Datasets:
The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: CastError
Message: Couldn't cast
FID: string
Volcano_Number: int64
Volcano_Name: string
Eruption_Number: int64
Activity_Type: string
ExplosivityIndexMax: double
ExplosivityIndexModifier: string
ActivityArea: string
ActivityUnit: string
StartEvidenceMethod: string
StartDateYearModifier: string
StartDateYear: int64
StartDateYearUncertainty: double
StartDateDayModifier: string
StartDateMonth: double
StartDateDay: double
StartDateDayUncertainty: double
EndDateYearModifier: string
EndDateYear: double
EndDateYearUncertainty: double
EndDateDayModifier: string
EndDateMonth: double
EndDateDay: double
EndDateDayUncertainty: double
GeoLocation: string
has_volcano_metadata: bool
Country: string
Region: string
Subregion: string
Primary_Volcano_Type: string
Tectonic_Setting: string
Latitude: double
Longitude: double
Elevation: double
Major_Rock_Type: string
pre_holocene_outlier: bool
start_date_precision: string
-- schema metadata --
pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 5081
to
{'FID': Value('string'), 'Volcano_Number': Value('int64'), 'Volcano_Name': Value('string'), 'Eruption_Number': Value('int64'), 'Activity_Type': Value('string'), 'ExplosivityIndexMax': Value('float64'), 'ExplosivityIndexModifier': Value('string'), 'ActivityArea': Value('string'), 'ActivityUnit': Value('string'), 'StartEvidenceMethod': Value('string')}
because column names don't match
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
return get_rows(
dataset=dataset,
...<4 lines>...
column_names=column_names,
)
File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
return func(*args, **kwargs)
File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
yield from ds.decode(False) if ds.features else ds
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2406, in _iter_arrow
pa_table = cast_table_to_features(pa_table, self.features)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2280, in cast_table_to_features
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
FID: string
Volcano_Number: int64
Volcano_Name: string
Eruption_Number: int64
Activity_Type: string
ExplosivityIndexMax: double
ExplosivityIndexModifier: string
ActivityArea: string
ActivityUnit: string
StartEvidenceMethod: string
StartDateYearModifier: string
StartDateYear: int64
StartDateYearUncertainty: double
StartDateDayModifier: string
StartDateMonth: double
StartDateDay: double
StartDateDayUncertainty: double
EndDateYearModifier: string
EndDateYear: double
EndDateYearUncertainty: double
EndDateDayModifier: string
EndDateMonth: double
EndDateDay: double
EndDateDayUncertainty: double
GeoLocation: string
has_volcano_metadata: bool
Country: string
Region: string
Subregion: string
Primary_Volcano_Type: string
Tectonic_Setting: string
Latitude: double
Longitude: double
Elevation: double
Major_Rock_Type: string
pre_holocene_outlier: bool
start_date_precision: string
-- schema metadata --
pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 5081
to
{'FID': Value('string'), 'Volcano_Number': Value('int64'), 'Volcano_Name': Value('string'), 'Eruption_Number': Value('int64'), 'Activity_Type': Value('string'), 'ExplosivityIndexMax': Value('float64'), 'ExplosivityIndexModifier': Value('string'), 'ActivityArea': Value('string'), 'ActivityUnit': Value('string'), 'StartEvidenceMethod': Value('string')}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Global Volcanic Eruptions: Holocene Record
11,089 documented volcanic eruptions from 1,196 volcanoes worldwide, spanning the Holocene period (~12,000 years), sourced directly from the Smithsonian Institution's Global Volcanism Program (GVP) via their official GeoServer web service.
One row per eruption event, joined to per-volcano location and geological metadata. Every ambiguous or edge-case record is flagged rather than dropped, so the dataset preserves the real shape of the historical record instead of a cleaned-up illusion of it.
TL;DR
- 11,089 eruptions across 1,196 volcanoes, Holocene scope (plus one flagged pre-Holocene outlier).
- Two joined GVP tables (eruptions plus volcanoes) on
Volcano_Number; zero duplicate IDs, zero full-duplicate rows. - Uncertain eruptions, missing VEI, missing volcano metadata and imprecise dates are kept and flagged, not silently removed.
- Not MIT. GVP data are U.S. Government works under the Smithsonian Terms of Use: free for personal, educational and non-commercial use with attribution. See Licensing.
