-
Notifications
You must be signed in to change notification settings - Fork 4
Future Work
ANZSoilML has defined a rich and robust conceptual model for the description, sampling and analysis of soils, and the modelling of their distribution, productivity or health. The next steps for the ANZSoilML community are to:
- repackage the data model in a way that separates concepts (the 'data dictionary') from technology (encodings, and web services and their APIs)
- define a modular framework that allows deployment of a range of tools that use technology and data that are appropriate to different communities of users
These communities will be varied. They may be made up of any combination (often in the same individual) of:
- Data providers who must publish and exchange rich, well documented data in a way that preserves the quality of the data without the loss of any content.
- Data engineers who work with data providers to publish data in ways the meet the needs of users or with users to extract and transform data according to their needs.
- Data and research scientists who may need fast access to raw data for models and simulations with optional access to rich metadata to help explain anomalies in results; or Semantic Web inferencing tools to discover new patterns in data. Traditional data and policy analysts most likely have similar needs.
- Web developers who need simple, terse and fast interfaces that use widely used and supported technology (ReST, JSON etc).
The end result of this work should be that an end user is only exposed to the ANZSoilML model in a form that they need. For example:
- Web developers will only need to work with definitions provided as HTML documents or JSON files.
- Maintainers of existing systems can continue to work with the GML/XML implementation.
- Research scientists can use Semantic Web tools.
Work on 'ANZSoilML 3.0' will begin in January 2019 as part of several projects:
- TERN Soil Data Infrastructure (Australia)
- Soil CRC Visualizing Australasian Soils (Australia and New Zealand)
- ACIAR Pacific Soil Portal (Australia and New Zealand)
- New Zealand National Soils Data Repository (New Zealand)
The OGC is conducting the ELFIE initiative - a series of Interoperability Experiments (IEs) designed to establish web-friendly implementations of linked environmental data. Our vision for ANZSoilML as a usable standard closely aligned with other environmental data standards matches the long term vision for ELFIE. We are active participants and expect to use the findings of ELFIE to design implement and the new version.
ANZSoilML 2.0.1 has been developed as a GML Application Schema according to a suite of ISO/TC 211 specifications:
- ISO19103 - Conceptual Schema Language
- ISO19109 - Rules for Application Schema
- ISO19136 - Geographic information – Geography Markup Language
We want to compliment this approach by providing versions that have value for other communities and support publication in ways that don't require specialist knowledge of tools such as the Unified Modelling Language (UML) and XML Schema Documents (XSD). Also, useful encodings like JSON-LD require models to be defined using the Resource Description Framework (RDF) making the provision of the model as an OWL ontology necessary.
- Create an ANZSoilML OWL ontology from the ANZSoilML 2.0.1 UML model according to the approach described by BRGM in the first ELFIE Engineering Report (OGC18-097 Annex A).
- Publish the ontology in a browsable environment. For example, a tool such as the CSIRO implementation of the Linked Data Registry.
- Provide machine-readable representations of definitions according to the encoding specifications discussed below.
In addition to the use of GML, we need to define an additional set of practices for publishing data in a manner best suited to other user types. For widespread adoption it is essential that ANZSoilML supports widely used encodings, currently JSON, however there are small, but very important, sets of users that must also be supported.
The XML implementation of ANZSoilML is very tightly bound its conceptual and logical models, and this makes it difficult to implement as the XML is therefore very complex. Little can be done about the complexity of the conceptual model (it needs to be, soils are complex) it does not follow that the XML schema used to encode data is as complex. A simpler XML encoding specification should be provided.
- Support JSON-LD encodings according to ELFIE findings. Requires the creation of JSON-LD context files linked to the ANZSoilML OWL Ontology.
- Define a simpler ANZSoilML XML encoding specification by combining classes explicitly defined at a conceptual or logical level in a single XML class if this can be done without loss of information. This is an equivalent practice to the Observations and Measurements virtual typing strategy OGC10-025r1, section 7.2.
A GML Profile is a logical restriction of an application schema. These allow communities refine an implementation by specifying only the parts of the schema they need.
Soil data need not be served according to the ANZSoilML XSD. Much of the analysis of soil data only needs values for a few properties at a given sampling location - the Observations and Measurements (O&M) standard meets this need with a simple data model. As no soil-specific classes are used the data can be quickly integrated with other environmental data provided using O&M.
- Define a profile of Observations and Measurements 2.0 for the provision of soil property data.