What the pack contains

File	One row =	Volume	See an extract
haiti_570_sections_communales_gps.csv	A communal section (admin 3)	570 rows · 13 columns	10 real rows
haiti_140_villes_chefs_lieux_gps.csv	A chief town, with its status	140 rows · 13 columns	10 real rows
haiti_lieux_habites_geonames_gps.csv	A populated place, attached to its communal section	10,945 rows · 14 columns	10 real rows
hti_admin1.geojson	A department (polygon)	10 entities	sheet for one entity
hti_admin2.geojson	A commune (polygon)	140 entities	sheet for one entity
hti_admin3.geojson	A communal section (polygon)	570 entities	sheet for one entity

Real preview — communal sections

Exact extract of the delivered file, columns included. The full sample (10 rows) downloads without an account.

departement_ht	commune_ht	section_communale_fr	section_communale_ht	pcode_section	superficie_km2	latitude	longitude
Lwès	Pòtoprens	1re Section Turgeau	1ye Seksyon Turgeau	HT0111-01	18.35	18.536790	-72.322631
Lwès	Pòtoprens	3e Section Martissant	3yèm Seksyon Martissant	HT0111-03	8.95	18.518016	-72.359607
Lwès	Dèlma	1re Section St Martin	1ye Seksyon St Martin	HT0112-01	26.04	18.551396	-72.287185
Lwès	Leyogàn	11e Section Gros Morne	11yèm Seksyon Gro Mòn	HT0121-11	28.37	18.433665	-72.590076

Download the sample (CSV, 10 rows)

What it is for

Power BI choropleth maps — join your indicators to the polygons through pcode_section, with no manual geocoding.
NGO territorial targeting — cross your beneficiary databases with the official p-codes used by humanitarian partners.
Survey geocoding — attach field records to an identified communal section.
Distance calculations — sales points, schools, health centres: town ↔ section distances from the coordinates.
Dissertations and research — a stable, quotable administrative division, with its areas.
From a locality up to its section — the 10,945 populated places carry their section's p-code: a village name in your data becomes an administrative zone.
Maps and reports in kreyòl — the GeoJSON carry the kreyòl names in their properties: labels come out directly, with no join.

Sources and method

Administrative boundaries, centroids, areas, p-codes and names — CNIGS, distributed by UN OCHA (COD-AB Haiti), published 26 January 2026, CC BY 3.0 IGO licence.
Populated places — GeoNames, CC BY 4.0 licence.
Atmart's work — recovering the kreyòl names of the communes from the chief-town layer, kreyòl normalisation of the sections, attaching the 10,945 populated places to their section by point-in-polygon computation, harmonising accented spellings, producing the CSV and the labelled GeoJSON. July 2026 rebuild, fully automated and reproducible.

The kreyòl names, and where they come from

The CNIGS provides a kreyòl field but does not fill it everywhere. Version 2 goes to fetch it where it really exists, and says clearly what comes from the source and what comes from us.

Level	Kreyòl in the pack	Provenance
10 departments	10 / 10	CNIGS — official names
140 communes	140 / 140	CNIGS — recovered from the chief-town layer
140 chief-town cities	140 / 140	CNIGS — official names
570 communal sections	570 / 570	Documented Atmart normalisation, CNIGS toponym preserved

The rule applied to the sections is deterministic: the ordinal moves into kreyòl (1re → 1ye, 10e → 10yèm), “Section” becomes “Seksyon”, the French preposition is dropped, and the toponym stays exactly as the CNIGS has it — except where the CNIGS itself gives an official kreyòl form (Gros Morne → Gro Mòn, Pointe à Raquette → Pwent a Rakèt). A section_ht_provenance column carries the information row by row. The official French name stays in the file: it is the one of record.

Known limits

Population of populated places. GeoNames gives a population for only 114 places out of 10,945 (about 1 %). This file is for locating, not for sizing.

Administrative attachment of populated places. 10,808 places (98.7 %) fall inside their section's polygon. The other 137 (1.3 %) are attached to the nearest section, because GeoNames often rounds its coordinates to 1/60^th of a degree. The rattachement column says so row by row — filter it if your use requires strict location.

Accents. The CNIGS writes part of the names without accents (104 communes out of 140, 501 sections out of 570). We keep its spelling: correcting it would mean departing from the source of record.

Administrative divisions. CNIGS vintage in force on 29 November 2018, published by OCHA on 26 January 2026. It says nothing about communes created or merged afterwards.

Data sheet (Fèy Done)

Version	v2.2026.07 — entirely rebuilt from the CNIGS
Next update	at the CNIGS's next COD-AB release; included at no extra cost for buyers of the year
Geographic coverage	Haiti — 10 departments, 140 communes, 570 communal sections, 10,945 populated places
Formats	3 CSV (UTF-8 with BOM) + 3 GeoJSON (WGS84) + LISEZMOI + Fèy Done in 4 languages
Coordinate system	WGS84 (EPSG:4326)
Join key	OCHA p-codes: `pcode_departement`, `pcode_commune`, `pcode_section` — present in all three CSV and all three GeoJSON
Confidence level	A for the reference layer and the French names; B for the kreyòl names of the sections (Atmart normalisation) — see the scale
Personal data	None
Licence	Single-organisation use, resale and redistribution forbidden; attribution “Pack Géo Haïti — Atmart / CNIGS-OCHA / GeoNames”
Contact	sales@atmart.ltd

Citation

Atmart LLC (2026). Pack Géo Haïti — communal sections, towns and populated places, v2.2026.07 [dataset]. Atmart Data. Primary sources: CNIGS, distributed by UN OCHA (COD-AB Haiti); GeoNames.

Haiti Geo Pack — every administrative division with GPS coordinates