I.  Abstract

Zarr provides efficient chunked storage for n-dimensional arrays but do not provide with the semantic constructs required for geospatial and scientific data workflows.

GeoZarr defines an abstract data model and a set of conventions for representing geospatial and scientific datasets in the Zarr format:

By providing a standardized framework for geospatial semantics, GeoZarr enables scientific and geospatial applications to fully utilize cloud-native storage architectures while maintaining the rich metadata and coordinate referencing required for Earth observation workflows. The result is a modern, scalable approach to storing and accessing geospatial data that meets the needs of both data providers and consumers.

III.  Preface

The GeoZarr Standard defines a layered, standards-based framework for representing and encoding geospatial and scientific datasets in the Zarr format. The purpose of this document is to provide implementation guidance and normative structure for consistent, interoperable adoption of GeoZarr across tools, platforms, and services. This work extends prior standardisation efforts within the OGC, including OGC API – Tiles, the Tile Matrix Set Standard, and EO metadata conventions, and anticipates integration with catalogue systems such as STAC.

This Standard has been developed in collaboration with contributors from Earth observation, climate science, geospatial analysis, and cloud-native geodata infrastructure communities. Future work may extend this model to additional storage formats, API services, and semantic layers.

1.  Scope

The GeoZarr Standard defines a conceptual and implementation framework for representing and encoding geospatial and scientific datasets using the Zarr format. The scope of this Standard includes the definition of a format-agnostic data model, the specification of its encoding into Zarr Version 2 and Version 3, and a set of extensions to support affine transformations and overviews.

These capabilities are necessary for geospatial data because Zarr does not provide semantic constructs for geospatial data interpretation. Applications need to understand not just array shapes and values, but coordinate meanings, projection parameters, and scientific metadata. GeoZarr fills this gap without compromising Zarr’s performance characteristics.

2.  Conformance

The GeoZarr Unified Data Model is structured around a modular set of requirements classes. These classes define the conformance criteria for datasets and implementations adopting the GeoZarr specification. Each class provides a distinct set of structural or semantic expectations, facilitating interoperability across a broad spectrum of geospatial and scientific use cases.

The Core requirements class defines the minimal compliance necessary to claim conformance with the GeoZarr Unified Data Model. It is intentionally open and permissive, supporting incremental adoption and broad compatibility with existing Zarr tools and data models based on the Unidata Common Data Model (CDM).

Additional requirements classes are defined to support enhanced functionality, semantic richness, and interoperability with established geospatial conventions and systems. These include extensions for time series, coordinate systems, affine transformations, and multiscale tiling.

Table 1 — Requirements Classes Overview

Each requirements class is independently defined. Implementations may declare conformance with any subset of classes appropriate to their use case. All classes build upon the Core model.

Associated conformance tests for each class are detailed in Annex A.

3.  Normative references

The following documents are referred to in the text in such a way that some or all of their content constitutes requirements of this document. For dated references, only the edition cited applies. For undated references, the latest edition of the referenced document (including any amendments) applies.

Miles, A., et al.: Zarr Specification Version 2. Zarr Developers. https://zarr.readthedocs.io/en/stable/spec/v2.html

Zarr Community: Zarr Specification Version 3. https://zarr-specs.readthedocs.io/en/latest/v3.0

Unidata: The Common Data Model. https://docs.unidata.ucar.edu/netcdf-java/5.0/userguide/common_data_model_overview.html

Rew, R., Davis, G.: NetCDF: An Interface for Scientific Data Access. IEEE Computer Graphics and Applications, 10(4), 76–82 (1990). https://doi.org/10.1109/38.56302

CF Community: Climate and Forecast (CF) Metadata Conventions, Version 1.10. https://cfconventions.org/

GDAL Developers: GDAL/OGR Version 3.8 Documentation. Open Source Geospatial Foundation. https://gdal.org

Open Geospatial Consortium: OGC Two Dimensional Tile Matrix Set and Tile Pyramid (OGC 17-083r2). https://docs.ogc.org/is/17-083r2/17-083r2.html

STAC Community: STAC Specification v1.0.0. https://stacspec.org/en/

Open Geospatial Consortium: OGC Compliance Testing Policies and Procedures, OGC 08-134r10. https://portal.ogc.org/files/?artifact_id=55184

4.  Terms, definitions and abbreviated terms

This document uses the terms defined in OGC Policy Directive 49, which is based on the ISO/IEC Directives, Part 2, Rules for the structure and drafting of International Standards. In particular, the word “shall” (not “must”) is the verb form used to indicate a requirement to be strictly followed to conform to this document and OGC documents do not use the equivalent phrases in the ISO/IEC Directives, Part 2.

This document also uses terms defined in the OGC Standard for Modular specifications (OGC 08-131r3), also known as the ‘ModSpec’. The definitions of terms such as standard, specification, requirement, and conformance test are provided in the ModSpec.

For the purposes of this document, the following additional terms and definitions apply.

5.  Conventions

This section describes the conventions used throughout this Standard, including identifiers, metadata schemas, and referencing mechanisms relevant to the GeoZarr Unified Data Model.

