proxy (Global Domain)" for more details.

Figure 3.4: Global Energy Monitor HP plants operating (blue) and in construction (red). The bigger the size of each plant (dot), the bigger the Installed Capacity.

Electricity and Energy Demand Indicators

Electricity Demand Model (European Domain)

Input data:

• Climate variables aggregated at ADM0 level and at daily temporal resolution:

  • 2 m temperature (TA)

  • Surface solar radiation downwards (GHI)

  • Wind speed at 10 m (WS10)

• Electricity load data from ENTSO-E:

  • ENTSO-E Power Statistics: historical data from 2006 to 2014

  • ENTSO-E Transparency Platform: operational data from 2015 onward

Model overview:

A Generalised Additive Model (GAM) is used to estimate daily electricity demand across 34 European countries. The model captures the nonlinear relationship between climate variables (e.g., temperature, solar radiation, wind) and electricity load, accounting for seasonal cycles and day types. Input data are preprocessed (including detrending) to isolate climate-driven variability. Country-specific models are trained and validated using cleaned historical datasets.

Output data:

Electricity demand time series (ADM0 level, CSV format) at daily, monthly, seasonal and annual temporal resolution.

See Electricity and Energy Demand Models, Section 1 "Electricity Demand Model (European Domain)", for more details.

Figure 3.5 below shows an example of time series of annual means for the EDM.

Figure 3.5: Example of time series of annual means for the Electricity Demand Model (EDM) indicator, aggregated at country level (ADM0) over Belgium (BE). Historical values are based on the ERA5 reanalysis (blue solid line; time period: 1995-2025), while climate projections are based on the CMIP6 AWCM climate projection model, showing the four future scenarios (period: 2016-2100). Line colours are specified in the plot legend.

Energy Demand Proxy (Global Domain)

Input data:

• 2 m air temperature (daily resolution)

• Gridded population data (for country-level aggregation)

Model overview:

The model estimates energy demand using a proxy metric called Energy Degree Days (EDD), calculated as the sum of Heating Degree Days (HDD) and Cooling Degree Days (CDD). These are derived from daily mean temperatures following definitions aligned with IEA standards (HDD: T<15°C, reference 18°C; CDD: T>24°C, reference 21°C). HDD and CDD are first computed at the grid level using CMIP6 data and then aggregated to country level using population-weighted averaging. The resulting EDD index serves as a proxy for energy demand and is especially valuable in data-scarce regions, though it is also applicable in data-rich countries.

Output data:

HDD, CDD and EDD time series (ADM0 level, CSV format) at monthly, seasonal and annual temporal resolution.

See Electricity and Energy Demand Models, Section 2 "Energy Demand Model (Global Domain)" for more details.

Figure 3.6 below shows examples of time series of annual means for the CDD, HDD and EDD.

Figure 3.6: Example of time series of annual sumss for Cooling Degree Days (CDD), Heating Degree Days (HDD) and Energy Degree Days (EDD), aggregated at country level (ADM0) over Australia (AU). Historical values are based on the ERA5 reanalysis (blue solid line; time period: 1995-2025), while climate projections are based on the CMIP6 AWCM climate projection model, showing the four future scenarios (period: 2015-2100). Line colours are specified in each plot legend.

Gridded and Aggregated Energy Indicators

Spatial and Temporal Aggregation Procedures

The spatial and temporal aggregation procedures applied to energy indicators follow the same principles used for climate indicators (see Section 2.5.1). Gridded data are aggregated over standard administrative regions, ADM0 (national) and ADM1 (sub-national), and resampled to multiple time scales, including daily, monthly, seasonal, and annual resolutions, depending on the nature of each indicator.

More detailed explanations are provided in the dedicated documentation pages:

• Spatial Aggregation Procedure

• Temporal Aggregation Procedure

Exclusion Areas

Some energy indicators, particularly those related to renewable generation, such as solar and wind, may be affected by geographic constraints that limit the suitability of certain locations for energy production. To address this, the Service provides a set of exclusion area layers that define geographic zones where energy production is restricted or unlikely, such as protected areas, urban zones, steep slopes, and polar regions.