Dataset structure
A single CSV, global_volcanic_eruptions_holocene.csv (~4.3 MB), one row per eruption, 37 columns. The most important columns:
| Column | Type | Description |
|---|---|---|
FID |
str | Internal GVP feature ID from the source web service. Not meaningful for analysis — safe to drop. |
Volcano_Number |
int | GVP unique volcano identifier. The join key between the eruption and volcano tables. |
Volcano_Name |
str | Name of the volcano. |
Eruption_Number |
int | GVP unique eruption identifier. |
Activity_Type |
str | Confirmed Eruption or Uncertain Eruption. |
ExplosivityIndexMax |
float | Volcanic Explosivity Index (VEI), 0–8 scale. Null for 24.3% of eruptions (see caveats). |
ExplosivityIndexModifier |
str | Qualifier on the VEI value (mostly null; populated only when GVP has a specific caveat). |
ActivityArea |
str | Named vent, crater or sub-feature within the volcano (often null). |
ActivityUnit |
str | More granular sub-unit of ActivityArea, e.g. a tephra layer (91.8% null). |
StartEvidenceMethod |
str | Dating method for the eruption's start (historical observation, radiocarbon, tephrochronology, …). |
Additional columns in the file cover eruption start/end dates, geographic coordinates and per-volcano geological metadata, plus the cleaning flags described below (has_volcano_metadata, pre_holocene_outlier, start_date_precision).
How it was built
Two source tables from the GVP GeoServer WFS service a Holocene Eruptions layer (one row per eruption event) and a Holocene Volcanoes layer (location/geological metadata per volcano) joined on Volcano_Number. No fuzzy matching: the join is on GVP's own integer keys and both eruption and volcano IDs are unique with no full-duplicate rows.
Data cleaning & decisions
Every judgement call is surfaced and edge cases are flagged, not deleted.
Uncertain eruptions are retained, not filtered. 9,918 Confirmed Eruption plus 1,171 Uncertain Eruption. "Uncertain" reflects unverifiable historical evidence, not fabrication; dropping it would discard 10.6% of real data. Filter Activity_Type yourself if you want confirmed-only.
Missing volcano metadata is flagged. 266 eruptions (2.4%) have no matching volcano record, likely subsidiary vents or synonyms not separately catalogued. Kept and flagged via has_volcano_metadata rather than dropped.
The one pre-Holocene outlier is flagged, not removed. A single eruption (~55,500 years ago, Golden Trout Creek, dated via cosmic-ray exposure) predates the nominal Holocene window. It's a real, sourced data point, flagged via pre_holocene_outlier.
Missing VEI is preserved as null. ExplosivityIndexMax is null for 24.3% of eruptions — VEI can't always be determined for older or less-observed events. Kept as NaN rather than dropped or imputed, to avoid biasing the set toward recent, well-studied eruptions.
Date precision varies and is explicitly flagged — read this before any fine-grained temporal analysis. Many historical eruptions are known only to the year or month, not the day and the source uses placeholder dates rather than blanks:
- July 2nd appears to serve as a year-only-precision placeholder (confirmed via evidence method, 95% of these cite vague historical "Observations: Reported").
- The 16th of the month appears to serve as a month-only-precision placeholder (confirmed by its even spread across all 12 months, rather than genuine clustering).
A start_date_precision column flags these as year (probable placeholder), month (probable placeholder) or day (assumed exact). This cannot be determined with 100% certainty per row, so treat day/month-level precision with caution for fine-grained work.
Negative years mean BCE. e.g. -8300 = 8300 BCE, not an error.
Validation checks passed. VEI values within the valid 0–8 scale; coordinates within valid lat/lon ranges; no eruption logically ends before it starts.
Example usage
import pandas as pd
from datasets import load_dataset
ds = load_dataset("xquantize/global-volcanic-eruptions-holocene", split="train")
df = ds.to_pandas()
# Confirmed eruptions with a known VEI, day-level date precision
clean = df[
(df["Activity_Type"] == "Confirmed Eruption")
& df["ExplosivityIndexMax"].notna()
& (df["start_date_precision"] == "day")
]
print(len(clean), "high-confidence eruptions")
Not intended for: commercial redistribution without observing the Smithsonian terms; fine-grained daily-resolution analysis without honoring start_date_precision; or treating missing VEI as zero.
Provenance
Smithsonian Institution, Global Volcanism Program, Volcanoes of the World (VOTW) database.
Queried directly via the GVP GeoServer WFS web service, pulling the Holocene Eruptions and Holocene Volcanoes layers, then joined on Volcano_Number. Snapshot corresponds to VOTW v. 5.3.6 (2025).
Licensing
License: other Smithsonian Terms of Use for GVP Content.
GVP data are produced by employees of the U.S. Government and are provided under the Smithsonian Terms of Use, which invite use for personal, educational and other non-commercial purposes with attribution. This is not an MIT/BSD software license and not a blanket public-domain dedication, please observe the Smithsonian terms and cite GVP as below. If your use is commercial, review the Smithsonian Terms of Use and contact GVP as needed.
- Downloads last month
- 19