Look up the average July temperature for a city, or how many rainy days Madrid usually gets in November, and the figures almost certainly trace back to a dataset called ERA5. It underpins work at national weather services, climate research groups, and a long list of public weather tools, including several apps in the SimpleMeteo family. This article covers what ERA5 is, how it is built, and why it has become the default source for historical climate data.

What Is ERA5?

ERA5 is the fifth-generation atmospheric reanalysis of the global climate produced by the European Centre for Medium-Range Weather Forecasts (ECMWF). It reaches the public through the Copernicus Climate Change Service (C3S), which ECMWF runs on behalf of the European Union.

The dataset gives hourly values for a wide range of atmospheric, land, and ocean variables, covering the whole planet from January 1940 up to the recent past, with new data added as it becomes available. It replaced ERA-Interim, an earlier reanalysis that ran from 1979 to 2019 and is no longer updated.

The methods behind ERA5 and the testing that went into it are set out in its main reference paper: Hersbach et al. (2020), "The ERA5 global reanalysis," Quarterly Journal of the Royal Meteorological Society, 146, 1999–2049, doi:10.1002/qj.3803.

What Is Reanalysis?

"Reanalysis" is what separates ERA5 from a plain archive of weather records.

Direct measurements come from surface stations, weather balloons, ships, aircraft, and, since the 1970s, satellites. They are real, but they are scattered unevenly: dense over Europe and North America, thin over the oceans, the mountains, and much of the southern hemisphere, with coverage that changes over the years. That makes raw station data awkward for climate comparisons. A coastal city with a hundred-year record is not really comparable to a mountain valley that had no instruments until 1970.

Reanalysis fills those gaps. Every available observation is fed into a numerical weather prediction model through a process called data assimilation. The model carries a physically consistent estimate of the atmosphere forward in time, and the observations pull that estimate toward reality wherever real measurements exist. What comes out is a complete, gap-free grid of past weather, with no blank spots in space or time and every value tied together by the same physical equations.

One detail matters a great deal for climate work. ERA5 runs the entire 1940-to-present record through a single, fixed version of the model. Operational forecasting systems are upgraded constantly, which leaves subtle steps in their long records. ERA5 avoids that, so a temperature trend measured across decades reflects the climate itself rather than a change in the software.

How ERA5 Is Produced

ERA5 is built with 4D-Var, a variational form of data assimilation. For each twelve-hour window it adjusts the modelled state of the atmosphere to fit all the observations gathered during that window at once, while keeping the result consistent with the physics built into the model. It is expensive to run, but it produces estimates that hold together cleanly across space and time.

The observations come from many sources:

Surface stations on land, measuring temperature, pressure, wind, and humidity
Weather balloons (radiosondes), launched twice a day to profile the atmosphere from the ground upward
Aircraft, reporting temperature and wind along their routes
Ships and moored buoys out over the oceans
Satellites, whose numbers grew sharply from 1979 onward, which is the main reason ERA5 is more reliable from 1979 to the present than in its earliest decades

The full list of inputs and the processing chain are documented on the dataset's page in the Copernicus Climate Data Store.

What ERA5 Covers

Horizontally, ERA5's atmospheric fields come on a regular 0.25° latitude–longitude grid, cells of around 28 km, derived from a model that runs at roughly 31 km. A separate product, ERA5-Land, reworks the land-surface variables onto a finer 0.1° grid of about 9 km. Vertically, the atmosphere is divided into 137 levels reaching from the ground up to around 80 km, enough to cover everything from near-surface turbulence to the stratosphere.

Everything is reported hourly. That means you can recover not just monthly and daily averages but the daily cycle itself, such as how far the temperature usually drops overnight in a given place and month. The list of variables runs long: two-metre air temperature, total precipitation, wind at several heights, humidity, sea-surface temperature, snow cover, soil moisture, and the components of incoming solar radiation, among many more.

Where ERA5 Has Limits

It helps to be clear about what ERA5 cannot do.

For one, it is a model output rather than a direct reading. Each grid value is the model's best estimate, shaped by observations rather than measured by an instrument standing on that spot. Where the observing network is dense, as in Europe, North America, or Japan, the model is held tightly to the measurements and the two agree closely. In thinly observed regions, and in the decades before satellites, the estimates are looser.

There is also the matter of resolution. A cell roughly 28 km across cannot resolve a single alpine valley, a coastal cliff, or a narrow plateau; it returns an average over the whole cell. A city sitting at 3,600 m among grid cells that average 2,400 m will read systematically too warm. Anyone using ERA5-derived figures for high or sharply enclosed places should keep that in mind.

And the earliest data is the shakiest. From 1940 to 1978 there were far fewer observations to anchor the model, especially over the oceans and the southern hemisphere, so values from that stretch carry more uncertainty than the satellite-era record that follows. ECMWF flags this in its documentation for the back-extension.

Getting Hold of ERA5

ERA5 is free. The full archive lives in the Copernicus Climate Data Store, where a free account lets you pull data through the web interface or the CDS API (the cdsapi Python client), in NetCDF or GRIB format.

For developers who want something lighter, the Open-Meteo API serves ERA5 history, alongside other reanalysis and forecast data, as plain JSON over HTTP, with no account needed for non-commercial use.

ERA5 and the SimpleMeteo Apps

Several SimpleMeteo tools sit on top of Copernicus and ECMWF data.

WeatherJourney draws on ERA5 back to 1940 to show the climate history of more than 6,000 cities: monthly temperature and rainfall averages, year-by-year trends, and how each place has warmed over the decades. Because the whole record comes from one consistent dataset, you can line up a city's climate in the 1950s against today and trust that you are comparing like with like.

UV Index Today and AirIndex are built on the Copernicus Atmosphere Monitoring Service (CAMS), ECMWF's atmospheric-composition arm, which covers ultraviolet radiation and air quality. Pollen Today runs on the CAMS European pollen forecast. All of them reach you through the same Copernicus and ECMWF infrastructure that produces ERA5.

Further Reference Documents

ERA5 on the Copernicus Climate Data Store — official dataset page, documentation, and API access
ECMWF ERA5 overview — ECMWF's own description of ERA5 and where it sits in the reanalysis lineage
Copernicus Climate Change Service: reanalysis — programme overview and background on reanalysis
Hersbach et al. (2020), "The ERA5 global reanalysis," Q.J.R. Meteorol. Soc., 146, 1999–2049 — doi:10.1002/qj.3803 — the primary peer-reviewed description of ERA5

ERA5 Explained: What Climate Reanalysis Is and How It Works