While these exclusion layers were not used in the standard modelling of energy indicators in this dataset, they are made available to users to enhance post-processing, filtering, or custom analyses.

For a full description of available exclusion criteria, NetCDF files, and variable names, see the dedicated documentation page: Exclusion Areas Computation.

Overview of Energy Indicators

The energy indicators included in this dataset are derived from the Climate Indicators and the Energy Data previously described, and through the application of specific Energy Conversion Models. Table 3.1 summarises the complete list of energy indicators, detailing the associated units, temporal and spatial resolution, domain of application, and original data sources.

Table 3.1: Gridded and aggregated energy indicators in the projections data stream.

Known issues

Extreme Daily Precipitation Values in Bias-Adjusted Projections

Status: Closed

Impacted Variables: Total daily precipitation (TP)
Spatial resolution: Gridded and aggregated products
Temporal period: Future projections (2015–2100)

Description:
In some regions, the bias-adjusted precipitation from climate change projection models can have unrealistically high daily values. These extreme values occur primarily over regions where precipitation is rare or highly irregular, such as arid and semi-arid areas, small islands, and other observation-poor regions.

This issue arises from limitations in the bias-adjustment procedure when applied to precipitation. In particular, when the reference dataset used for bias adjustment exhibits higher mean precipitation than the historical climate model simulation, the distribution-based adjustment can overcorrect already wet projection data. As a result, daily precipitation totals may reach unphysical values at isolated grid points and time steps.

Action taken and user guidance:

Several mitigation strategies were tested, including clipping precipitation values based on thresholds derived from the reference climatology and applying corrections only over dry regions. While these approaches reduced extreme values in some cases, they did not provide a robust and consistent improvement across all regions, climate models, and projection periods.

To avoid introducing additional artefacts, the only post-processing correction applied to gridded bias-adjusted TP data was to impose an upper bound of 2 m day⁻¹. While effective at removing very extreme unphysical values (> 2 m day⁻¹), this method does not modify the overall TP distribution. Users are therefore advised to interpret precipitation projections with caution in sensitive regions and to carefully assess the suitability of the data for analyses involving local precipitation extremes. Where necessary, users may also refer to the original (non–bias-adjusted) climate projection data.

Appendix

Filenames Convention and Characteristics

This paragraph aims to explain the filename convention used for the dataset described in this Product User Guide. Table 5.1 details the structure and possible fields of the filenames. Specifically, the last column indicates the corresponding section of the CDS catalogue where users can personalise their choice. If "Not applicable" is indicated, it means that the user cannot modify this field, and the data are downloaded with fixed characteristics that are not customizable. Table 5.2 details the structure and filenames of the ancillary NetCDF files (described in Table 5.3) that are available in the CDS under the widget "Weights and masks".

Table 5.1: Filename convention used in the dataset described in this Product User Guide.

Examples of filenames: 

