Climate proxies
Plant leaves keep a growing-season count of pores. Antarctic ice keeps bubbles of old air. Nineteenth-century chemists titrated the air itself. Mauna Loa is still measuring. Temperature on the Common Era chart is a see-through overlay: European summers (Luterbacher 2016, no thermometer splice) and Ljungqvist’s hemisphere composite. Other temperature series stay off.
1860–2026
Ice climbs from ~286 to 370 ppm by the late 1990s. Mauna Loa and the NOAA marine network continue the record to 2026 (~429 ppm). Stomatal density tracks the rise once a 3-point mean is taken. Beck’s 2007 chemical compilation spikes through 430 ppm around 1942 — a peak ice, stomata and the modern analyser record do not share. Temperature (HadCRUT5, UAH) is off by default — turn it on in Records.
Dashed sage = stomata 3-point means. Dotted copper = Beck 2007. Ice samples are points. Mauna Loa / NOAA marine run through 2026. Temperature is off until you turn it on — HadCRUT5 analysis vs non-infilled, UAH satellite TLT.
1–2026
Koutsoyiannis 2024 sources: Kouwenberg’s two-millennium Tsuga reconstruction against Law Dome / WAIS ice. Translucent rust is Luterbacher 2016 European summer temperature (no instrumental splice). Pale gold is Ljungqvist 2010 extra-tropical NH. Olive dashed line is Stomata T — the leaf CO₂ mean drawn as temperature (CPS to HadCRUT5 1860–2000). PAGES 2k stays off.
Translucent rust = Euro-Med JJA 31-year mean (Luterbacher 2016, noInst). Pale gold = Ljungqvist 2010 extra-tropical NH. Olive dashed = Stomata T. Faint dash is the archived leaf series once the method slider is off zero.
Not an added offset. Kouwenberg’s needles were already a 3-point mean from a straight-line TSDL fit on herbarium leaves spanning 290–370 ppm. The slider applies the later steps that raise adherence to the ice core: a longer smooth, shrinkage of departures from that 330 ppm training center (stronger outside the window, where a linear calibration is weakest), then a partial anchor to coeval Law Dome. The solid curve is that processed series. The faint line stays on the archive. Around AD 225 the archive is 357 ppm.
Holocene · ka BP
As far back as stomata still have clustered, reproducible Holocene records. Ice (Dome C, WAIS) is continuous and tight until the firn. Mauna Loa is the thin CO₂ spike on the far right. Temperature (Dome C deuterium, HadCRUT5, UAH) is off by default.
Axis is thousands of years before present; the last tick is now. Temperature (Dome C ΔT, HadCRUT5, UAH) is off until you turn it on.
25 series
Pre-industrial Law Dome wanders ~15 ppm. Kouwenberg over the same centuries wanders ~106 ppm. Stomata authors quote ±36 to ±40 ppm reconstruction error — larger than ice-core analytic precision (~1 ppm). Hover year 225.
Ice cores are NOAA originals. The two-millennium stomata curve is the 3-point mean from Kouwenberg 2004 Fig. 5.4 — the series Koutsoyiannis 2024 pairs with ice. van Hoof 2008 is pinned to published extrema. Wagner 2004 supplies the 8.2 ka Holocene snapshot. Beck 2007 is the 11-year chemical compilation on the industrial chart. Mauna Loa and the NOAA marine network run through 2026.
Buried western-hemlock needles, ~AD 1–2000. Four CO₂ minima of 250–275 ppm (ca. AD 730, 1040, 1550, 1800) and a Roman-era maximum of ~357 ppm — about 80 ppm above coeval Law Dome ice. Linear TSDL–CO₂ calibration on herbarium material over 290–370 ppm.
Kouwenberg 2004 PhD thesis, LPP Contributions 16; Kouwenberg et al. 2005, Geology 33:33–36
Digitized from a published figure
Species Tsuga heterophylla
The Holocene ice series Kouwenberg plots as white diamonds. Superseded here by the Bereiter 2015 Dome C / WAIS compilation, which includes that interval.
Nature 398:121–126
The ice series Kouwenberg plots as white squares. Rubino et al. 2019 (Law Dome spline) is the revised version of the same cores.
J. Geophys. Res. 101:4115–4128
Splan Pond, 11,620–10,920 cal BP, with a CO₂ minimum at 11,120 cal BP. Same amplitude as the 8.2 ka panel plotted above; not vector-clean enough to extract as a second line.
Quat. Sci. Rev. 23:1947–1954
Salix stomata across the YD stadial. A rare case where stomata and Antarctic ice both show a decline — still with larger stomatal scatter.
J. Quaternary Sci. 17:21–29
The longest Holocene stomatal-index series (Salix herbacea, northern Sweden). Not in NOAA Paleo-pCO₂ (that archive is Cenozoic) and not digitized here.
The Holocene 9
Station-level quality-controlled unadjusted monthly temperatures. NOAA does not publish a ready-made global mean of the unadjusted file; a cosine-weighted global average from ~27,000 stations is a separate reconstruction, not plotted here. HadCRUT5 non-infilled is the official ‘no empty-cell filling’ option.
