OGC Testbed-21 Call for Sponsors

GeoDataCubes are multi-dimensional arrays of data that have one or more spatial dimensions. Advancement of GeoDataCubes offers several potential benefits. First, it could improve the ability to integrate information from multidimensional datasets and thus lead to greater efficiency. Second, it could help to grow the community of scientific users of Earth Observation data.

Progress from Testbed-20 included development of a GeoDataCube API Profile [1], testing of the usability of GeoDataCubes [2], and demonstration of how to handle the provenance of GeoDataCubes [3]. While significant progress has been made on the definition of GeoDataCubes, there has not been sufficient progress to advance the GeoDataCube concept towards standardization. Consequently, the following ideas for Testbed-21 have been suggested jointly by OGC staff, participants of previous Testbeds, and the sponsor community.

Respondents to the Testbed-21 RFI highlighted the need for lowering the access barrier to geospatial data for the wider scientific community, while preserving reliable, trustable and explainable results, especially for situations where geospatial data is leveraged for the training of Artificial Intelligence (AI) models or decision-making support.

The Committee on Earth Observation Satellites (CEOS) defines Analysis Ready Data (ARD) as "satellite data that have been processed to a minimum set of requirements and organized into a form that allows immediate analysis with a minimum of additional user effort and interoperability both through time and with other datasets." In the previous OGC Testbed-19 initiative, this definition was expanded to define ARD "as geospatial data that have been processed to a minimum set of requirements and organized into a form that enables immediate analysis with a minimum of additional user effort and interoperability both through time and with other datasets."

These definitions highlight that ARD has the potential to offer several benefits to a wide variety of users. First, it could increase the speed at which users are able to analyze data. Second, it could reduce the risk of analysts missing key information in the data that they analyze. Third, it could improve the ability of cross-functional teams to collaborate in analysis across disciplines.

In discussions with Strategic and Principal Members, they expressed interest in experimentation with existing standards to facilitate the description and publication of ARD, as well as clearly defining the scope and levels of readiness. Such experimentation could also involve the following activities recommended by previous OGC Testbeds for further investigation:

Respondents to the Testbed-21 RFI highlighted the increasing role of reanalysis model data in Earth Observation applications across atmospheric, marine, and terrestrial domains. Both ARD and the OGC Training Data Markup Language for AI (TrainingDML-AI) Standard could play a role in such reanalysis use cases.

The adoption of Artificial Intelligence (AI) in geoinformatics has progressed at an exemplary speed over the past decade. Whereas many of the early applications that leveraged AI focused feature extraction and classification supported by Machine Learning (ML), many of the recent applications have explored productivity of the data user - supported by Generative AI. There is therefore wide-ranging interest, from prototyping of AI-focused extensions of existing OGC Standards (e.g., facilitating query); to prototyping of new AI-focused OGC Standards.

AI offers several benefits to geospatial applications. First, the ability to automate functionality offers a potential increase in productivity. Second, greater automation also offers the potential for cost savings. Third, the capabilities of generative AI applications offer the potential to increase the creativity of geospatial users involved in regional planning, or Architecture, Engineering and Construction (AEC).

Extensive research has been undertaken on the topic of AI for urban geography [4] . Many of those studies focus on socio-economic aspects, such as population flow analysis, urban growth, and social differentiation. The variability of the studies and the semantics they use highlight the need for common semantics where possible. Some of the advances made in combining Machine Learning and Semantic Web technologies, such as those by Breit et al. [5], suggest that there could be an opportunity to leverage a common model such as CityGML within the context of AI for Urban Geography. To this end, work on prototyping CityRDF, an RDF encoding of CityGML and CityJSON, could play a role in this endevor.

