SCOPE-ERA5: Station-Calibrated Outputs for Planning & Engineering-ERA5

SCOPE-ERA5 (Station-Calibrated Outputs for Planning & Engineering-ERA5) is a global, observationally calibrated daily resolution version of ERA5 at the location of weather stations. This dataset uses a multivariate bias adjustment method (MBCn; Cannon et al., 2018) to correct key ERA5 thermodynamic variables—such as near-surface air temperature, humidity, pressure—and wind speed based on daily observations from more than 15,000 weather stations around the world. SCOPE-ERA5 was developed by Degree Day LLC, an independent provider of climate risk data and analytics (https://www.degreeday.org).

SCOPE-ERA5 provides temporally complete, thermodynamically consistent daily time series that better reflect observed local conditions compared to raw ERA5 at weather station locations. It is designed for use in applications requiring local accuracy, such as engineering design, energy systems modeling, and other climate risk assessment purposes. Additional technical details are provided in the supporting article.

This Zenodo deposit includes both the full SCOPE-ERA5 daily timeseries dataset (restricted) and a subset covering only stations in the United Kingdom (unrestricted). The full dataset is available for non-commercial use. Each station’s data is provided in an individual NetCDF file. This Zenodo deposit also includes several climatological indicators calculated from SCOPE-ERA5 daily timeseries at the full record of weather stations.

Key Features:

• Bias-adjusted ERA5 data using observed daily station records (1979–2025).
• Use of a multivariate correction that preserves inter-variable physical relationships. 
• Daily timeseries available for >15,000 stations globally.
• Variables include: temperature (tas, tasmax, tasmin), dew point (dew_point), relative humidity (hurs), specific humidity (huss), wet-bulb temperature (wet_bulb), heat index (heat_index), wind speed (sfcWind), wind chill index (wind_chill), surface pressure (ps), and more.
• Indicators computed on SCOPE-ERA5, raw ERA5, and station observations using identical methodology, so users can directly quantify the bias-correction effect.
• Format: NetCDF4 and MS Excel; one file per station.

What’s new in v2.0:

• Station count expanded from 7,115 to 15,744 via multi-source ingest (HadISD, GHCNh, GHCNd, ECAD, GSOD, USCRN, CDS), deduplicated at 1 km haversine plus station-ID cross-source matching.
• Time period extended through 2025.
• Pairwise SNHT homogeneity testing with ERA5 as a synthetic neighbour (Menne & Williams 2009; Alexandersson 1986), detecting breaks at sparse-station and high-latitude locations missed by v1’s HadISD-only PHA list.
• Joint training-window selection: a single 20-year window per station, prioritised tas > dew_point > sfcWind, replacing v1’s per-variable-independent selection.
• Break-magnitude rule: training windows whose cumulative |Σ break magnitude| exceeds 2 °C are rejected (temperature variables only).
• Dunn (2014) / HadISD v3 station QC applied per-station before homogeneity testing (spike, day-to-day-change, frequent-value, constant-difference, dtr-collapse, Pettitt step, day-of-year outlier, variance-shift, world-record bounds), with Köppen-Geiger-zone-aware thresholds.
• ERA5 surface pressure hypsometric fallback for stations with insufficient ps coverage; ps is no longer a category-gating variable.
• Substantially expanded indicator set (up to ~50 annual indicators per Category 3 station), plus monthly means of every bias-corrected variable and a per-year valid-day fraction.
• Comparison indicators added: the same indicator set is also computed on raw ERA5 and on station observations for direct evaluation of the bias-correction effect.

More Details:

The set of available variables differs by weather station, since not all stations recorded each of the foundational variables (from which other variables were derived) or met the required data completeness thresholds. To balance data quality with spatial coverage, stations were grouped into three hierarchical categories, each representing a different "package" of coincident foundational variables, depending on data completeness and station data homogenization.

Category 1 stations (N = 10,492) met completeness criteria for dry-bulb temperature alone.
Category 2 stations (N = 2,329) additionally included complete records of dew point temperature.
Category 3 stations (N = 2,923) further required complete 10-meter surface wind speed observations.

In v2, relative humidity (hurs) is derived from (tas, dew_point) rather than being a category-gating variable. Surface pressure (ps) is bias-corrected when station coverage is sufficient and otherwise falls back to elevation-adjusted ERA5 ps, so ps is no longer category-gating either.

Included Supplemental and Derived Variables:

The table below is an overview of climate variables available in the dataset by category. Daily resolution.

Note on surface pressure (Category 1): for Category 1 stations, ps is taken from elevation-adjusted ERA5 (hypsometric fallback) and is not bias-corrected from station observations.

Units convention: daily timeseries files use SI units throughout (temperatures in kelvin), so they plug directly into bias-adjustment / quantile-mapping toolchains. Indicator files store every temperature-valued indicator both in degC (<var>) and in degF (<var>_F); threshold-count indicators are unitless day counts.

