Which early works are cited most frequently in climate change research literature? A bibliometric approach based on Reference Publication Year Spectroscopy

1. Introduction

• The analysis seeks early works that remain frequently cited and help identify the origins and intellectual roots of climate-change research.Historians have cited signpost works inconsistently, motivating an approach based on the research community’s references.Source: §1 · p. 3
• RPYS treats cited references as a community-generated basis for identifying works most relevant to a research field.The approach uses publication-year peaks among references as indicators of frequently cited early publications.Source: §1 · pp. 3–4
• RPYS-CO addresses the dilution of historical references by selecting references co-cited with a prominent marker work.The method uses the marker’s citation environment to focus analysis on a specific historical topic, here the greenhouse effect and carbon dioxide.Source: §1 · p. 4
• The authors use Arrhenius’s 1896 paper as the marker for their focused analysis of greenhouse-effect research.The proposed approach selects a smaller, topic-specific publication set based on co-citations with the marker.Source: §1 · pp. 4–5
• The study presents RPYS and RPYS-CO results for climate-change research, using 10,932,050 total references cited within a carefully searched set of 222,060 relevant papers; it also examines 1,658 co-cited references of a single paper.Source: §1 · pp. 4–5

2.1 Climate change research literature in total

• The search produced 222,060 climate-change articles and reviews published from 1980 to 2014.The search used interactive query formulation and keywords developed from a preliminary set of key papers.Source: §2.1 · p. 5
• The publication set is not exhaustive, though the authors expect it to include most relevant papers, especially key climate-change studies.They expect moderate incompleteness not to bias RPYS, while broader queries could add non-relevant records.Source: §2.1 · p. 5
• The analysis drew on 10,932,050 references from the 222,060-paper set and restricted the imported references to publication years before 1971.The pre-1971 restriction reduced the reference volume for processing.Source: §2.1 · p. 5
• Reference variants were algorithmically clustered, with manual clustering applied to references published before 1900.For the 1000–1900 period, references cited fewer than 10 times were removed, leaving 152 references for further analysis.Source: §2.1 · p. 6

4. For detecting important historical publications newer than 1900, we selected the reference

• For references published from 1901 to 1970, the authors retained works cited at least 100 times to reduce the number analyzed.The passage describes this threshold as a way to reduce the flood of references.Source: §4 · p. 6

2.2 Climate change research literature citing the marker paper

• The RPYS-CO analysis targets the greenhouse effect and carbon dioxide by retaining references co-cited with Arrhenius’s 1896 paper.The resulting set contains 1,658 co-cited references and has no minimum reference-count threshold.Source: §2.2 · pp. 6–7
• Arrhenius’s paper received fewer than ten citations per year until 1990, after his theory had been widely refused in the first half of the 20th century.The passage notes that his work later became a cornerstone in climate-change research.Source: §2.2 · p. 7
• RPYS-CO reveals early works relevant to greenhouse-effect history that generally do not form distinct peaks in the full RPYS, apart from Arrhenius and Chamberlin.The focused analysis is intended to identify forerunners and later works that took up Arrhenius’s impetus.Source: §2.2 · p. 7

3.1 Historical roots of climate change research

• 1686 is the earliest visible cited-reference year, while the original set reaches back to 1002, with no years between them exceeding 10 citations.Source: §3.1 · pp. 7–8
• Peaks become less distinct for more recent references as the growing number of different references per year smooths both counts and deviations from the median.Source: §3.1 · p. 10
• Pronounced peaks occur across the reference-year distributions, and a peak can contain multiple important references, as in 1941 and 1965.The study notes that peaks comprise most or all cited references for their corresponding years.Source: §3.1 · p. 10
• Reference counts in the table are lower than total citation counts because many citations fall outside the climate-change set, and counts across publication years are not comparable.The publication set also excludes some relevant papers lacking abstracts or query terms in their titles or abstracts.Source: §3.1 · p. 18
• Many frequently cited early works are major publications in science generally, only loosely connected to climate change, including Linnaeus (CR3), Malthus (CR4), and Darwin (CR6).Other disciplinary foundations include meteorology works by Halley (CR1) and Hadley (CR2), and Bergmann’s biology work (CR5).Source: §3.1 · pp. 18–19
• Arrhenius’s 1896 paper (CR8) is identified as the origin of modern climate-change research, following the early-19th-century discovery of ice ages and the greenhouse effect.The paper is also used as the marker reference for the co-citation analysis.Source: §3.1 · p. 19
• Frequently cited works from the first half of the 20th century are seminal contributions in meteorology, agronomy, and climatology, including Walker and Bliss (CR11), Jenny (CR12), Penman (CR16), and Thornthwaite (CR15).Later cited works span geoscience and paleoclimate; the paper also identifies Milankowitch (CR13) and Lorenz (CR22) as cornerstones with strong climate or meteorology connections.Source: §3.1 · p. 19

