Historical weather data is freely available from several government, academic, and commercial sources, with records stretching back more than 80 years in some cases. The best source for you depends on whether you need data for a specific U.S. location, global coverage, or a developer-friendly format you can plug into an application.
NOAA Climate Data Online
For U.S. locations, the single best starting point is NOAA’s Climate Data Online portal, hosted by the National Centers for Environmental Information (NCEI). It provides free access to quality-controlled daily, monthly, seasonal, and yearly measurements of temperature, precipitation, wind, and degree days, along with 30-year climate normals. You can search by station, zip code, or city, and the archive includes thousands of weather stations across the country, some with records going back well over a century.
The data downloads in CSV format, which opens easily in Excel or Google Sheets. If you’re trying to answer a simple question like “how much rain fell in Denver on June 3, 2014” or “what was the average temperature in Chicago during the winter of 1985,” this is the place to go. NCEI also hosts radar data and station history information, so you can check whether a station relocated or changed instruments during the period you’re looking at.
Global Reanalysis: ERA5
If you need weather data for a location that doesn’t have a nearby weather station, or you need consistent global coverage, the ERA5 dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF) is the gold standard. ERA5 covers the entire planet on a 31-kilometer grid, with data going back to January 1940 and running to the present. It resolves the atmosphere using 137 vertical levels from the surface up to 80 kilometers high.
Rather than relying on individual station measurements, ERA5 blends millions of observations with atmospheric models to produce a complete, gap-free picture of past weather. This makes it especially useful for ocean locations, remote areas, and anywhere station coverage is sparse. The dataset is free to access through the Copernicus Climate Data Store, though the files are large and typically require some comfort with scientific data formats like NetCDF. Python users commonly work with ERA5 through libraries like xarray.
NASA POWER for Solar and Agricultural Data
NASA’s POWER project fills a specific niche: solar radiation and meteorological data derived from satellite observations and assimilation models. The parameters are tailored to three communities: renewable energy, sustainable infrastructure, and agroclimatology. If you’re sizing a solar panel system, estimating crop water needs, or evaluating a building’s energy performance, POWER provides the right variables in the right format.
Data is available at hourly, daily, monthly, annual, and climatological time scales. You can search by keyword (temperature, solar, wind) or browse by category, and the platform filters parameters by what’s actually available at your chosen time resolution. The interface is straightforward enough for non-programmers, and there’s also an API for automated access.
International Station Records
For historical weather data from countries outside the United States, the World Weather Records collection (also hosted by NCEI) compiles station data supplied by national meteorological services from around the world, many of them members of the World Meteorological Organization. These records are organized by decade and cover standard measurements like temperature and precipitation.
Another resource is the KNMI Climate Explorer, maintained by the Royal Netherlands Meteorological Institute. It hosts a variety of climate indices, interpolated datasets, and station metadata. While its primary focus is the Netherlands, the Climate Explorer tool itself can access global datasets and is widely used by researchers to analyze climate variability and trends.
Severe Weather and Radar Archives
If you’re looking for historical storm data rather than daily temperature and precipitation, the Iowa Environmental Mesonet (IEM) at Iowa State University maintains one of the most comprehensive public archives of National Weather Service products. This includes watches, warnings, and advisories (searchable by county or zone), special weather statements, and Storm Prediction Center convective outlooks.
The IEM also archives NEXRAD radar data. For Level II radar files (the full-resolution data), they point users to Amazon’s S3 NEXRAD archive as the primary source. The IEM’s own archive includes NEXRAD composite images and NIDS products from select sites, with complete product archiving going back to mid-April 2002. If you want to see what a particular storm looked like on radar or verify whether a tornado warning was issued for your county on a specific date, this is where to look.
Commercial APIs for Developers
When you need historical weather data delivered through an API for software applications, dashboards, or automated workflows, commercial providers offer a more streamlined path than downloading government files.
OpenWeatherMap’s One Call 3.0 API includes historical data reaching back more than 47 years. The first 1,000 API calls per day are free, with each additional call costing $0.0015. Students get expanded free access, including the History API and Statistical Weather Data API, with limits of 60 calls per minute and 1 million calls per month. Higher-volume professional plans start at $470 per month for up to 1 billion calls.
Visual Crossing is another popular option, particularly for users who want to download bulk data without writing code. You can export results in CSV, JSON, or Excel format with no setup required, which makes it easy to connect to business intelligence tools or data warehouses. Their web interface lets you pull historical data for a date range and location, then download the file directly.
Pre-Satellite Era and Ship Logs
Weather records from before the mid-20th century get increasingly sparse, especially over oceans. The Old Weather project, supported by NOAA, addresses this gap through citizen science. Volunteers transcribe weather observations from 19th-century ship logbooks preserved by the U.S. National Archives and other repositories. These logs contain temperature, pressure, wind, and sea state observations that, once digitized, can be fed into computer models to improve our understanding of historical climate patterns. You can browse completed datasets or contribute to the transcription effort at oldweather.org.
Choosing the Right Source
- Specific U.S. station data: NOAA Climate Data Online. Free, simple, and covers most needs.
- Global gridded data or remote locations: ERA5 through the Copernicus Climate Data Store. Free but requires some technical skill.
- Solar radiation and crop modeling: NASA POWER. Free, with a user-friendly interface.
- Storm history and radar imagery: Iowa Environmental Mesonet. Free, with deep NWS archives.
- App integration or quick prototyping: OpenWeatherMap or Visual Crossing. Free tiers available, with paid plans for higher volume.
- International station records: World Weather Records via NCEI, or the KNMI Climate Explorer for European data and global indices.