Lilambd/world-ground

World Ground Ledger: how many people live near volcanoes, coasts and plate boundaries, and on what soil

🌐 日本語 · 中文 · Español · Português · Français · Deutsch · 한국어 · हिन्दी · العربية

• 日本語 — World Ground Ledger：火山・海岸・断層の近くに住む人口と土壌: 世界の4,182の州・省ごとに、火山・海岸・プレート境界の近くに何人が住み、どんな土壌の上にいるかを集計しました。
• 中文 — World Ground Ledger（全球地表暴露与土壤数据）: 全球 4,182 个州和省中，有多少人居住在火山、海岸和板块边界附近，以及脚下是什么样的土壤。
• Español — World Ground Ledger (población expuesta y suelos por región): Cuántas personas viven cerca de volcanes, costas y límites de placas, y sobre qué suelo, en 4,182 estados y provincias del mundo.
• Português — World Ground Ledger: população perto de vulcões, costas e falhas: Quantas pessoas vivem perto de vulcões, do litoral e de limites de placas tectônicas, e sobre que tipo de solo, em 4,182 estados e províncias do mundo.
• Français — World Ground Ledger : population exposée aux volcans, côtes et failles: Combien de personnes vivent près des volcans, des côtes et des frontières de plaques, et sur quel sol, pour 4,182 États et provinces du monde.
• Deutsch — World Ground Ledger: Bevölkerung an Vulkanen, Küsten und Plattengrenzen: Wie viele Menschen in der Nähe von Vulkanen, Küsten und Plattengrenzen leben und auf welchem Boden, für 4,182 Bundesstaaten und Provinzen der Welt.
• 한국어 — World Ground Ledger: 화산·해안·단층 인근 인구와 토양: 세계 4,182개 주·도별로 화산, 해안, 판 경계 가까이에 사는 인구와 그 땅의 토양을 정리했습니다.
• हिन्दी — World Ground Ledger (भूमि और जोखिम खाता): दुनिया के 4,182 राज्यों और प्रांतों में कितने लोग ज्वालामुखियों, तटों और प्लेट सीमाओं के पास रहते हैं, और किस मिट्टी पर।
• العربية — سجل الأرض والسكان (World Ground Ledger): كم عدد من يعيشون قرب البراكين والسواحل وحدود الصفائح التكتونية، وعلى أي تربة، في 4,182 ولاية ومقاطعة في العالم.

For 4,182 states, provinces and prefectures of the world, computed on a 0.1° land grid and summed per region:

feature	unit	what
pop_volcano50_2020	persons	people living within 50 km of a volcano with a recorded eruption (NOAA NCEI)
pop_coast10_2020	persons	people living within 10 km of the coast
pop_plate100_2020	persons	people living within 100 km of a tectonic plate boundary
coast_km_popw	km	population-weighted mean distance to the coast
plate_km_popw	km	population-weighted mean distance to a plate boundary
built_2025	km²	built-up area in 2025 (GHSL nowcast)
soc_0_5	g/kg	soil organic carbon, 0-5 cm, area mean
ph_0_5	pH	soil pH in water, 0-5 cm, area mean
ocs_0_30	kg/m²	soil organic carbon stock, 0-30 cm
macro_cover	share	share of the region covered by a Macrostrat geologic column (North and South America, Caribbean, New Zealand)

Nothing is generated by a model. Population is GHSL 2020 on each grid cell; distances are to the nearest point.

Data as of 2026-10-04.

Quick start

from datasets import load_dataset
g = load_dataset("Lilambd/world-ground", "admin1_ground", split="train")

admin1_ground is one row per region (wide). features_long is the same in long form (a1, feature, value, year, unit, source).

What the numbers say

• 213 million people live within 50 km of an erupting volcano; the top regions are all on Java: West Java 37.2 M, East Java 18.2 M, Central Java 12.0 M.
• 1.29 billion people (16.6%) live within 10 km of the coast.
• 1.44 billion people (18.5%) live within 100 km of a plate boundary.

Shared keys

a1 is the same admin-1 id as in Lilambd/world-hazard and Lilambd/world-earth, so volcano and coast exposure join with earthquakes, warming and forest loss row by row. Also iso_3166_2, wikidata, adm0_a3, and cell_05deg (0.5° grid of the region's centre, row*720+col).

