Most ways of getting OpenStreetMap data hand you the database and wish you luck. The HOT Export Tool, built by the Humanitarian OpenStreetMap Team for its own operations, hands you a deliverable: a defined area, in the formats your software eats, containing the features your work needs, packaged as files a field team can use the hour they land. The difference is the difference between a quarry and a cut stone. Where the generic download routes, the Export tab, Geofabrik's extracts, Overpass queries, take the database home whole or in regional slabs, the Export Tool performs the humanitarian workflow's specific alchemy: clip to an area of interest, filter to the themes that matter, convert to the formats the field actually runs on, and zip the result. This guide teaches the system in its own order: the area, the formats, the themes and the YAML configurations beneath them, the run itself, and the documented ceiling that planners must respect. The generic doors are owned by the download guide linked below; this tool is the filtered, formatted, field-grade one.
Home / GIS and Open Map Data for Beginners
HOT Export Tool: Filtered, Formatted OpenStreetMap Extracts for Field Work

Photo: Anna Shvets / Pexels
Ship a Deliverable: The Export Checklist
Check off each step as you finish it. Your progress stays on this page.
0 of 7 complete
0% done
The Area: Four Ways to Say Where
An export begins with an area of interest, and the tool accepts it in the four forms field work actually produces. You can draw a box on the map for the fast rectangular answer. You can draw a polygon when the area has a shape, a district, a watershed, a camp's catchment. You can upload a GeoJSON boundary when the area was defined elsewhere, by a GIS team or an official source, and must be honored exactly. Or you can type a bounding-box string when the coordinates arrive in an email and precision matters more than drawing. The choice is not cosmetic: the area is the export's first filter, and everything outside it is excluded from every format and theme downstream, which is what keeps field files lean. The professional habit is to define the area from the operation's real geography rather than from a convenient rectangle, and to remember that the area also drives the ceiling discussed below: a generous polygon over a dense city can hold more data than the tool will export, and discovering that at the run is worse than discovering it at the drawing.
The Formats: Five Doors Out
The tool's format list is a field kit in miniature, each format a different consumer's native tongue. OSM PBF is the raw extract, the full-fidelity slice for tools that speak OpenStreetMap natively. Shapefile is the universal donor format, the family of files every GIS and half the world's spreadsheets-adjacent tools can open, with the format's known limits riding along. GeoPackage is the modern single-file container, the connoisseur's choice for QGIS and mobile GIS. KML speaks to Google Earth and the lightweight viewers that field managers actually open. MBTiles carries a rendered tile package for offline map apps, the format our MBTiles guide teaches in full. An export can request several formats at once, and the result arrives as per-format ZIP files, one package per format, each containing the area's data in that shape. The selection discipline is audience discipline: choose the formats your consumers run, not every format the tool offers, because each additional door is a file somebody must download over the connection the field actually has.

The Themes: What the Export Contains
Inside the area and formats sits the export's intelligence: which features. The tool offers curated data themes, humanitarian-shaped selections of the database: the roads and buildings a logistics planner needs, the health facilities and water points a WASH team needs, the full base map for general work. Themes exist because the whole database is usually the wrong answer: a field team needs the features of its mission, in files sized for its bandwidth, without the region's every bench and bus stop. Choosing a theme is therefore a requirements decision disguised as a dropdown, and the project conversation, what will you actually do with this data, belongs before the export, not after the download disappoints. For many operations a theme is exactly right, maintained by HOT to match how its responders work, and the tool's theme list is a small ethnography of humanitarian data needs, worth reading even by users who will end up configuring their own.
The YAML Configuration: Themes You Write Yourself
Where no theme fits, the tool accepts a YAML configuration file that defines the selection in the tool's own grammar: which feature types to include, selected by their OpenStreetMap tags, and which attributes to carry as columns in the tabular formats. This is the export system's power tool, and it rewards the tag literacy this site teaches elsewhere: a configuration is, in effect, a standing query written in YAML, saying take these keys and values, give me these fields, for this area, in these formats. A well-kept configuration is an operational asset: the same export, refreshed monthly, identical in structure, so a program's data pipeline can depend on it. The discipline mirrors the theme discipline at a finer grain: include the attributes your analysis joins on, exclude the ones that only add weight, and test the configuration on a small area before trusting it on the operation's whole geography. HOT's documentation and training materials teach the configuration format's details; this guide's claim is the architectural one, that the YAML file is where an export stops being a download and becomes a specification.
The Run, and the Ceiling
Creating the export is deliberately plain: name it, describe it if the record matters, Create Export, and the tool processes the job and offers the per-format downloads when ready. Plainness ends at the ceiling, which HOT's own training material states as a number and this guide repeats exactly: an export area may contain at most 10,000,000 OpenStreetMap nodes. The ceiling is not a file-size limit but a data-volume one, and it bites in dense urban areas at polygon sizes that look modest on the map. The planner's responses are the standard ones: shrink the area and run neighboring exports, narrow the theme or configuration so the counted data is the data you need, or graduate to the generic heavy machinery, Geofabrik extracts and local tooling, when the job is genuinely a database job. Knowing the ceiling before the deployment meeting, rather than during it, is the difference between a data plan and a data hope, and it is why this guide prints the number in the body rather than a footnote.