3.2 Climate research literature citing the marker paper

• The Arrhenius co-citation spectrogram shows two peaks before 1896 and six after it, including pronounced peaks in 1908, 1924, 1938, and 1956/1957.Only Arrhenius’s 1896 paper and Chamberlin’s 1899 paper appear in both the broad and marker-focused reference lists.Source: §3.2 · pp. 20–21, Figure 3 · p. 21
• The 1800–1850 analysis highlights Fourier’s 1824 and 1827 works: he explained Earth’s warmth through atmospheric transparency to solar radiation and absorption of outgoing infrared radiation, identifying the natural greenhouse effect.The 1827 paper reproduces the 1824 work, which is based on Fourier’s 1822 analytical theory of heat.Source: §3.2 · pp. 20, 22
• Fourier’s correct reference year, 1824, was confused with 1924 in 17 of 25 references, an error traced to the database producer.The comparison with the original papers indicated that citing authors did not cause the mistake.Source: §3.2 · p. 24
• Fourier (COR1-2) is identified as the first to detect the greenhouse effect, while Tyndall (COR3-4) first measured greenhouse gases’ infrared heat absorption.The cited works include carbon dioxide among the greenhouse gases measured by Tyndall.Source: §3.2 · p. 25
• Chamberlin (COR6-8) proposed CO2-related climate change, Callendar (COR10-11) attributed prior warming to CO2, and Plass (COR12-14) predicted a 3-4°C rise from doubled CO2.Plass calculated radiation transmission through Earth’s atmosphere.Source: §3.2 · p. 25
• Revelle and Suess (COR15) described fossil-fuel use as a human-initiated “large-scale geophysical experiment.”The passage presents this as their first such observation.Source: §3.2 · p. 25

4. Discussion

• The broad RPYS analysis draws on references from 222,060 climate-change papers published between 1980 and 2014.Source: §4 · p. 25
• The analysis presents 35 prominent early publications, from fundamental nineteenth-century science to climate-related cornerstones, with later works concentrated in geosciences and related fields.Agriculture, paleoclimatology, and climate-change vulnerability also feature among the cited research.Source: §4 · p. 25
• Dansgaard (CR23) is important for reconstructing past climate, while Nash (CR35) is among the first papers on climate-change vulnerability.Source: §4 · pp. 25–26
• The bibliometric results differ from expert historical overviews partly because the approaches select and emphasize different works, including later papers by authors whose earlier work RPYS identifies.Le Treut et al. (2007) cites 13 of 46 references in TAB 1 and 4 of 15 in TAB 2.Source: §4 · p. 26
• Keeling’s contributions appear in the dataset but not as pronounced peaks, and their importance may be obscured because the data record spans decades rather than a single paper.The discussion describes this as possible “obliteration by incorporation.”Source: §4 · p. 26
• Using Arrhenius (1896) as a marker, RPYS-CO identifies greenhouse-effect and carbon-dioxide history works that mostly do not appear as distinct broad-RPYS peaks.Arrhenius (CR8) and Chamberlin (CR9) are exceptions.Source: §4 · p. 27
• Fourier’s 1822 and 1824 works appear as distinct peaks, while preceding contributions can be identified through the reference list; simple discovery claims are difficult to sustain.Fleming (1998) is cited in connection with the warning about discovery claims.Source: §4 · p. 27
• The two RPYS analyses are not expected to produce similar results: one covers many disciplines, while the other restricts attention to historically oriented climate-change publications.The broad analysis presents 35 prominent works out of almost 11 million cited references.Source: §4 · p. 28

• RPYS reveals publications important to the evolution of the climate change research field.Source: §5 · p. 28
• Historical overviews may omit relevant literature; a Web of Science search found only two papers citing both Fourier and Tyndall, and three citing both Fourier and Arrhenius.Source: §5 · p. 28
• RPYS provides a detailed list of potentially important publications, detected from references cited by the relevant community without further assumptions.Source: §5 · p. 28
• Because citation counts do not always reflect importance, experts are needed to determine cited papers’ roles in historical context.Source: §5 · p. 28
• The authors invite researchers to combine expert knowledge and bibliometrics for meaningful historical analyses.Source: §5 · p. 28