OGC was one of the first standardization organizations to publish a standard for describing training data, the OGC Training Data Markup Language for AI (TrainingDML-AI) Standard. Whereas TrainingDML-AI has addressed many of the challenges related to the description of training data used for AI, some challenges remain. For example, the use of common identifiers in describing the provenance of AI/ML models and the storage of such models continue to be areas of active interest. Testbed-21 could therefore explore the following:

Other ideas for AI-related topics that Testbed-21 could explore include:

A Building Block is a package of functionality defined to meet specific business needs. Many OGC Standards are structured with modular sets of requirements (or requirement classes) that collectively function as a reusable building block (OGC 05-020r29). A building block fulfils a function that can operate in the larger context of an implementation, including through combination with other OGC building blocks to create novel implementations. Building blocks developed for one Standard can be reused in another Standard.

The benefits of the building blocks approach are numerous. First, the approach could potentially offer greater availability of the full provenance of components of a standard. Second, the modularity could enable easier integration of components into larger solutions. Third, the wider pool of potential components could potentially offer greater reliability of systems.

Recently published research by OGC Staff and OGC Members considered the different approaches for conceptualizing data spaces, their structures and components [6]. The research also mapped data space components and structures to well-established data standards. A previous initiative, OGC Testbed-19, identified the following characteristics as generic for a building block (OGC 23-050):

To support the development of building blocks, OGC has been developing a Building Blocks register. Whereas the register is already showing great potential for facilitating interoperability, there still exists a need to enhance its ability to support Spatial Data Infrastructure (SDI) initiatives. In parallel with the research presented by Noardo et al. [6], which focused on building blocks for data spaces, Testbed-21 could explore how the same concepts transfer to spatial data infrastructure such as the Canadian Geospatial Data Infrastructure (CGDI) or the US National Spatial Data Infrastructure (NSDI).

Citizen-centric data has long been an active area of research, particularly in relation to Smart Cities. Amongst its many benefits are increased public engagement, and the potential for greater trust in data. Some previous relevant work on citizen-centric data includes research by Aguilera et al. [7] and Ju et al. [8]. Aguilera et al. [7] explored citizen centric data services for Smart Cities, whereas Ju et al. [8] considered the role of citizen-centered big data analysis. What has changed in recent years has been the increasing need to make information more available and understandable to the citizen, not just the expert. This means enabling users to understand the lifecycle and provenance of their data. More recently, Meenken et al. [9] proposed a lifecycle for citizen-centric data.

Citizen-centric data also presents opportunities for the GeoPackage Standard as the Standard could improve the ability of basic users to store data on personal devices or even to move data from one device to another. For instance, Testbed-21 could build upon the security approaches developed in previous Testbeds such as Testbed-17 [10] and Testbed-18 [11] in order to prototype an approach for storing citizen-centric data in a standards-based way. Moreover, respondents to the Testbed-21 RFI highlighted the need to explore GeoParquet or GeoZarr support in cloud computing environments. Such exploration could potentially enable basic users to move their geospatial data between cloud computing environments more efficiently.

In recent years, there has been a growing need to improve the discoverability and accessibility of scientific publications based on their geographic relevance. This need has, in part, been fueled by the increasing use of location-referenced data in scientific studies. Consequently, there have been calls for the integration of geospatial metadata into scientific articles. One possible approach to achieve this is the Digital Object Identifier (DOI) system. The DOI system is used by many research publishers such as universities, as well as government organizations such as NASA, ESA, European Commission, and CNES. Some of the benefits of DOIs include persistent identification, enhanced discoverability, greater integration with journal systems, and more efficient tracking of metrics.

The DOI system is a persistent identifier used to uniquely identify documents, datasets, and other types of objects. The specification for DOIs is standardized by the International Organization for Standardization (ISO) as ISO 26324:2012 Information and documentation — Digital object identifier system. By associating DOIs with geographic information, resource publishers would then be able to carry out more effective querying, visualization, and analysis of scientific publications. Testbed-21 could explore possible solutions for associating DOI records with geospatial data. This could, for example, involve prototyping an extension of the DataCite and Crossref metadata schemas. DataCite and Crossref are some of the DOI registration agencies that have published their metadata schemas.