Where the Tool Sits
Close with the map of the territory. The generic doors take the database home: the Export tab for a quick slice, Geofabrik for regional extracts, Overpass for query-shaped data, all owned by the download guide. The Tasking Manager puts data in, square by validated square. The Export Tool is the third point of the triangle: data out, but shaped, filtered to themes or to your own YAML specification, converted to the field's formats, and clipped to the operation's true area. Humanitarian work made that shape necessary, but nothing about it is exclusive to emergencies: any project that repeatedly needs the same slice of OpenStreetMap, in the same formats, with the same features, is describing this tool. Learn its order, area, formats, themes, run, and you hold the deliverable end of the OpenStreetMap pipeline.
Repeat Exports: Freshness as an Operational Habit
An export is a snapshot, and operations that depend on one inherit a freshness question the tool makes easy to answer well. Because an export is defined by its area, formats, and theme or YAML configuration, re-running the same specification produces the same-shaped dataset with newer content, and the craft is in treating the specification as the durable asset: save the configuration file, record the area's definition, name exports so their date and scope are legible in the filename and the export's own description, and keep the previous run until the new one is verified, because a field pipeline that swaps snapshots atomically never discovers a bad export by depending on it. The rhythm that emerges in mature programs is calendrical: the export refreshes on the operation's cycle, weekly in a response, quarterly in a development program, and each refresh is also a quiet audit, since a feature count that moves strangely between runs is the data telling you something about the mapping, or about your configuration, before it tells your users. Snapshots age; specifications endure. That is the Export Tool's final lesson, and it applies to every extract workflow this site teaches.
Reading the Result: A First-Export Checklist
A finished run deserves the same verification discipline as any data delivery, and five checks catch nearly everything. Open every format once: the GeoPackage in QGIS, the shapefile in a viewer, the KML in Google Earth, because each consumer reads with different strictness and a file can be valid in one and broken in another. Count features against expectation: the theme promised buildings and roads; if the building count is zero, the area or the theme is wrong, not the city. Check the columns: the tabular formats should carry the attributes your configuration named, spelled as the receiving software expects. Verify the clip: features should end at your area's edge, cleanly, with nothing from the neighboring province riding along. Record the run: date, area definition, theme or configuration version, somewhere the next export's operator will find it, because the repeat-export habit taught above starts with a first export that was written down. Exports verified this way become infrastructure; exports merely downloaded become archaeology.
One audience deserves explicit mention because it justifies the tool's existence in a single image: the field officer on a borrowed laptop, on a connection shared with an entire office, who needs the district's roads and clinics tonight. Every design decision in this guide, the clip, the themes, the per-format zips, is that officer's evening going well instead of badly, and tools should be judged, as this one can be, by the evenings they save.
Related Guides
If this page answered part of the question, these GIS guides cover the rest of the same ground:
- OpenStreetMap Download: The Generic Doors: The unfiltered routes this tool is the shaped alternative to.
- HOT Tasking Manager: Coordinated Mapping, Square by Square: The pipeline's other end: data in, validated.
- OpenStreetMap MBTiles Download: One SQLite File Holding a Whole Map: One of this tool's output formats, taught as its own system.
- The Humanitarian OpenStreetMap Team: The Institution: The organization whose operations this tool was built to serve.
Frequently Asked Questions
What does the HOT Export Tool do that a normal download does not?
It shapes the data: you define an area, choose formats (OSM PBF, shapefile, GeoPackage, KML, MBTiles), and choose data themes or a custom YAML tag configuration, and it returns per-format ZIP files of just that selection, ready for field use.
How do I define the export area?
Four ways: draw a box, draw a polygon, upload a GeoJSON boundary, or type a bounding-box coordinate string. The area is the first filter; everything outside it is excluded from all formats.
What is the YAML configuration for?
Writing your own theme: a file that specifies which OpenStreetMap features, selected by tags, and which attributes as columns, the export should contain. A saved configuration makes an export repeatable as a standing specification.
Is there a size limit?
Yes, a data-volume ceiling documented in HOT's training material: an export area may contain at most 10,000,000 OpenStreetMap nodes. Dense urban polygons reach it sooner than their map size suggests; shrink the area or narrow the selection when you approach it.
Which format should I choose?
The one your consumer runs: GeoPackage for modern GIS, shapefile for universal compatibility, OSM PBF for OSM-native tools, KML for Google Earth, MBTiles for offline map apps. Requesting several formats at once is supported and returns one ZIP per format.
Photo credits: Anna Shvets (Pexels); Anna Shvets (Pexels). Photos are served from this site; follow the credit to the library page for the original.