Menne et al. 2018; ncei.noaa.gov/pub/data/ghcn/v4
Nature wavelet reconstruction mixing low-resolution lake/ocean proxies with tree rings. Larger millennial swings than Mann; Medieval warmth similar to the 20th century before 1990. Weaker Roman Warm Period than Ljungqvist. NOAA file is available; not plotted to keep one translucent overlay.
Nature 433:613–617; ncei.noaa.gov …/moberg2005/nhtemp-moberg2005.txt
Non-tree-ring 2000-year reconstruction. Shows a strong Medieval peak. Not in the NOAA contributions_by_author tree as a clean txt, so not plotted; Ljungqvist is the archived series that names both Roman and Medieval warm periods.
Energy & Environment 18 and 19
Each series as a CSV, plus a zip of those files with the original NOAA tables. Ice and Mauna Loa are tidy extracts of public archives. Stomata and Beck have no NOAA time series — those CSVs are our digitizations, so the paper link is the source of record. Some publisher pages are paywalled; the NOAA file links are the ones that resolve without a login.
Public archive table
Public archive table
Public archive table
Public archive table
Public archive table
Public archive table
Public archive table
Public archive table
Public archive table
Public archive table
Digitized from a published figure
Digitized from a published figure
Digitized from a published figure
Digitized from a published figure
Reconstructed from published extrema
Digitized from a published figure
Public archive table
Public archive table
Public archive table
Public archive table
Public archive table
Reconstructed from published extrema
Public archive table
Public archive table
Public archive table
Pairing and Common-Era window follow Koutsoyiannis, D. (2024), Stochastic assessment of temperature–CO₂ causal relationship in climate from the Phanerozoic through modern times, Mathematical Biosciences and Engineering 21:6560–6602. That paper compiles Kouwenberg 2004 (stomata, from ~AD 200), van Hoof et al. 2008 (oaks, 1000–1500), and Antarctic ice (Indermühle 1999, Etheridge 1996 / Francey 1999).
Compilation and Common-Era pairing follow Koutsoyiannis 2024, Math. Biosci. Eng. 21:6560–6602: stomata from Kouwenberg 2004 (thesis Fig. 5.4) and van Hoof et al. 2008, ice from Etheridge/Rubino Law Dome and Indermühle/Monnin Antarctic cores.
Ice-core series are original NOAA NCEI / NOAA GML files. The Kouwenberg 0–2000 curve is the 3-point mean extracted from the vector figure in the 2004 thesis — the same figure Koutsoyiannis digitized.
van Hoof 2008 Table S1 is not archived as a NOAA time series; the curve here is pinned to the published 292.3–319.2 ppm extrema and the 1000–1500 shape described in the PNAS paper.
Further back than ~2,000 years, stomata coverage is no longer a continuous millennial series. Wagner 2004 still has dense, multi-site Holocene snapshots (Preboreal, 8.2 ka, last millennium). Those event windows are plotted against continuous Dome C / WAIS ice.
Ice cores average air over years to a decade as bubbles close. Stomata record a growing-season snapshot with large biological scatter. A 3-point mean is what the stomata authors compare to ice.
Beck 2007 compiled ~90,000 chemical (Pettenkofer-style) analyses, 1812–1961, into a northern-hemisphere yearly series. The curve plotted here is that paper’s 11-year smooth. These are local, near-ground samples, not well-mixed marine air; the 1942 maximum (>420 ppm) is the feature ice cores and stomata do not reproduce.
Instrumental CO₂ is NOAA GML: Mauna Loa annual means 1959–2025 and the global marine-boundary network 1979–2025. 2026 is year-to-date from the NOAA deseasonalized monthly trend (Mauna Loa through July, global through May). Ice cores do not reach 2026 — bubble close-off ends the high-resolution Law Dome record in the 1990s. Stomata series end with their published material (Finsinger 2000, Kouwenberg ~2000).
Temperature is on the right axis. Common Era defaults to two translucent overlays: Luterbacher 2016 Euro-Med summer (noInst, 31-year mean) and Ljungqvist 2010 extra-tropical NH. Stomata T is an artificial series: the unweighted mean of independent stomata CO₂ (Kouwenberg, van Hoof, Finsinger), composite-plus-scaled to HadCRUT5 over 1860–2000 — industrial mean and spread match the thermometer by construction; the Roman/Medieval shape is the leaves. Christiansen & Ljungqvist 2012 LOC and PAGES 2k stay off. HadCRUT5 and UAH stay off. None of the published proxy temperature series are spliced to thermometers.
NOAA GML analyser files: trends data page. Law Dome 2018: NCEI study 25830. Kouwenberg thesis: Utrecht dspace. van Hoof 2008: PNAS. Temperature: HadCRUT5 (analysis and non-infilled) , UAH v6.1 TLT, Euro-Med 2k (noInst), Ljungqvist 2010, PAGES 2k, EPICA Dome C ΔT.
The industrial rise is the one feature both families agree on. Where they disagree is amplitude, and that disagreement grows as the window opens. Pre-industrial Law Dome is almost flat. Kouwenberg’s Roman maximum sits ~80 ppm above the same ice. Wagner’s 8.2 ka Betula snapshot swings ~80 ppm while Dome C moves a few.