Climate and weather datasets for research
Pick a source by what kind of climate question you are asking. For a spatially continuous historical record, use ERA5 reanalysis. For actual station measurements at fixed points, use NOAA's Global Historical Climatology Network. For future projections under different emissions scenarios, use CMIP6. For satellite-observed variables like precipitation, land cover or fire, use NASA Earthdata. All four are free; the differences are in format, resolution and how much preprocessing you will need to do yourself.
The short answer
ERA5, from the Copernicus Climate Data Store, is the standard general-purpose reanalysis: continuous global gridded data from 1940 to near-present, free with a Climate Data Store account. NOAA's GHCN and Climate Data Online give you real station observations rather than modeled grids, also free with a registration token. CMIP6, distributed through the Earth System Grid Federation, is where you go for future climate projections used in IPCC assessments. NASA Earthdata is the hub for satellite-derived variables — precipitation, vegetation, land surface temperature, fire — and requires a free Earthdata Login. None of the four charge for access; the real cost is download volume and file format handling, which can be substantial for gridded and satellite products.
Reanalysis: spatially continuous historical data
ERA5 (Copernicus Climate Data Store)
ERA5 is the fifth-generation ECMWF atmospheric reanalysis, produced by the Copernicus Climate Change Service and covering January 1940 to near-present with hourly estimates of atmospheric, land and ocean variables on a roughly 31km global grid, with atmospheric fields also available on 37 pressure levels. Because it blends a physical model with observations, it gives continuous coverage everywhere on Earth, including oceans and regions with sparse station networks, which raw station data cannot do. Data are free to download from the Climate Data Store after creating an account; a Python API (cdsapi) is provided for programmatic and bulk retrieval, and there is a companion higher-resolution ERA5-Land product for land-surface applications. New days are added with about a five-day lag behind real time. The tradeoff is that ERA5 values are model-informed estimates, not direct measurements, so for point-level station validation work you still need a source like GHCN. Official site: cds.climate.copernicus.eu.
Station observations
NOAA Global Historical Climatology Network (GHCN)
GHCN, maintained by NOAA's National Centers for Environmental Information, integrates daily, monthly and hourly climate observations from roughly 30 underlying data sources. The daily version (GHCNd) covers more than 90,000 stations worldwide, about two-thirds of which report precipitation only, with more than 25,000 stations updated within roughly the last month; it includes maximum and minimum temperature, precipitation, snowfall and snow depth. The dataset is rebuilt weekly with quality-assurance checks applied across the full record. Raw files are free to download as plain text requiring your own parsing, or you can use NCEI's ordering and mapping system to export a subset by station, city or region directly as CSV. This is the right source when your research needs actual instrument readings and station metadata rather than a modeled grid. Official site: ncei.noaa.gov — GHCN-Daily.
NOAA Climate Data Online (CDO) API
Climate Data Online is NOAA's web service layer on top of GHCN and related NCEI holdings, meant for developers who want to query specific stations, locations or data types programmatically rather than downloading bulk files. It is free but requires registering for an API token by email; each token is limited to five requests per second and 10,000 requests per day, which is enough for most research pulls but worth knowing if you plan a very large automated extraction. Responses are JSON over a REST API. This is a practical complement to raw GHCN files when you need a scripted, repeatable pull for a defined set of stations rather than a full archive download. Official site: ncei.noaa.gov/cdo-web.
Future climate projections
CMIP6 via the Earth System Grid Federation (ESGF)
CMIP6, the Coupled Model Intercomparison Project Phase 6, coordinates output from dozens of global climate models run by modeling centers worldwide under a common set of experiment protocols, including historical simulations and future scenarios expressed as Shared Socioeconomic Pathways (SSPs). It is the primary data basis for IPCC assessment report climate projections. The archive is distributed and federated rather than centralized: you search and download through ESGF portals such as PCMDI/LLNL (United States), IPSL (France), DKRZ (Germany) or CEDA (United Kingdom), which all index the same underlying federated holdings, though download speed varies by node. Access is free but requires accepting CMIP6's terms of use, and bulk downloads are typically done via Wget scripts generated by the search interface, or through Globus for large volumes. Expect to spend real time learning the model/experiment/variant naming conventions before your first successful query, since the file-naming scheme encodes model, scenario, ensemble member and variable all at once. Official site: pcmdi.llnl.gov/CMIP6.