The Testbed could also examine possible alignment with OGC Standards, for example:

The GEOINT Imagery Media for ISR (GIMI) Profile of the ISO base media file format (ISOBMFF) standard is a candidate standard that provides a modular container format for both still and time-based multimedia data such as single images, motion imagery, tiled imagery, and audio. GIMI is also a profile of the High Efficiency Image File Format (HEIF). GIMI supports the incorporation of security marking, timing, and content identification, as well as other information. Some of the benefits envisaged for GIMI include support for long term sustainment of Enterprise software capabilities, support for integration of current and future imagery domain metadata models, minimizing of the risk of file size bloat, and so on.

To support the development of GIMI, Testbed-20 defined a candidate Standard for Geographic High Efficiency Image Format (GeoHEIF) [12]. The Testbed also identified several lessons and best practices [13], supported by experiences from extending open source software during the Testbed [14]. The candidate Standard specified requirements and encoding rules for using the High Efficiency Image File Format (HEIF) for the exchange of georeferenced or geocoded imagery. The Testbed proposed that the candidate Standard should be extended to support use cases based on Motion Imagery (video, streaming). Further, the Testbed proposed that future work, for example in Testbed-21, could specify a binary encoding to describe dimensions and properties of the images. The addition of the offset and scale factor in the description of the continuous properties could allow for a more compact binary encoding. For example, temperature is typically represented as a floating number but using a scale factor could be encoded as an integer.

Testbed-21 could explore inclusion of a tiling mechanism to address requirements for making GeoHEIF a Cloud Optimized format, a concept referred to in the previous Testbed as COHEIF (Cloud Optimized GeoHEIF). Testbed-21 could therefore involve benchmarking of COHEIF and contrasting with Cloud Optimized GeoTIFF (COG), both in a local drive and over the Internet using HTTP range use cases. This could build on the benchmarking successes of Testbed-20 [15]. Furthermore, the use of HEIF with a high efficiency codec can produce significant file size reduction over PNG tiles, and in combination with a low-overhead image file format, it is likely that it will probably be of the same order as PNG even for very simple files. Therefore, further benchmarking of compression algorithms, geo-enabled HEIF drivers against COG drivers, and video metadata handling would improve the understanding of available options for optimizing the performace of GeoHEIF. The benchmarking could inform future use cases, building on the use cases identified in the previous Testbed [16].

Another area of potential development of GIMI is in relation to the ontology used for describing metadata. There are at least three related standardization efforts that could facilitate development of an ontology for GIMI. The first is the on-going standardization of GeoDCAT. The second is the Semantic Sensor Network Ontology (SSN), a standard jointly developed by OGC and the World Wide Web Consortium (W3C). The third relates to the National System for Geospatial Intelligence (NSG) and the associated NSG Enterprise Ontology (NEO). In previous Testbed work, OGC members prototyped NEO Application Profiles that facilitate Linked Data applications in the NSG [17]. Since then, the NEO has been registered as a standard for the NSG. Therefore, Testbed-21 could examine the potential for an ontology that leverages NEO and the SSN to support the embedding of metadata in GIMI files whilst also being in alignment with GeoDCAT.

There is a growing need for data integrity, provenance, and trust (IPT) across the various domains that make use of geospatial data. Therefore, agile reference architectures that are reinforced with IPT frameworks are essential. These frameworks are well-positioned to expand into areas beyond geospatial data, as demonstrated in the OGC Testbed 20 IPT activity. Some of the areas that could benefit from the deployment of interoperable IPT frameworks include public health, and defense, where continuous monitoring, data integration, and interoperability are crucial.

Testbed-21 activities could involve experimentation with approaches for validating resource integrity, and building trust, as well as ensuring provenance verification through the implementation of metadata and OGC API Standards. For example, a Testbed activity could involve development of additional IPT Building Blocks that remain agnostic to specific encoding formats. By leveraging IPT frameworks, it is envisaged that geospatial solutions will be able to support the management and use of data across dynamic and complex networks, thus ensuring that data remains reliable and trustworthy. Furthermore, as highlighted by Testbed-20, by applying IPT in public health the risk of zoonotic spillover could potentially be mitigated [18].