6.  Overview

The GeoZarr Standard defines an abstract data model and a set of conventions for representing and describing geospatial and scientific datasets using the Zarr format.

Zarr provides efficient, chunked storage for n-dimensional arrays but does not include the semantic constructs required for geospatial and scientific data workflows. The Unidata Common Data Model (CDM) addresses this gap by introducing essential concepts that structure information through variables, groups, coordinates, and metadata. This abstract data model provides the semantic framework that enables structured interpretation of array-based data on top of Zarr’s storage foundation.

The primary objective of GeoZarr is to specify how the CDM is encoded within Zarr. GeoZarr provides normative rules for encoding these CDM concepts in Zarr and thereby standardises the encoding practices already adopted by CDM-compatible libraries such as xarray and nczarr, promoting consistent interpretation and interoperability across tools and platforms.

By defining an abstract model based on the CDM and a corresponding encoding for Zarr, GeoZarr establishes an explicit relationship between the conceptual structure of the data and its physical storage representation. Zarr defines how data are stored and accessed as chunked, hierarchical arrays, while GeoZarr specifies how this stored structure represents the scientific and geospatial meaning of the dataset..

As a secondary objective, GeoZarr extends the CDM base layer with additional capabilities required for geospatial and cloud-native applications. These extensions include multiscale overviews, which enable the representation of data at multiple levels of detail, and affine transformations, which define the spatial relationship between data coordinates and real-world locations. All extensions remain fully aligned with the CDM framework.

The CDM base layer also provides a generic framework capable of hosting metadata from a wide range of community standards. GeoZarr encourages the use of the Climate and Forecast (CF) Conventions, which are themselves defined around the CDM model, without imposing them as mandatory. This flexibility also supports metadata from other domain-specific standards such as GeoTIFF, GDAL, and similar geospatial conventions.

7.  GeoZarr Data Model

@startuml
class Group {
}

class Variable {
}

class Dimension {
}

class Attribute {
}

class DataArray {
}

class CoordinateVariable {
}

class Subgroup {
}

Group "1" *-- "0..*" Variable
Group "1" *-- "0..*" Dimension
Group "1" *-- "0..*" Attribute
Group "1" *-- "0..*" Subgroup
Variable "1" *-- "1" DataArray
Variable <|-- CoordinateVariable

@enduml

@startuml
skinparam classAttributeIconSize 0
skinparam linetype ortho
skinparam packageStyle rectangle
skinparam backgroundColor #FFFFFF
title GeoZarr Extensions – Geospatial Enhancements

package "Common Data Model (CDM)" {
  class Group
  class Variable
  class Dimension
  class Attribute
}

package "GeoZarr Extensions" {
  class AffineTransform
  class Overview
}

Variable --> AffineTransform : georeference
Group --> Overview : provides
Variable --> Overview : provides

@enduml

8.  Encodings for Zarr

This clause defines the normative mapping between the GeoZarr Data Model and the Zarr storage format. It specifies how the structural elements of the Common Data Model (CDM) — groups, variables, dimensions, and attributes — are encoded in Zarr v2 and Zarr v3, and identifies additional constraints introduced by GeoZarr.

GeoZarr’s encoding rules are limited to cases where explicit guidance is required for interoperability. GeoZarr does not redefine how CF, GDAL, or other metadata conventions map to CDM constructs — these mappings are already implemented in community libraries such as xarray, GDAL, and netCDF-Java.

The GeoZarr encoding rules therefore focus on CDM structure and semantics, with additional subsections specifying any Zarr-specific requirements for supported metadata conventions.

{
  "zarr_format": 3,
  "node_type": "array",
  "shape": [180, 360],
  "dimension_names": ["lat", "lon"],
  ...
}

{
  "zarr_format": 3,
  "node_type": "array",
  "shape": [180],
  "dimension_names": ["lat"],
  "data_type": "float32",
  "attributes": {
    "standard_name": "latitude",
    "units": "degrees_north",
    "axis": "Y"
  }
}

{
  "zarr_format": 3,
  "node_type": "array",
  "shape": [12, 180, 360],
  "dimension_names": ["time", "lat", "lon"],
  "attributes": {
    "standard_name": "air_temperature",
    "units": "K",
    "long_name": "Surface air temperature",
    "_FillValue": -9999.0
  }
}

{
  "zarr_format": 3,
  "node_type": "group",
  "attributes": {
    "title": "Example Dataset",
    "summary": "Multidimensional Earth Observation data",
    "institution": "Example Space Agency",
    "Conventions": "CF-1.10"
  }
}

{
  "multiscales": {
    "tile_matrix_set": "WebMercatorQuad",
    "resampling_method": "nearest"
  }
}

{
  "zarr_format": 3,
  "node_type": "group",
  "attributes": {
    "multiscales": {
      "tile_matrix_set": "WebMercatorQuad",
      "resampling_method": "nearest"
    }
  }
}

"tile_matrix_set_limits": {
  "1": {
    "min_tile_col": 0,
    "max_tile_col": 1,
    "min_tile_row": 0,
    "max_tile_row": 1
  }
}