inaturalist-clumper 0.1

Group iNaturalist sightings into clumps by space and time

inaturalist-clumper

Group iNaturalist sightings into clumps, represented as JSON.

Given one or more iNaturalist user logins, this CLI fetches every public observation those users have recorded and groups sightings that happened within ~5 km and ~3 hours of each other into "clumps" — useful for reconstructing a single hike, birding session, or tide-pool visit as one record.

See simonw/inaturalist-clumps/blob/main/clumps.json for example output from this tool.

The output JSON file records, for every clump:

• start/end timestamps, duration, centroid, bounding box, span
• a species roll-up
• a location block: the mode of the observations' place_guess plus the most-specific iNaturalist place that every observation in the clump falls inside (with a breadcrumb of ancestor place IDs)
• per-observation: timestamp, latitude/longitude, identified taxon, user-typed place_guess, iNat place_ids, and photo URLs (thumbnail / large / original)

A top-level places dictionary caches the iNat place metadata (name, display_name, admin_level, ancestor_ids) for every ID that appears in a clump's breadcrumb, so the file is self-contained. Incremental runs reuse this cache and only fetch newly-referenced places.

Install

pip install inaturalist-clumper

Or:

uv tool install inaturalist-clumper

Usage

inaturalist-clumper simonw --output clumps.json

You can pass more than one iNaturalist username to gather sightings from multiple users accounts.

Options:

• USERNAME ... — required
  One or more iNaturalist user logins to fetch.
• --output — default: clumps.json
  Output JSON path. Used as the basis for incremental runs if it exists.
• --distance-km — default: 5.0
  Spatial threshold for linking two observations into the same clump.
• --hours — default: 3.0
  Temporal threshold for linking two observations into the same clump.
• --full-refresh — default: off
  Ignore any existing observations in the output file and re-fetch fully.

Incremental runs

A second invocation against the same --output file reads the existing data and asks the iNaturalist API for any observations that have been created or edited since the previous run's generated_at timestamp (using the updated_since parameter). Edited records overwrite the cached copy on merge, so corrected taxa, new photos, or fixed coordinates flow back in. The full set is then re-clumped and the file rewritten — so a new sighting that bridges two previous clumps will merge them.

Use --full-refresh to ignore the existing file and start over.

Development

Clone the repo and then:

uv run pytest

To run the development version:

uv run inaturalist-clumper --help