• P_CMI6_ECEC_ECE3_TP-_0000m_Glob_025d_S205001010000_E205001310000_ACC_MAP_01d_NA-_cdf_org_NA_SP245_NA—_NA—_v1.00.nc
  This file contains projection data (P) from CMIP6 models (CMI6). The climate producing centre is EC-Earth Consortium (ECEC), and the climate model used is EC-Earth3 r1i1p1f1 (ECE3). The variable is total precipitation (TP-) at 0 m height (0000m). The coverage is the global domain (Glob) with a 0.25° spatial resolution (025d). Data span from 01/01/2050 at 00:00 UTC (S205001010000) to 31/01/2050 at 23:00 UTC (E205001312300). The data are accumulated (ACC), provided as gridded data (MAP), with an hourly temporal resolution (01h). The lead time is not available (NA-). Bias adjustment has been performed using the cumulative distribution function method (cdf), and the data are provided at their original hourly resolution without additional temporal aggregation (org). The ensemble number is not available (NA). The emission scenario is SSP2-4.5 (SP245). The technology specification and energy conversion model are not applicable (NA---_NA---). The file version is (v1.00) and the file format is NetCDF (.nc).
• P_CMI6_BCC-_BCCS_WS-_0010m_Glob_ADM0_S210012010000_E210012312100_INS_TIM_01m_NA-_cdf_avg_NA_SP245_NA---_NA—_v1.00.csv
  This file contains projection data (P) from CMIP6 models (CMI6). The climate producing centre is Beijing Climate Centre (BCC-), and the climate model used is BCC-CSM2-MR r1i1p1f1 (BCCS). The variable is wind speed (WS-) at 10 m height (0010m). The coverage is the global domain (Glob), aggregated at the ADM0 (country) level. Data span from 01/12/2100 at 00:00 UTC (S210012010000) to 31/12/2100 at 21:00 UTC (E210012312100). The data are instantaneous (INS), provided as time series (TIM), with a monthly temporal resolution (01m). The lead time is not available (NA-). Bias adjustment has been applied using the cumulative distribution function method (cdf), and the values are averaged over the selected temporal resolution (avg). The ensemble number is not available (NA). The emission scenario is SSP2-4.5 (SP245). The technology specification and energy conversion model are not applicable (NA---_NA---). The file version is (v1.00) and the file format is CSV (.csv).
• P_CMI6_BCC-_BCCS_WS-_0100m_Glob_025d_S210012010000_E210012312100_INS_MAP_01h_NA-_noc_org_NA_SP245_NA---_NA—_v1.00.nc
  This file contains projection data (P) from CMIP6 models (CMI6). The climate producing centre is Beijing Climate Centre (BCC-), and the climate model used is BCC-CSM2-MR r1i1p1f1 (BCCS). The variable is wind speed (WS-) at 100 m height (0100m). The coverage is the global domain (Glob) with a 0.25° spatial resolution (025d). Data span from 01/12/2100 at 00:00 UTC (S210012010000) to 31/12/2100 at 21:00 UTC (E210012312100). The data are instantaneous (INS), provided as gridded data (MAP), with an hourly temporal resolution (01h). The lead time is not available (NA-). No bias adjustment has been applied (noc) since the data are obtained through wind power law extrapolation from ERA5 bias-adjusted 10m wind speed, and the data are provided at their original hourly resolution without additional temporal aggregation (org). The ensemble number is not available (NA). The emission scenario is SSP2-4.5 (SP245). The technology specification and energy conversion model are not applicable (NA---_NA---). The file version is (v1.00) and the file format is NetCDF (.nc).

• P_CMI6_AWI-_AWCM_WON_0135m_Glob_025d_S210012010000_E210012312100_CFR_MAP_01h_NA-_noc_org_NA_SP126_WP014_PhM02_v1.00.nc

  This file contains projection data (P) from CMIP6 models (CMI6). The climate producing centre is Alfred Wegener Institute (AWI-), and the climate model used is AWI-CM-1-1-MR r1i1p1f1 (AWCM). The variable is onshore wind power capacity factor (WON) at 135 m hub height (0135m). The coverage is the global domain (Glob) with a 0.25° spatial resolution (025d). Data span from 01/12/2100 at 00:00 UTC (S210012010000) to 31/12/2100 at 21:00 UTC (E210012312100). The data are expressed as capacity factor (CFR), provided as gridded data (MAP), with an hourly temporal resolution (01h). The lead time is not available (NA-). No bias correction has been applied (noc), and the data are provided at their original hourly resolution without additional temporal aggregation (org). The ensemble number is not available (NA). The emission scenario is SSP1-2.6 (SP126). The technology specification is (WP014) (NREL Bespoke_6MW_170, onshore, hub height 135 m, installed capacity 6 MW). The energy conversion model used is Physical Model method 2 (PhM02). The file version is (v1.00) and the file format is NetCDF (.nc).