The included *.csv files provide metadata for weather stations used in the dataset (subset and full dataset). Each row corresponds to a unique station, and the columns are defined as follows:

• Station_ID: Unique identifier assigned to each station in the dataset (typically a concatenation of WMO and WBAN codes if available).

• Category: Station category based on data completeness and availability (e.g., Category 1, 2, or 3), indicating which variables are available and meet quality criteria.

• Lat: Latitude of the station in decimal degrees (positive for North, negative for South).

• Lon: Longitude of the station in decimal degrees (positive for East, negative for West).

• Elevation: Elevation of the station above mean sea level in meters.

• WMO: World Meteorological Organization station ID, if available.

• WBAN: U.S. Weather Bureau Army Navy (WBAN) station code, if available.

• Station Name: Official station name, typically corresponding to the reporting airport, city, or region.

• Country: Country where the station is located.

• Province: State, province, or administrative region (if applicable).

• City: Nearest city or urban center (if applicable).

• County: County or district (if applicable).

• Continent: Continent on which the station is located (e.g., North America, Asia).

• Subregion: More specific geographic region within the continent (e.g., Southeast Asia, Western Europe).

• Train_start: First year of the 20-year period used for bias adjustment training at the station (joint window across primary variables in v2).

• Train_end: Last year of the 20-year training period.

Climatological Indicators

Several climatological indicators have been derived from the daily data from SCOPE-ERA5. These are described below and are available for download in this repository. In v2, the same indicator set is also computed on raw ERA5 and on cleaned station observations using identical methodology, so users can directly quantify the bias-correction effect.

Temperature Extremes (ETCCDI)

Energy-related Indicators

Heat-related Indicators

Cold-related Indicators

Wind-related Indicators (Category 3 only)

Additional fields: indicator files also include monthly means of every bias-corrected variable present at the station and a per-year valid-day fraction for each primary variable (mask for sparse station-years).

Version History:

• V2.0-complete (2026-04-28)
  • Major release of the full SCOPE-ERA5 daily timeseries.
  • Updates:
    • Stations expanded from 7,115 to 15,744 via multi-source ingest (HadISD, GHCNh, GHCNd, ECAD, GSOD, USCRN, CDS), with 1 km haversine deduplication and station-ID cross-source matching.
    • Time period extended through 2025.
    • Pairwise SNHT homogeneity testing with ERA5 as a synthetic neighbour, replacing v1’s HadISD-only PHA breakpoint list.
    • Joint training-window selection (single 20-year window per station across primary variables) replacing per-variable-independent selection.
    • Break-magnitude rule: training windows whose cumulative |Σ break magnitude| exceeds 2 °C are rejected.
    • Dunn (2014) / HadISD v3 station QC applied per-station before homogeneity testing, with Köppen-Geiger-zone-aware thresholds.
    • ERA5 hypsometric fallback for surface pressure; ps is no longer category-gating.
    • Category 2 now requires complete dew point observations (relative humidity is derived from tas + dew_point).
    • Daily-only-station tasskew fix: stations where tas is reconstructed from (tasmax + tasmin)/2 in ingest now publish tasskew = 0.5 throughout the record.
    • Substantially expanded indicator set (up to ~50 annual indicators per Category 3 station): ETCCDI temperature indices; HDD/CDD at three SI bases (10 °C, 15 °C, 18 °C); four-threshold tx_days_above, tn_days_above, and hi_days_above families; monthly means; per-year valid-day fractions.
    • Comparison indicators added: identical methodology applied to raw ERA5 and to cleaned station observations.
    • Indicator temperature units: every temperature-valued indicator is now stored in both °C and °F.

• V1.1-complete (2025-12-07)
  • Initial Release of the full SCOPE-ERA5 daily timeseries.

• V1.1-subset (2025-12-07)
  
  • Updates:
    • Added separate tracking of daytime and nighttime heatwaves to capture locations experiencing divergent trends (e.g., increasing nighttime heat stress without corresponding daytime changes).
    • Fixed calculation error in coldwave and freeze-thaw cycle indicators.
  • Format change: NetCDF (.nc) format discontinued for climatological indicators. All future releases will use Excel (.xlsx) format only for indicators.
  • Note: UK daily timeseries data unchanged from previous version.

• V1.0-subset (2025-06-30)
  • Initial Release.
    
    

How to cite the dataset:

Rasmussen, D. J. (2026). SCOPE-ERA5: Station-Calibrated Outputs for Planning & Engineering-ERA5 (v2.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.3735533

How to cite the Methodology: 

Rasmussen, D. J. (2026). Multivariate bias correction of ERA5 using in-situ observations for planning and engineering. Environmental
Research: Climate. https://doi.org/10.1088/2752-5295/ae63ee