Everything cave data needs.
A full-featured research platform built specifically for how cavers actually work: in the field, offline, and across organizations.
The map is the center of the platform. Topo quads, karst geology, LiDAR hillshade. Stack as many layers as you need. Overlay your own GPX tracks, draw areas of interest, and view every cave you have access to.
The extraction pipeline reads your uploads and pulls out every cave name, coordinate, depth, length, species observation, and person name it finds. Results are flagged for human review before being linked to site records.
No query syntax to learn. Type what you mean. The platform figures out the rest.
Install KEEP on your phone as a web app. Sites, documents, and photos cache locally so you can pull up the record at the entrance with no signal. Log observations with GPS autofill. Everything syncs when you get back in range.
No forced taxonomy. Pick from a shared list of descriptors other cavers have already used, or type your own and add it with one click. The library grows with how the community actually talks.
Description
A free-text field for narrative notes — history, character, anything that doesn't fit a picker.
Status
Confirmed, unconfirmed, or not-a-cave. Every claim also carries its own confidence level (verified, high, medium, low, unverified).
Entrance & cave descriptors
Two pickers for entrance and cave character. Select from existing terms or add a custom one — your new term becomes available to everyone else.
Access conditions & hazards
Gated, permit required, flooding risk, landowner permission, seasonal closure — plus anything else you need to add.
Privacy tier
Strictly Private (only you), Contributed (others can see the name + region and request access), or Public (visible to all registered users). Set per site.
Every export is scoped to your access level. Sites you can't see aren't in the file. All exports are audit-logged.
| Format | Use case | Extension | Scope |
|---|---|---|---|
| CSV | Spreadsheet analysis, GIS import | .csv | Per-search or full selection |
| KML / KMZ | Google Earth, mapping apps | .kml / .kmz | Mapped sites with geom |
| GPX | GPS devices, navigation apps | .gpx | Waypoints & tracks |
| GeoJSON | Web maps, PostGIS, QGIS | .geojson | All site geometry |
| Compass .dat | Compass survey software | .dat | Survey shots per cave |
| Walls .srv | Walls survey software | .srv | Survey shots per cave |
| Bibliography PDF | Formatted reference list of sources | Documents linked to selection |
When you enter a cave or upload a document, KEEP cross-references it against every uploaded document and every existing site record. Different name, slightly different coordinates? It still surfaces the possible match — then a human decides whether to link.
You add data
Enter a cave name and coordinates, upload a PDF trip report, import a GPX track, or drop in a scanned survey map.
The engine suggests matches
Fuzzy name matching and proximity search surface candidate sites and mentions across uploaded sources. "Bear Hollow Cave" surfaces "Bear Hollow Cavern" as a likely match.
A trustee reviews and links
Matches stay in a review queue until a user with access confirms them. Nothing is auto-merged, and linking a cave doesn't grant you access to the underlying record — that's a separate request.
Nothing gets overwritten. Every claim keeps its source, author, and date.