GeoZarr SWG Monthly Meeting, 2 June 2026

Meeting Purpose

Review and align on proposed conventions for defining array coordinates.

Key Takeaways

• Two coordinate conventions were presented: Christophe's domain-agnostic coords and Patrick's OGC/CF-based CS (Coordinate Set).
• The core debate is between a composable, simple-first approach (coords) and a comprehensive, OGC-aligned superset (CS).
• The group agreed to update the SWG charter to define GeoZarr's scope and design philosophy, which will guide the convention's direction.

Topics

The problem: limited geospatial support

• Current GeoZarr conventions (multiscales, spatial, proj) support a narrow range of geospatial data, primarily affine-transformed GIS/GeoTIFF-style data.
• This leaves complex datasets (e.g., CMIP6, ROMS, EOPF) unaddressed, forcing workarounds like data duplication.
• The two proposed conventions aim to fill this gap but represent fundamentally different philosophies.

Proposed solution 1: coords convention

• Philosophy: Domain-agnostic, composable, and simple-first.
• Goal: Provide foundational building blocks that can be extended with domain-specific conventions (e.g., spatial, proj).
• Descriptor types:
  • array: Standard explicit coordinate array.
  • inline: Embeds small coordinate lists directly in metadata (e.g., RGB bands).
  • interval: Describes regularly spaced coordinates via start, end, and step.
  • reference: Points to another convention (e.g., spatial for affine transforms).
• Auxiliary coordinates: Supports mapping one or more coordinates to one or more dimensions, similar to CF's auxiliary coordinates.

Proposed solution 2: CS convention

• Philosophy: A comprehensive superset of coords and spatial, based on OGC and CF standards.
• Goal: Manage the inherent complexity of geospatial data by adopting established, structured standards.
• Structure:
  • Organizes coordinates by OGC Coordinate Reference System (CRS) type (e.g., spatial, temporal).
  • This structure enables unique CRS identifiers (e.g., EPSG codes) for each component.
• Capabilities:
  • Supports complex grids (curvilinear, parametric vertical coordinates).
  • Allows out-of-group references to avoid data duplication (e.g., in EOPF).
  • Includes CF-style formula definitions for parametric coordinates.

The debate: composability vs. comprehensiveness

• Core question: Should GeoZarr offer multiple conventions for similar tasks, or define one canonical way?
• Argument for composability (coords):
  • Users only need to learn the conventions relevant to their use case.
  • Avoids forcing users with simple data to adopt complex standards.
• Argument for comprehensiveness (CS):
  • The complexity is in the data, not the convention; a robust standard is necessary to manage it.
  • Leverages decades of knowledge from OGC and CF for interoperability.
• Counterpoint: Standards development is about restricting possibilities to ensure interoperability, not enabling all of them.

Next Steps / Action Items

• Max
  • Draft an updated GeoZarr charter defining GeoZarr's scope, design goals, and philosophy on convention composability; post on GitHub for SWG review.
  • Send these minutes to the SWG, including Christophe's Medium write-up on coords and Patrick's CS one-pager.
  • Update the GeoZarr page to reflect Christophe stepping down as co-chair.
• All
  • Discuss the charter update on GitHub to resolve the composability vs. comprehensiveness debate.