Gaps and known limits

• 414 of the 4,596 Natural Earth admin-1 regions have no row; the source grid model did not cover them.
• About 70 regions have a blank soil value: SoilGrids gave no value for their cells.
• macro_cover is only meaningful in the Americas and New Zealand; elsewhere it is 0 because Macrostrat has no columns, not because there is no rock.
• Seismic hazard (peak ground acceleration) and mine counts exist in the source model but are not included here: the hazard map is licensed for non-commercial use only, and the mine list has mixed sources. Earthquake counts per region are in world-hazard.
• Plate boundaries follow Bird (2003) PB2002; only computed distances and counts are shared, not the boundary geometry.

Use it in production

For AI agents: 14 more Lilambd open-data APIs (airports, regions, markets, security, travel and more), pay per call with x402 (from $0.005 USDC on Base, no signup), free sample with sample=1, also as an MCP server: https://api.lilambd.com/llms.txt

Need a per-place answer, a field that is not here, or a finer grid? Write to hello@lilambd.com.

Quick start: datasets, pandas, DuckDB

import pandas as pd  # hf:// paths need `pip install huggingface_hub`
g = pd.read_csv("hf://datasets/Lilambd/world-ground/data/admin1_ground.csv",
                keep_default_na=False, na_values=[""])
print(g.nlargest(10, "pop_volcano50_2020")[["country", "name", "pop_volcano50_2020"]])
print(g.groupby("country")["pop_coast10_2020"].sum().sort_values(ascending=False).head(10))

-- DuckDB 0.10+ reads hf:// paths directly
-- people near plate boundaries vs. M7+ counts (joins world-hazard on a1)
SELECT g.country, g.name, g.pop_plate100_2020, q.m6, q.m7
FROM 'hf://datasets/Lilambd/world-ground/data/admin1_ground.csv' g
JOIN 'hf://datasets/Lilambd/world-hazard/data/admin1_quake.csv' q USING (a1)
ORDER BY q.m7 DESC LIMIT 10;

A runnable notebook is in notebooks/quickstart.ipynb.

Use cases

• Exposure screening: people within 50 km of volcanoes, 10 km of coasts or 100 km of plate boundaries.
• Coastal and sea-level risk ranking per region.
• Soil features (organic carbon, pH) per region for agriculture or climate models.
• Combine with world-hazard and world-earth on a1 for multi-hazard indices.

Citation

@misc{lilambd2026_world_ground,
  author       = {Lilambd},
  title        = {Lilambd World Ground Ledger},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/Lilambd/world-ground}},
  url          = {https://huggingface.co/datasets/Lilambd/world-ground}
}

Please also credit the original sources listed in this card.

Sources and licence

• Population and built-up: European Commission JRC, GHSL GHS-POP / GHS-BUILT-S R2023A. CC BY 4.0.
• Volcano locations: NOAA NCEI / WDS Global Volcano database. Public domain.
• Coastline and regions: Natural Earth 10m. Public domain.
• Plate boundaries: Bird, P. (2003) An updated digital model of plate boundaries, G-cubed 4(3), 1027 (PB2002). Distances only.
• Soil: ISRIC SoilGrids 2.0 (Poggio et al. 2021). CC BY 4.0.
• Geologic columns: Macrostrat API v2. CC BY 4.0.

The derived tables are released under CC BY 4.0. Please cite "Lilambd World Ground Ledger" and the original sources.

Maintained by Lilambd. Questions: hello@lilambd.com

Changelog

• 2026-11-03 quality fixes.
  • The person counts pop_volcano50_2020, pop_coast10_2020 and pop_plate100_2020 are now whole numbers, rounded to the nearest person from the fractional GHSL gridded estimates. 2,192 of 4,182 rows in admin1_ground had a fractional value: 1,684 in coast, 912 in plate and 309 in volcano. In features_long, 2,905 rows had one. Totals moved by less than 15 people per column. Distance, area, soil and share columns are means or densities and were not changed.
  • iso_3166_2: 138 Natural Earth placeholder codes (…~, e.g. AW-X01~) were cleared to empty. 3 of them were filled with the correct code (Gbapolu LR-GP, River Gee LR-RG, Tadzhikistan Territories TJ-RA). The codes now match Lilambd/world-hazard admin1 row for row.

Paid editions & related

• Five famous mountains, 150 mm 3D-printable terrain ($15): https://payhip.com/b/nvLdh
• Terrain lithophane 6-pack ($14): https://payhip.com/b/3onpq
• Real Mountains, Layered: 5 laser-cut models ($15): https://payhip.com/b/5PMsa
• Same data as an API (free sample=1, pay-per-call via x402): https://api.lilambd.com/llms.txt
• All Lilambd shops: https://payhip.com/Lilambd and https://lilambdcrossing.itch.io

If this dataset helped you, a heart (like) on this page helps others find it.