• P_CMI6_AWI-_AWCM_WON_0084m_Glob_025d_S210012010000_E210012312100_CFR_MAP_01h_NA-_noc_org_NA_SP126_WP002_PhM02_v1.00.nc
  This file contains projection data (P) from CMIP6 models (CMI6). The climate producing centre is Alfred Wegener Institute (AWI-), and the climate model used is AWI-CM-1-1-MR r1i1p1f1 (AWCM). The variable is onshore wind power capacity factor (WON) at 84 m hub height (0084m). The coverage is the global domain (Glob) with a 0.25° spatial resolution (025d). Data span from 01/12/2100 at 00:00 UTC (S210012010000) to 31/12/2100 at 21:00 UTC (E210012312100). The data are expressed as capacity factor (CFR), provided as gridded data (MAP), with an hourly temporal resolution (01h). The lead time is not available (NA-). No bias correction has been applied (noc), and the data are provided at their original hourly resolution without additional temporal aggregation (org). The ensemble number is not available (NA). The emission scenario is SSP1-2.6 (SP126). The technology specification is (WP002) (Gamesa G132/3300, onshore, hub height 84 m, installed capacity 3.3 MW). The energy conversion model used is Physical Model method 2 (PhM02). The file version is (v1.00) and the file format is NetCDF (.nc).

Table 5.2: Filename convention for ancillary data used in the historical data stream and that are available in the CDS under the widget "Weights and masks".

Example of filename for the ancillary data: ANCI_LATW-coeff_C3S2LOT1_025d_v1.00.nc.

This file contains ancillary data (ANCI) used to adjust the gridded data with the proper latitudinal weights (LATW-coeff) during the spatial aggregation procedure. The file has been created within the framework of the C3S2 Lot1 contract (C3S2LOT1). The spatial resolution is 0.25° (025d); the file version is v1.00, and the file format is NetCDF (.nc).

Table 5.3: Description of the ancillary data and their characteristics. These files are available for download in the CDS under the widget "Weights and masks".

Metadata

The header of the time series CSV files contains the following metadata descriptors. An example of a Cooling Degree Days variable is presented below.

# General
## Title
### Cooling degree days - C3S Energy Lot1 file version v1.00
## Abstract
### CMIP6; t2m bias adjustment: cdf; energy conversion model: PhM03
## Date
### 2025-02-20
## Date type
### Publication: Date identifies when the data was issued
## Unit
### °C
## URL
### https://cds.climate.copernicus.eu/
## Data format
### CSV
## Keywords
### CMI6 : Copernicus : C3S : C3S Energy : ICS
## Producer
### Inside Climate Service
## Point of contact
### ECMWF User Support, https://confluence.ecmwf.int/site/support
# Usage
## Access constraints
### Intellectual property rights: The IP of these data belongs to the EU Copernicus programme
## Use constraints
### CC-BY 4.0, https://creativecommons.org/licenses/by/4.0/
## Temporal extent
## Begin date
### 2015-01-01 00:00
## End date
### 2015-01-31 23:00
## Temporal resolution
### 01m
## Geographic bounding box
### westBoundLongitude 0.5
### eastBoundLongitude 359.5
### southBoundLatitude -89.5
### northBoundLatitude 89.5
## Spatial resolution
### ADMIN0
# Lineage Statement
## Original Data Source
## Statement
### The original data sources are CMIP6_BCC-CSM2-MR_ssp126_r1i1p1f1_Projection (available at: https://esgf-data.dkrz.de/esg-search/)

References

Beyer, H G, Heilscher, G, Bofinger, S, “A robust model for the MPP performance of different types of PV-modules applied for the performance check of grid connected systems”, EuroSun 2004 conference, pp. 3064-3071, Germany, June 2004.

Blanc, P., Wald, L., “The SG2 algorithm for a fast and accurate computation of the position of the Sun for multi-decadal time period”, Solar Energy, vol. 86, pp. 3072-3083, 2012.
(https://doi.org/10.1016/j.solener.2012.07.018)

Courtney, J., Steve Buchan, Randall S. Cerveny, Pierre Bessemoulin, Thomas C. Peterson, Jose M. Rubiera Torres, John Beven, John King, Blair Trewin, and Kenneth Rancourt, 2012: Documentation and Verification of the World Extreme Wind Gust Record:  113.3 m s-1 on Barrow Island Australia during passage of Tropical Cyclone Olivia, Australian Meteorological and Oceanographic Journal, 62(1): 1-9.