Satellite-derived data
NASA Earthdata (MODIS, GPM and related missions)
NASA Earthdata is the umbrella access point for NASA's Earth science data holdings, more than 12,400 datasets spanning land, ocean, atmosphere and cryosphere, all free and openly available. Two commonly used products are MODIS, which provides land surface temperature, vegetation indices and fire detection from the Terra and Aqua satellites, and GPM (Global Precipitation Measurement), a joint NASA-JAXA mission providing global rain and snow estimates used in hydrology and disaster-response work. The Worldview tool lets you browse and visualize more than 1,200 satellite imagery products without logging in, and links through to Earthdata Search for the underlying data files. Downloading the actual data requires a free Earthdata Login account. This is the right first stop for any research question framed around a satellite-observed variable rather than a surface station or reanalysis grid. Official site: earthdata.nasa.gov.
Climate and weather sources side by side
| Source | Data type | Coverage | Access |
|---|---|---|---|
| ERA5 | Gridded reanalysis, hourly | Global, 1940–near present, ~31km grid | Free, CDS account + API |
| NOAA GHCN | Station observations, daily/monthly/hourly | 90,000+ stations worldwide | Free, raw files or export tool |
| NOAA Climate Data Online | Station data via API | Same underlying NCEI archive | Free, API token by email |
| CMIP6 / ESGF | Global climate model output, future scenarios | Dozens of models, multiple SSPs | Free, federated portals, terms of use |
| NASA Earthdata | Satellite-derived variables | 12,400+ datasets, global | Free, Earthdata Login required |
How to choose
If you need a value at every point on Earth for a historical period — no gaps over oceans or remote regions — use ERA5. If your research question depends on what an actual instrument measured at a specific place, use GHCN, and reach for the Climate Data Online API instead of raw file parsing if you only need a defined set of stations. If the question is about the future rather than the past, CMIP6 through ESGF is close to the only comprehensive option, and you should decide early which SSP scenarios and which subset of models you need, since the full archive is far larger than any single project requires. If your variable of interest is something a satellite observes directly — precipitation, vegetation, fire, sea ice — start at NASA Earthdata rather than trying to derive it from reanalysis or station data.
FAQ
What is the difference between ERA5 and NOAA GHCN data?
ERA5 is a gridded reanalysis that blends model output with observations for continuous global coverage, including oceans and station-sparse regions. GHCN is raw station observation data with no interpolation. Use ERA5 for spatially continuous coverage; use GHCN for real station-level measurements and metadata.
Where do I get future climate projections instead of historical data?
Use CMIP6, distributed through the Earth System Grid Federation. It contains output from dozens of global climate models under multiple future emissions scenarios (SSPs) and underlies most IPCC assessment report projections.
Do I need an account to download NASA satellite climate data?
Yes. A free Earthdata Login is required to download most NASA Earthdata datasets, including MODIS and GPM products, through Earthdata Search. Worldview lets you browse imagery without logging in, but downloading the underlying files still requires an account.
Not sure which climate dataset fits your research question?
If you are deciding between reanalysis, station data, projections or satellite products, or need to confirm a specific region, variable or time range is actually covered before you commit, you can give us your research question and required conditions, and we run a data availability assessment (3 free credits on sign-up) — searching these sources for real and judging matches and gaps against each requirement. Even when no perfectly fitting dataset is found, we present the search directions, approximate sources and item-by-item findings honestly for your reference.
New accounts get 3 free searches; every result comes with verifiable source links.