Collections

Collections are named groups of datasets. Unlike a saved search (which is a query that re-runs against the catalog), a collection has explicit membership: you decide what’s in it. Collections have their own list, reached from the top navigation bar, and each one gets a landing page that lists its member datasets, with a box that filters that list by name.

What collections are

A collection is a small piece of metadata pointing at a list of dataset IDs. There’s no copy of the data: adding a dataset to a collection doesn’t duplicate it; removing a dataset from a collection doesn’t delete the dataset. The collection is purely a curation surface.

Typical use cases:

Project deliverables: group all the datasets you produced for a specific project so a stakeholder can browse them in one place.
Thematic catalogs: “Population & demographics”, “Hydrology”, “Land cover 2024” (small curated sub-catalogs organized by theme).
Quarterly reviews: collect the datasets in scope for a specific review and share the URL with reviewers.
Onboarding: give a new team member a “Start here” collection of the most-used datasets.

A dataset can belong to any number of collections; collections can overlap. Collections themselves can’t be nested (no collections-of- collections), but tags work as a coarser grouping above the collection level if you need a hierarchy.

Browsing collections

The Collections link in the top navigation bar opens the collection list: every collection visible to you, sorted by recently-updated.

Each collection card shows:

Collection name and description
Dataset count (e.g., “12 datasets”), plus an Empty badge when the collection has no members
The members’ combined temporal range, when they have one
The created date, and a preview of the collection’s combined spatial extent

Click a card to open the collection’s detail page.

Collection detail

The collection detail page lists every member dataset you can see, 20 per page, under a Filter datasets by name box (each row links to the dataset, with its geometry type and feature count).

That box narrows the rows on screen by title. It filters the page you’re looking at rather than searching the whole collection, so on a collection with more than 20 members it won’t reach the pages you haven’t loaded. To search across a collection’s contents with the full filter rail, use the Collection filter on the catalog page instead.

Creating a collection

Creating a collection requires the editor or administrator role. Viewers can read collections but cannot create them. If you don’t see a New collection button on the Collections page, your role cannot create collections on this instance. See User management & RBAC for how roles are assigned.

To create a collection:

Open Collections in the top navigation bar.
Click New collection.
Fill in the create dialog:
- Name (required): short, human-readable name.
- Description: an optional plain-text description.
Click Create. The collection is saved and the dialog closes; the new collection appears in the Collections list. Open it to add member datasets. Collection names must be unique, so a name that’s already taken is rejected.

Who can see a collection

A collection has no visibility setting of its own: there’s no public/private/restricted toggle on a collection. Every collection is listed to anyone who can browse collections. What governs access is the visibility of each member dataset: when a viewer opens a collection, GeoLens filters the member list down to the datasets that viewer is allowed to see (RBAC). A private member dataset is hidden from viewers who lack access, while the collection itself remains listed.

If you need access control around a set of datasets, set it on the datasets themselves. See the Access tab on each dataset’s detail page.

Managing membership

Editing a collection — renaming it, deleting it, adding or removing members — is restricted to the person who created it, plus admins. The editor role lets you create collections; it doesn’t let you modify someone else’s. Collections with no recorded owner (seeded data, or an owner whose account was deleted) can only be modified by an admin.

If you can modify a collection, its detail page gives you:

Edit in the page header: opens a dialog for the name and description.
More actions next to it: holds Delete, which asks you to type the collection name to confirm. Deleting a collection removes the grouping only; the member datasets are not affected.
An X on each member row: removes that dataset from the collection, one per click.
Add Datasets, a panel below the member list: type a name and press Search; the top 10 matching datasets you can access appear, minus the ones already in the collection. Click Add on a row to add that dataset. The UI adds one dataset at a time; the API takes a batch (POST /api/catalog/collections/{id}/datasets/ accepts a list of dataset_ids).

Members are listed in the order they were added; there’s no manual ordering.

A dataset’s Overview tab shows an Appears in card listing the collections it already belongs to, each linking to that collection. Adding a dataset happens from the collection side, not from the dataset.

Saved searches vs collections

It’s worth being explicit about the distinction, because the two features look similar at first glance:

A saved search is a stored query. Its membership changes as the catalog changes: new datasets matching the query auto-appear, deleted datasets disappear. Use it when the criteria are what matter (“every hydrology dataset in California”).
A collection is an explicit list of dataset IDs. Membership only changes when you add or remove a dataset. Use it when the specific datasets are what matter (“the seven datasets we used for the Q3 review”).

If you need both behaviors (a query result that’s also a stable shareable list), start with a saved search, then materialize the result into a collection. The collection captures a snapshot; the saved search keeps tracking the live query.

See Search & Discovery -> Saved searches for the saved-search side.

Collection URLs are stable: https://your-instance/collections/<id>. Anyone who can browse collections can open the URL; the collection page renders the same for everyone, and the editing controls appear only for the collection’s owner and for admins. Which member datasets each viewer sees still depends on the per-dataset visibility (RBAC), as described above.

To watch or mirror a collection programmatically, fetch its member list from the API: GET /api/catalog/collections/{id}/datasets/ returns the member datasets the caller is allowed to see, paginated with skip and limit (default 50, maximum 200).

Tips

Naming: collection names are clamped to one line on the collection card, so 2 to 4 words read best. Names must also be unique across the instance.
Keep the description short. It renders as plain text under the collection title, and the collection card clamps it to two lines, so a sentence or two lands better than a long block.
Lean on descriptive titles. The collection landing page lists each member dataset by title with its geometry type and feature count, so clear, descriptive dataset titles make a collection easy to scan.