Respondents to the Testbed-21 RFI highlighted the need for definition of metadata requirements, guidance and representations needed to achieve IPT, according to the relevant application use cases for AI and ARD. They also highlighted the need to determine how to best manage provenance in a standards-based way. They noted that the Testbed could examine the specification of provenance business logic such that such logic could be implemented in geospatial applications in order to automate the documentation of a chain of trust in data processing. They reiterated the role of Federated Agile Collaborating Trusted Systems (FACTS) and APIs in helping to implement IPT frameworks.

Respondents to the Testbed-21 RFI also expressed interest in decentralized solutions for data storage and identity management which are related to the "agile reference architecture" and IPT topic. They gave an example of exploring use of a decentralized Interplanetary File System (IPFS) solution with Amazon S3 data backend stores and applying decentralized authentication through SOLID.

There are at least three areas of standards development emerging in relation to metadata and data infrastructures. First, the Features and Geometry JSON (JSON-FG) candidate Standard is nearing completion. The candidate Standard is now stable and thus offers a path towards a JSON encoding for new data models such as MUDDI, as well as long-established data models that use other encodings. Examples of those long-established data models include the GeoSciML, IndoorGML, PipelineML, GroundWaterML, WaterML, InfraGML and CityGML Standards - all of which currently solely use the Geography Markup Language (GML).

The second area of standards development is that of metadata infrastructures. Within this area, the development of ISO 19115-4 (the JSON encoding of ISO 19115) is one of the most significant activities. Once completed, ISO 19115-4 will offer a path for profiles of ISO 19115-1 standards such as the NSG Metadata Foundation (NMF) and DGIWG Metadata Foundation (DMF) specifications.

The third area of standards development relates to infrastructures for High-Performance Computing (HPC). The increasing complexity and volume of geospatial data necessitates the application of HPC for efficient analysis and processing. Testbed-20 noted that research into the potential use of existing cloud-optimized formats as HPC-optimized formats, best practices for geospatial data indexing and partitioning, and optimal solutions for an HPC-optimized format could guide future developments in the field of HPC [19].

Respondents to the Testbed-21 RFI highlighted the need to experiment with interoperability challenges of data lakehouse solutions, with a focus on OGC Standards. A data lakehouse is data architecture that combines the capabilities of a data lake with those of a data warehouse. As a data lakehouse can hold both structured and unstructured data, the RFI respondents highlighted the issues surrounding the documentation of metadata about data quality and confidence of processing results. Respondents to the Testbed-21 RFI also highlighted the need to address the exact mechanisms to find-bind interfaces that conform to OGC API Standards, especially given the emerging GeoDCAT candidate Standard and its potential support in OGC API - Records.

The Testbed-21 RFI respondents also noted the potential to integrate the OGC 3D Tiles Community Standard and other OGC Standards into the draft Hyperspace Modelling Language (HSML) standard of the IEEE. The requirements for HSML are listed in the IEEE P2874 Standard for Spatial Web Protocol, Architecture and Governance.

Calls for data-driven policy making have grown over the past few decades [20] [21]. This has led to the growth of the market for services supporting the reporting of global indicators. The scale and growth of the market suggests significant global demand for geospatial services that support reporting against global schemes, both related to corporate financial disclosures and targets for progress against the United Nations Sustainable Development Goals (UN SDGs). However, currently there lacks the standardization needed for robust diligence, monitoring, and verification processes that could support data-driven policy making. To address this need, the OGC established the OGC Geospatial Reporting Indicators Standards Working Group to develop standards that support transparency in international reporting indicators which rely on geospatial data, referred to as Geospatial Reporting Indicators (GRI). The initial focus areas of the working group is the standardization of indicators associated with Land Degradation Neutrality (LDN) and Climate resilience. To facilitate the standardization activities of the SWG, Testbed-21 could explore both the infrastructure needs and the data model requirements for GRI.

