Skip to content
Units

WeatherArchive / About the data

ERA5 reanalysis explained: where our weather history comes from

Every past-weather number on WeatherArchive comes from ERA5, produced by the European Centre for Medium-Range Weather Forecasts (ECMWF) for the Copernicus Climate Change Service, and served by Open-Meteo. Here is what that means, and when to trust it.

What a reanalysis is

A reanalysis re-runs a modern weather-forecast model over the past, hour by hour, and at every step nudges it towards whatever was actually observed: weather stations, radiosonde balloons, ships, buoys, aircraft and, since the 1970s, satellites. The result is a complete, physically consistent picture of the atmosphere on a regular grid, with no gaps where there were no stations.

Step through one reanalysis cycle

Dashed: model forecast

The model starts from the previous state of the atmosphere and runs forward twelve hours. Its forecast (dashed) is a good guess, but drifts from reality.

What it is good at

Temperature and humidity over flat or gently rolling land, pressure, wind at larger scales, and long consistent series for comparing years or places. That makes it ideal for climate averages, frost dates, degree days and “how did this week compare with normal” questions.

Where it falls short

Small-scale rain. A thunderstorm a few kilometres wide is smeared across a grid cell, so daily totals can be too low at the storm and too high nearby. Mountains and coasts. A grid cell averages valley and peak, land and sea; expect differences of a few degrees. Extremes. Record station highs and lows are usually more extreme than the grid-cell value. Early decades. Before satellites (roughly pre-1979), the model had far fewer observations, especially in the southern hemisphere and over oceans.

How WeatherArchive uses it

We query the Open-Meteo Historical Weather API, which combines ERA5, ERA5-Land and, for the most recent days, ECMWF’s operational model, and delivers it in the place’s local time. Requests go straight from your browser to Open-Meteo when you press a button; we don’t store your queries. Recent results are cached in your browser so repeat views are instant. Sun times are computed locally with NOAA’s solar equations, and nothing about them depends on weather data.

We never fill gaps: if the source has no value for a variable, the page says “no data”.

Credit and licence

Weather data by Open-Meteo.com, licensed under CC BY 4.0. Contains modified Copernicus Climate Change Service information; neither the European Commission nor ECMWF is responsible for any use of it. Hersbach, H. et al. (2020), “The ERA5 global reanalysis”, Quarterly Journal of the Royal Meteorological Society 146: 1999–2049. Geocoding by Open-Meteo using GeoNames data (CC BY 4.0).

Questions people ask

What does ERA5 stand for?

It is the fifth generation of ECMWF’s atmospheric ReAnalysis of the global climate, covering 1940 to the present.

What is the resolution of ERA5?

About 31 km globally, with hourly output. ERA5-Land re-runs the land surface part at about 9 km.

Is ERA5 the same as observed weather?

No. It is a model estimate constrained by observations. It is usually close to what stations measured, but not identical, especially for rain and in complex terrain.

How soon is new data available?

The archive usually reaches to about two days ago; ERA5 itself follows with a delay of about five days, and preliminary values can be revised.