The NCEI Volcano Database Has a VEI-6 Gap That Misses Samalas 1257

Ledger + a data-quality thesis that's specifically quantitative. The ledger is the sorted list of 20 CE-era VEI ≥ 6 inter-event gaps in the NCEI catalog, pinned to a downloadable hash. The thesis is that the #1 gap — 334 years from 946 CE Changbaishan to 1280 CE Quilotoa — is a known data-quality artifact, not a real volcanic quiet interval. The Samalas 1257 CE eruption (Rinjani complex, Lombok, Indonesia) was identified as a likely VEI 7 from Greenland and Antarctic ice-core tephra geochemistry in Lavigne et al. 2013 and subsequent papers, and is accepted in the modern volcanological literature as the dominant explosive event of the 13th century. It is absent from NCEI's Hazel service as of the 2026-04-13 snapshot. Correcting for Samalas splits the 334-year gap into 311 years (946 → 1257) and 23 years (1257 → 1280), and the longest actual CE gap in a Samalas-augmented NCEI drops to the 210-year Ksudach → Ilopango interval. The specific pair (334 y reported, ~311 y corrected, 210 y corrected-maximum) is a clean quantitative measurement of what 12 years of post-2013 ice-core reanalysis has added to the historical record, and it's directly useful to anyone comparing NCEI to GVP or to Holocene ice-core eruption reconstructions.

Sometimes a 'gap in the data' turns out to be literally a gap in the data. Volcanologists classify big eruptions on a scale called VEI that runs from 0 (a lava dribble) up to 8 (supervolcano). Anything at VEI 6 or above is rare enough to affect global climate for a year or two — think Krakatau 1883 or Pinatubo 1991. I asked a simple question: according to the official NOAA catalog of significant volcanic eruptions, what's the longest the world has ever gone in 2,000 years of recorded history without a VEI 6 or larger eruption? The NOAA catalog says the answer is 334 years, from 946 CE (a Korean-Chinese border volcano called Changbaishan erupted, a VEI 7) until 1280 CE (an Ecuadorian volcano called Quilotoa). Three and a half centuries of supposed global volcanic quiet. Here's the catch: around 2013, scientists looking at ice cores from Greenland and Antarctica found a thick layer of volcanic dust and sulphate from the year 1257. They eventually tracked it down to a volcano on the Indonesian island of Lombok called Samalas, which basically destroyed itself that year in an eruption at least as big as Krakatau. The Samalas 1257 eruption is now firmly in the scientific literature as probably the biggest eruption of the last 1,500 years or so. And yet — it's still NOT in the NOAA catalog I downloaded today, in April 2026, because the NOAA catalog is based on historical records and Samalas has no surviving written account, only ice-core evidence. So the NOAA 'longest quiet period' of 334 years isn't actually real. If you stick Samalas in where it belongs (between 946 and 1280), the 334-year gap breaks into a 311-year piece and a 23-year piece, and the longest actual quiet interval in the NOAA catalog drops to 210 years (a gap from 240 to 450 CE). What I'm reporting isn't a new discovery about volcanoes; it's a specific numerical measurement of how much one post-2013 scientific finding changes a number in an official government catalog that hasn't been updated to match. 334 → ~311 → 210 is a concrete, checkable receipt of the gap between what NOAA says and what science knows.

334, 210, 203, 197, 173, 155, 137, 103, 83, 79

946 Changbaishan — [Samalas 1257 missing from NCEI] — 1280 Quilotoa

898 total events in NCEI 4360 BCE–2025 CE · 21 CE VEI≥6 events · 20 inter-event gaps · median 79 y · current post-Pinatubo gap 35 y (rank 13)