The representation of triaxial bodies (e.g., asteroids, moons, comets) in the context of the solar system is becoming increasingly important. OGC Testbed-18 evaluated the potential use of existing OGC/ISO Standards for applications on celestial bodies other than planet Earth. These applications include data for other planetary bodies, for objects in orbit around those planetary bodies, and for objects in free flight within our solar system. Building on the lessons identified by Testbed-18, Testbed-19 addressed three threads related to non-terrestrial applications. The first thread focused on coordinate reference systems and coordinate transformations. The second thread focused on the use of 4D GeoPose (NeoPose) for highly dynamic non-terrestrial and terrestrial applications. The third thread advanced the work on Moving Features and Sensor Integration with an eye toward incorporation of the work from the first two threads.

The modeling and visualization of celestial bodies in the solar system often involve complex three-dimensional (3D) shapes; therefore, there is a need for standardized methods for representing their geometries. Whereas previous Testbeds explored the representation of positions of celestial objects, interoperability challenges for the modeling and visualization of celestial bodies such as planets, moons, asteroids, and comets endure. These celestial bodies exhibit triaxial ellipsoidal forms due to their rotation and gravitational forces and thus involve complex three-dimensional (3D) shapes. Therefore, there is a need for standardizing the methods for representing the geometries of such celestial bodies in order to more effectively manage and visualize the objects in geospatial systems. To help address this need, Testbed-21 could establish a standards-based framework for managing triaxial bodies within the context of the solar system. Further, as highlighted by respondents to the Testbed-21 RFI, the Testbed could explore the potential use of Discrete Global Grid Systems (DGGS) to support 3D and 4D indexing of celestial bodies (not just the Earth). Furthermore, respondents to the Testbed-21 RFI also highlighted the need to research and document the effectiveness of 4D radiance field models for visual positioning use cases, e.g. responding to the presentation of a video or image with a sequence of OGC GeoPose objects or single GeoPose, respectively.

This topic seeks to advance data integration on multiple fronts. First, the Arazzo specification from the OpenAPI initiative potentially offers a means of chaining implementations of OGC API Standards together. A Testbed activity to help understand how the Arazzo specification could be applied by the OGC API - Processes Standard, as an alternative workflow specification amongst others such as the Common Workflow Language (CWL), would help to broaden the potential adopter community. Second, prototyping of a JSON encoding for Coordinate Reference System (CRS) definitions that is consistent with ISO 19111 could also be a Testbed activity. The CRS Standards Working Group has already started considering options for a CRS JSON encoding.

Finally, a series of activities that leverage or enhance OGC RAINBOW, the definitions server. Some ideas for those activities include:

The World Economic Forum recently stated that "The space economy is expected to be worth $1.8 trillion by 2035 as satellite and rocket-enabled technologies become increasingly prevalent, according to a new report." [22]. As the space economy grows, it can be expected that it will become increasingly important to provide the means to more effectively monitor, visualize and analyze the space domain. Such monitoring, visualization and analysis can help to achieve Space Situational Awareness (SSA). Therefore, Testbed-21 could demonstrate the potential role of OGC Standards in the location, modeling and representation of space objects, space debris and space weather phenomena.

Respondents to the Testbed-21 RFI highlighted the need for testing and demonstrating the application of Discrete Global Grid Systems (DGGS) to support the identification and tracking of space objects, space debris and space weather phenomena in space and time. Such an approach could potentially make it possible to assess and represent various risks in space such as collisions with space debris and extreme solar storms. Respondents to the Testbed-21 RFI also highlighted the need to explore the interoperability challenges of radiance field modelling. This could include the use of artificial neural networks, spatio-temporal elements such as 3D/4D Gaussians, and high-performance rendering via splatting or raytracing from a Radiance Field Modelling Service (RFMS) that supports the OGC GeoPose Standard.

Other topics have featured in recent discussions across the OGC. Those additional topics have included: