Lunar eclipses illuminate timing and climate impact of medieval volcanism

Abstract

Explosive volcanism is a key contributor to climate variability on interannual to centennial timescales¹. Understanding the far-field societal impacts of eruption-forced climatic changes requires firm event chronologies and reliable estimates of both the burden and altitude (that is, tropospheric versus stratospheric) of volcanic sulfate aerosol^2,3. However, despite progress in ice-core dating, uncertainties remain in these key factors⁴. This particularly hinders investigation of the role of large, temporally clustered eruptions during the High Medieval Period (HMP, 1100-1300 CE), which have been implicated in the transition from the warm Medieval Climate Anomaly to the Little Ice Age⁵. Here we shed new light on explosive volcanism during the HMP, drawing on analysis of contemporary reports of total lunar eclipses, from which we derive a time series of stratospheric turbidity. By combining this new record with aerosol model simulations and tree-ring-based climate proxies, we refine the estimated dates of five notable eruptions and associate each with stratospheric aerosol veils. Five further eruptions, including one responsible for high sulfur deposition over Greenland circa 1182 CE, affected only the troposphere and had muted climatic consequences. Our findings offer support for further investigation of the decadal-scale to centennial-scale climate response to volcanic eruptions.

Fig. 1. Representations of lunar eclipses in medieval manuscripts.

a, Commentary on the Apocalypse by Beatus of Liébana, from the monastery of Santo Domingo de Silos, near Burgos, Spain, 1090–1109 ce. Credit: British Library Board (Add. MS 11695, f108r). The text at the bottom of the miniature, between the dark circle on the left representing a total solar eclipse and the red circle on the right representing a total lunar eclipse, reads: “hic sol obscurabitur et luna in sanguine versa est” (“and the Sun was obscured and the Moon turned into blood”). The blood-red eclipsed Moon was seen as one possible sign of the Apocalypse. Lunar occultation descriptions from the Middle Ages often follow the Book of Revelation, suggesting that the Bible provided justification and inspiration for recording lunar eclipses and their colour. b, Thirteenth-century depiction of a lunar eclipse by Johannes de Sacrobosco. Credit: The New York Public Library (De Sphaera, MssCol 2557, f112v). c, Facsimile of the Meigetsuki (明月記) diary by Fujiwara no Teika (藤原定家) describing the total lunar eclipse of 2 December 1229 ce. Credit: Meigetsuki, vol. 4, pp. 517, 2000. Reizei-ke Shiguretei Bunko. Tokyo: Asahi Shinbunsha. Teika mentions this event twice. The figure shows the first entry: “[…] the sky was free of clouds into the distance and the Moon over the hills emerged in eclipse, total for a little while, [its light] meagre as on a dark night. About an hour later it brightened gradually, and after it was extinguished [during the eclipse] it was especially luminous”. The second entry, written four days later, details the unusual coloration of the Moon. Over the centuries, several portions of the Meigetsuki were cut apart, and the entry for 6 December 1229 ce is held in a private collection^,.

Fig. 2. Stratospheric turbidity derived from total lunar eclipse coloration and non-sea-salt sulfur records from polar ice cores.

a, Total lunar eclipse descriptions retrieved from European, Middle Eastern and East Asian historical sources from 1100 to 1300 ce (Supplementary Dataset S1), rated on the Danjon scale (right y axis), and converted to equivalent global mean SAOD₅₅₀ (SAOD at 550 nm; left y axis). b, Monthly resolved non-sea-salt sulfur concentrations from the Greenland NEEM-2011-S1 (blue line) and Antarctica WDC06A (grey line) ice cores. Source data

Fig. 3. Constraining the timing of HMP (1100–1300 ce) volcanic eruptions.

a, Residence time of volcanic stratospheric aerosols and time windows with SAOD exceeding about 0.1. The residence time of aerosols is based on global mean SAOD₅₅₀ time series from the Sato/GISS and GloSSAC v2 (ref. ) datasets (for the 1883 ce Krakatau and 1991 ce Pinatubo eruptions) and simulated by the EVA forcing generator^, (for UE2, UE4 and the 1257 ce Samalas eruption) and the IPSL-CM5A-LR model (for the Samalas eruption). Probability of occurrence of HMP eruptions based on the timing of dark lunar eclipse dates (b) and tree-ring records^,, (c). d, Integration of b and c to estimate the most probable time windows for UE2, UE4 and the Samalas eruption. Source data

Fig. 4. Revised chronology of explosive volcanism in the twelfth and thirteenth centuries.

Vertical bars are based on the eVolv2k volcanic forcing reconstruction and indicate the magnitude of VSSI. Using total lunar eclipse coloration (red and black dots) and Δ³³S isotope records, we discriminate between stratospheric (red bars) and tropospheric (blue bars) dust veils. Grey bars show uncertain events. Squares, circles and triangles refer to low-latitude, Northern Hemisphere extratopical and Southern Hemisphere extratropical eruptions, respectively. Source data

Extended Data Fig. 1. Testing the four-step procedure using the emblematic 1815 ce Tambora eruption.

A detailed description of the approach can be found in Methods. Source data

Extended Data Fig. 2. Potential dates of HMP (1100–1300 ce) volcanic eruptions (UE1 and UE5) using a four-step procedure.

a, Residence time of volcanic stratospheric aerosols and time windows with SAOD exceeding about 0.1. The residence time of aerosols is based on global mean SAOD₅₅₀ time series from the Sato/GISS and GloSSAC v2 (ref. ) datasets (for the 1883 ce Krakatau and 1991 ce Pinatubo eruptions) and simulated by the EVA forcing generator^, (for UE1 and UE5). Probability of occurrence of HMP eruptions based on the timing of dark lunar eclipse dates (b) and tree-ring records^,, (c). d, Integration of b and c to estimate the most probable time windows for the UE1 and UE5 eruptions. Source data