Eyring, V., Bony, S., Meehl, G. A., Senior, C. A., Stevens, B., Stouffer, R. J., and Taylor, K. E.: Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization, Geosci. Model Dev., 9, 1937–1958, https://doi.org/10.5194/gmd-9-1937-2016, 2016.

Gaillard, P.,; Goude, Y.,; Stoltz, G. A further look at the forecasting of the electricity consumption by aggregation of specialized experts. Technical report, École normale supérieure, Paris and EDF R&D, Clamart; 2011.

Goude, Y., Nedellec, R. & Kong, N. (2014) Local short and middle term electricity load forecasting with semi-parametric additive models. IEEE Trans. Smart Grid.

Gueymard, C. A., Lara-Fanego, V., Sengupta, M., & Xie, Y., “Surface albedo and reflectance: Review of definitions, angular and spectral effects, and intercomparison of major data sources in support of advanced solar irradiance modeling over the Americas”, Solar Energy, vol. 182, pp. 194–212, 2019. (https://doi.org/10.1016/j.solener.2019.02.040)

Ho, L.T.T.; Dubus, L.; De Felice, M.; Troccoli, 2020. A. Reconstruction of Multidecadal Country-Aggregated Hydro Power Generation in Europe Based on a Random Forest Model. Energies 2020, 13, 1786. https://doi.org/10.3390/en13071786

Holland, G.J., 1993: WMO/TC-No. 560; Report No. TCP-31; World Meteorological Organization; Geneva; Switzerland. 

IEA (2023), Weather for Energy Tracker, IEA, Paris https://www.iea.org/data-and-statistics/data-tools/weather-for-energy-tracker

IPCC (2021). Summary for Policymakers. In: Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [Masson-Delmotte, V., P. Zhai, A. Pirani, et al. (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 3−32, doi:10.1017/9781009157896.001 

Jourdier, B.: Evaluation of ERA5, MERRA-2, COSMO-REA6, NEWA and AROME to simulate wind power production over France, Adv. Sci. Res., 17, 63–77, https://doi.org/10.5194/asr-17-63-2020, 2020.

Klucher, T. M., “Evaluation of models to predict insolation on tilted surfaces”, Solar Energy, vol. 23, pp. 111–114, 1979. (https://doi.org/10.1016/0038-092X(79)90110-5)

Koivisto M., K. Plakas, E. R. Hurtado Ellmann, N. Davis, P. Sørensen, “Application of microscale wind and detailed wind power plant data in large-scale wind generation simulations”, Electric Power Systems Research, vol. 190, 106638, January 2021(https://doi.org/10.1016/j.epsr.2020.106638)

Lledó, Ll.; Torralba, V.; Soret, A.; Ramon, J.; Doblas-Reyes, F.J. "Seasonal forecasts of wind power generation," Renewable Energy, Elsevier, vol. 143(C), pages 91-100, 2019.

Liu, B., & Jordan, R., “Daily insolation on surfaces tilted towards equator”, ASHRAE Transactions, 10, pp. 526–541, 1961.

Martin, N., & Ruiz, J. M., “Calculation of the PV modules angular losses under field conditions by means of an analytical model”, Solar Energy Materials and Solar Cells, vol. 70, pp. 25–38, 2001. (https://doi.org/10.1016/S0927-0248(00)00408-6)

Martin, N., & Ruiz, J. M., Corrigendum to “Calculation of the PV modules angular losses under field conditions by means of an analytical model”, Solar Energy Materials and Solar Cells, vol. 110, pp. 154, 2013.

Michelangeli, P.-A., Vrac, M., and Loukos, H. (2009). Probabilistic downscaling approaches: Application to wind cumulative distribution functions, Geophys. Res. Lett., 36, L11708, doi:10.1029/2009GL038401.

Mortensen N. G., “Wind resource assessment using the WAsP software”, WindEurope DTU report, 2018 (https://backend.orbit.dtu.dk/ws/portalfiles/portal/164389714/Wind_resource_assessment_using_the_WAsP_software_DTU_Wind_Energy_E_0174_.pdf).

