Global Temperature Time Series
- Rows
- 323
- Columns
- 3
- Data range
- 1850-01-01 to 2026-01-01
- Auto-refresh
- Re-checks its source every 14 days
- Last refreshed with new data
- 2026-11-05
Annual and monthly global mean surface temperature anomalies (°C) from 1850–present. Sourced from NASA GISTEMP and NOAA Climate at a Glance (GCAG). Two series provided: global monthly mean and annual mean. Ideal for long-run climate trend analysis.
AI analysis
Global Temperature Time Series — Dataset Analysis
1. Dataset Overview & Value for Correlation Studies
This dataset captures annual global mean surface temperature anomalies (°C) spanning 1850 to 2026, drawing from two authoritative climate science institutions: NASA GISTEMP and NOAA's Global Climate at a Glance (GCAG) series. The dual-source structure — confirmed by the Source column holding exactly 2 distinct values — effectively doubles the row count to 323 across what would otherwise be ~176 years of observations, enabling side-by-side methodological comparison between the two measurement frameworks. The source URL (https://datahub.io/core/global-temp/) places this data on Datahub.io, a well-maintained open data registry known for versioned, machine-readable core datasets, which lends confidence to its freshness and reproducibility. The long temporal horizon makes this dataset exceptionally well-suited for studying secular climate trends, regime shifts, and the correlation of warming patterns with human activity indicators such as CO₂ emissions, industrial output, and land-use change.
2. Data Quality Observations
Data quality here is notably strong across all three columns. Total null cells equal zero, meaning no missing values exist in Year, Source, or Mean — a rare completeness for a dataset spanning 176 years of historical records. No type mismatches are indicated in the column metadata, with each field conforming cleanly to its declared type (Date, String, and Decimal respectively). The duplicate-row count is listed as pending Phase D recomputation against Parquet, so a definitive deduplication verdict is outstanding; however, the 323-row count against 177 distinct Year values and 2 distinct Source values is arithmetically consistent (177 × 2 = 354 theoretical maximum, with 323 actual rows suggesting ~31 year-source combinations may be absent, likely at the leading or trailing edges of one series). This warrants a row-completeness check — specifically whether GCAG or GISTEMP is missing early observations prior to a certain year.
3. Key Column Distributions
The Mean column is the analytical centerpiece and tells a compelling story on its own. The overall mean anomaly is just +0.0076°C, but the median sits at −0.1067°C, and together with a Q1 of −0.290°C and Q3 of +0.212°C, this reveals a distribution that is substantially left-anchored in cooler historical territory. The positive skew of 1.01 confirms a long right tail — recent decades are pulling the distribution upward with anomalies approaching and exceeding the maximum of +1.285°C, while the minimum of −0.597°C reflects the coldest pre-industrial baseline years. The 10 flagged outliers almost certainly cluster in the post-2000 period, representing the accelerating warming signal of the 21st century. The standard deviation of ±0.407°C across the full series underscores how dramatic a 1.2°C anomaly truly is in context — roughly 3 standard deviations above the long-run mean. The Year column's 177 distinct values against a declared range of 1850–2026 is internally consistent and functions reliably as a temporal index.
4. Recommended Join Key Columns
The optimal join strategy for cross-dataset correlation uses Year (as a 4-digit integer or standardized date) combined with Source as a composite key. For most external datasets that do not distinguish measurement source, joining on Year alone at annual granularity will be sufficient and is the most portable option. Analysts should normalize the Year column from its ISO timestamp format (1850-01-01T00:00:00) to a plain integer year before joining, as many partner datasets encode year as INT or YYYY string. If pairing with the monthly series also available from this publisher, a Year-Month composite key would unlock finer-grained lag-correlation analysis (e.g., ENSO cycles vs. temperature spikes).
5. Suggested Paired Datasets for Correlation Discovery
Several dataset categories would pair powerfully with this series:
- Atmospheric CO₂ Concentrations (e.g., Mauna Loa / NOAA): The canonical climate correlation — joining on Year would let analysts model the CO₂-temperature lag relationship and test radiative forcing hypotheses directly. - Global Fossil Fuel CO₂ Emissions (CDIAC / Our World in Data): Correlating cumulative emissions with temperature anomaly trajectory would illuminate the anthropogenic signal strength across industrial eras. - Arctic & Antarctic Sea Ice Extent (NSIDC): Temperature anomalies likely lead sea-ice loss; a lagged cross-correlation analysis could quantify the response time. - Global Sea Level Rise Records (NASA / CSIRO): Thermal expansion is a primary driver of sea level change, making this a natural dependent-variable pairing. - El Niño/La Niña Index (ENSO/ONI): Overlaying the MEI or ONI index on annual temperature anomalies would allow analysts to decompose natural variability from the long-term trend, significantly improving any regression model built on this data.
Columns
- Year (date)
- Source (string)
- Mean (decimal)