Mauna Loa Atmospheric CO₂ Annual Mean Concentrations
- Rows
- 67
- Columns
- 3
- Data range
- 1959 to 2025
- Auto-refresh
- Re-checks its source daily
Annual mean atmospheric CO₂ concentrations (ppm) measured at Mauna Loa Observatory, Hawaii, from 1959 to present – the iconic Keeling Curve dataset.
AI analysis
Mauna Loa Atmospheric CO₂ Dataset — Analytical Overview
1. Dataset Representation & Value This dataset captures the Keeling Curve — arguably the most important continuous environmental time series in modern climate science. Sourced directly from NOAA's Global Monitoring Laboratory (gml.noaa.gov), it records annual mean atmospheric CO₂ concentrations in parts per million (ppm) at Mauna Loa Observatory, Hawaii, spanning 1959 through 2025. The NOAA GML affiliation signals authoritative, peer-reviewed instrumentation and methodology, and the inclusion of 2025 data confirms the feed is current and actively maintained. The dataset's 66-year continuous record makes it exceptionally valuable for long-run correlation studies with industrial emissions, energy consumption, temperature anomalies, policy interventions, and economic indicators — essentially any annual global variable tied to human activity or climate response.
2. Data Quality Quality here is near-pristine. There are zero null cells across all 67 rows and 3 columns, and each year value is distinct (67 distinct values across 67 rows), confirming no duplicate records even before Phase D Parquet recomputation is finalized. All columns resolve to their expected types — Integer for year, Decimal for mean and unc — with no type-mismatch flags. The one notable quirk is the unc (uncertainty) column, which carries a constant value of 0.12 ppm across all 67 rows (σ = 0, skew = NaN due to zero variance, 1 distinct value). This column adds no analytical information and should be treated as metadata rather than a variable — it either reflects a standardized reporting convention or a data pipeline artifact worth verifying against the source CSV.
3. Key Column Distributions The mean CO₂ column tells a compelling story. Values range from 315.98 ppm (1959) to 427.35 ppm (2025), a rise of over 111 ppm in six decades. The dataset mean (361.25 ppm) sits noticeably above the median (356.54 ppm), and the mild positive skew of 0.379 reflects an accelerating upward trajectory — more recent years pull the distribution rightward. The interquartile range spans 331.58 to 386.74 ppm, with the lower quartile anchored in the mid-1970s and the upper quartile in the late 2000s, consistent with a near-linear but slightly steepening annual increase. There are no statistical outliers in the classical sense — every value is part of a monotonically increasing trend — but the rate of change itself (roughly 0.8 ppm/year in the 1960s vs. ~2.5 ppm/year recently) is the most analytically significant feature worth preserving in any derivative analysis. The year column is perfectly uniform (skew = 0, σ ≈ 19.5), confirming no gaps in the annual record.
4. Recommended Join Key year is the unambiguous join key for cross-dataset correlation. It is an integer, fully populated, has no duplicates, and spans a well-defined range (1959–2025) that overlaps with most modern annual global datasets. When joining, analysts should be mindful that 2025 data may be preliminary depending on ingestion timing, and should confirm whether partner datasets use calendar year, fiscal year, or a different annual convention before joining.
5. Suggested Pairing Datasets This dataset has exceptional correlation potential across several domains. Global mean temperature anomalies (NASA GISS or NOAA Global Surface Temperature) would allow direct testing of the CO₂–warming relationship year-by-year. Global fossil fuel CO₂ emissions (e.g., Global Carbon Project annual data) would illuminate whether atmospheric concentration tracks emissions linearly or shows absorption lag effects. Arctic/Antarctic sea ice extent (NSIDC annual records) could reveal lagged climate responses to concentration milestones. On the socioeconomic side, global GDP or energy consumption data (World Bank or IEA annual series) would help decompose emissions intensity trends, while renewable energy capacity additions (IRENA) could test whether the post-2015 clean energy build-out is beginning to bend the CO₂ curve. All of these datasets share the year integer as a natural join key.
Columns
- year (integer)
- mean (decimal)
- unc (decimal)