Murcia Leon J. P., M. J. Koivisto, P. Sørensen, P. Magnant, “Power Fluctuations In High Installation Density Offshore Wind Fleets”, Wind Energy Science, vol. 6, pp. 461–476, 2021. (https://doi.org/10.5194/wes-6-461-2021).

Murcia J. P., M. J. Koivisto, G. Luzia, B. T. Olsen, A. N. Hahmann, P. E. Sørensen, M. Als, “Validation of European-scale simulated wind speed and wind generation time series”, Applied Energy, vol. 305, 117794, January 2022 (https://doi.org/10.1016/j.apenergy.2021.117794).

Navarro-Racines, C., Tarapues, J., Thornton, P. et al. (2020). High-resolution and bias-corrected CMIP5 projections for climate change impact assessments. Sci Data 7, 7. https://doi.org/10.1038/s41597-019-0343-8

Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V. and Vanderplas, J., 2011. Scikit-learn: Machine learning in Python. the Journal of machine Learning research, 12, pp.2825-2830.

Pierrot. A., Goude, Y. Short-term electricity load forecasting with generalized additive models. Proceedings of ISAP power. 2011 Sep 2011.

pvlib,  python v0.9.3 documentation, Klucher irradiance transposition function. Retrieved November 30, 2022, from https://pvlib-python.readthedocs.io/en/stable/reference/generated/pvlib.irradiance.klucher.html, n.d.

pvlib,  python v0.9.3 documentation, ground diffuse irradiance function. Retrieved November 30, 2022, from https://pvlib-python.readthedocs.io/en/stable/reference/generated/pvlib.irradiance.klucher.html, n.d.

PyPi sg2 package entry. Retrieved November 30, 2022, from https://pypi.org/project/sg2/, n.d.

Ross, R.G., “Interface design considerations for terrestrial solar cell modules,” in Photovoltaic Specialists Conference Record, pp. 801-806, United States of America, 1976.

Saint-Drenan, Y.-M., Wald, L., Ranchin, T., Dubus, L., and Troccoli, A.: An approach for the estimation of the aggregated photovoltaic power generated in several European countries from meteorological data, Adv. Sci. Res., 15, 51–62, https://doi.org/10.5194/asr-15-51-2018, 2018.

Schmidt, H., Sauer, D. U., ‘‘Practical modeling and estimation of inverter efficiencies,’’ 9th Internationales Sonnenforum, pp. 550–557, Germany, 1994.

Scoccimarro, E., Cattaneo, O., Gualdi, S. et al. Country-level energy demand for cooling has increased over the past two decades. Commun Earth Environ 4, 208 (2023). https://doi.org/10.1038/s43247-023-00878-3

Skartveit, A., & Olseth, J. A., “A model for the diffuse fraction of hourly global radiation”, Solar Energy, vol. 38, pp. 271-274, 1987. (https://doi.org/10.1016/0038-092X(87)90049-1)

Swisher P., J. P. Murcia Leon, J. Gea-Bermúdez, M. Koivisto, H. Madsen, M. Münster, “Competitiveness of a low specific power, low cut-out wind speed wind turbine in North and Central Europe towards 2050”, Applied Energy, vol. 306, part B, 118043, January 2022 (https://doi.org/10.1016/j.apenergy.2021.118043).

Temps, R. C., & Coulson, K. L., “Solar radiation incident upon slopes of different orientations”, Solar Energy, vol. 19, pp. 179–184 (https://doi.org/10.1016/0038-092X(77)90056-1).

Wood, N.; Goude, Y.; Shaw, S. Generalized Additive Models for Large Data Sets, Journal of the Royal Statistical Society Series C: Applied Statistics, Volume 64, Issue 1, January 2015, Pages 139–155, https://doi.org/10.1111/rssc.12068

Williams, S. R., Betts, T. R., Helf, T., Gottschalg, R., Beyer, H. G., Infield, D. G., “Modelling long-term module performance based on realistic reporting conditions with consideration to spectral effects”, 3rd World Conference on Photovoltaic Energy Conversion, vol. 2, pp. 1908-1911, 2003.

Related articles

Related articles appear here based on the labels you select. Click to edit the macro and add or change labels.