Climate Data Sources

Updated May 2026
Climate science relies on a vast global observation network including surface weather stations, ocean buoys and floats, weather balloons, satellites, and ice core archives. Multiple independent datasets produced by different research groups using different methods yield consistent results, providing robust evidence of climate change. Understanding these data sources helps evaluate the reliability of climate science conclusions.

Surface Temperature Records

Four major groups independently produce global surface temperature datasets: NASA GISS (GISTEMP), NOAA NCEI (NOAAGlobalTemp), the UK Met Office with the University of East Anglia (HadCRUT), and Berkeley Earth. Each uses different station selections, quality control methods, spatial interpolation techniques, and ocean data treatments. Despite these methodological differences, all four produce nearly identical warming estimates (within 0.1 degrees), demonstrating the result is robust to analytical choices.

The raw data comes from roughly 30,000 land weather stations, ship-based measurements, and buoys. Quality control removes errors from station moves, instrument changes, urban heat contamination, and time-of-observation biases. Homogenization algorithms detect and correct artificial discontinuities by comparing each station with its neighbors. These adjustments are well-documented and publicly available for scrutiny.

Satellite Observations

Since 1979, microwave sounding units on polar-orbiting satellites measure atmospheric temperature profiles. Two main groups (UAH and RSS) process these data, producing lower troposphere temperature records that broadly agree with surface measurements though with greater uncertainty due to complex retrieval algorithms and instrument drift corrections. Satellites provide global coverage including over oceans and remote regions poorly served by surface stations.

Other satellite systems measure ice sheet mass (GRACE gravity missions), sea level (radar altimeters like Jason and Sentinel-6), sea ice extent and thickness (passive microwave and altimeters), vegetation indices, ocean color, radiation balance (CERES), and atmospheric CO2 (OCO-2, GOSAT). These provide global, continuous measurements impossible from surface networks alone.

Ocean Observation

The Argo network deploys roughly 4,000 autonomous floats throughout the global ocean, each profiling temperature and salinity from the surface to 2,000 meters every 10 days. Since reaching full deployment around 2007, Argo has revolutionized ocean heat content measurement. Deep Argo floats sampling to 6,000 meters are being deployed to capture warming below standard Argo depths.

Tide gauges at coastal stations have measured sea level since the mid-19th century. These long records are complemented since 1993 by satellite altimeters providing precise global measurements. The combination allows separation of local land subsidence from true ocean volume changes.

Atmospheric Composition

The Mauna Loa Observatory has continuously measured atmospheric CO2 since 1958 (the Keeling Curve), providing the iconic record of rising concentrations. A global network of background monitoring stations now tracks CO2, methane, nitrous oxide, and other gases at dozens of locations. Flask samples and continuous analyzers confirm that increases are truly global and not local contamination artifacts.

Reanalysis Products

Reanalyses combine all available observations (surface, upper air, satellite) with weather forecast models to produce spatially complete, physically consistent records of atmospheric conditions. Major products include ERA5 (ECMWF), MERRA-2 (NASA), and JRA-55 (Japan). While not pure observations, reanalyses fill gaps and provide consistent three-dimensional fields useful for climate monitoring and model evaluation.

Key Takeaway

Climate data comes from multiple independent systems (stations, satellites, ocean floats, ice cores) processed by different groups with different methods. Their consistent results demonstrate that observed warming is robust and not an artifact of any single measurement system or analytical approach.