Proposal for extension of LOCN with properties for Coordinate Reference System and Level of Detail

Introduction

This is a proposal for extending the Location Core Vocabulary with semantics for Coordinate Reference System (CRS) and spatial resolution. The reason for proposing these two extensions is that both can help in the automated processing of geographical data, both on the level of single geometries and the data set level. Usage on the data set level is discussed on the page about extensions of LOCN for metadata.

The two concepts have not been found to exist in other commonly used RDF vocabularies for geographical data, but they are basic properties of geographical data, hence the proposed addition to this core vocabulary.

CRS

The Coordinate Reference System (CRS) is modelled by a class (locn:CRS) and a property (locn:crs). The property locn:crs is proposed to allow associating a geometry with a CRS. It can also be used to make known that a dataset contains geometries that use this CRS, for example in the metadata of a dataset. The domain of locn:crs is a rdfs:Class, to ensure that a CRS is always identified by a URI. Preferably, this URI gives access to meaningful data about the CRS.

Definition

Qname

locn:CRS

Type

Class

Description

An identification of a reference system that is used for the coordinates of a geometry, or for a coordinate reference system that is used in a dataset containing geometrical geographic data. Preferably the URI resolves to meaningful data about the coordinate reference system.

Range

unspecified

Domain

unspecified

Qname

locn:coordinateReferenceSystem locn:crs

Type

Property

Description

Allows association of a geometry, a collection of geometries or a dataset with a CRS

Range

locn:CRS

Domain

unspecified

Questions

1. What is the best name: coordinateReferenceSystem or spatialReferenceSystem, or something else?

- Depends what you want to use it for. CRS is a subclass of Spatial Reference System. A gazetteer is another kind of spatial reference system, in which locations are given as names. Post-codes are another kind. If the intention is to support the general case, then the domain is probably not 'Geometry'. If the intention is only to support geometries encoded using coordinates, then stick with CRS.

1. The CRS can be applied to different things: geometry or a dataset. Perhaps the domain specification is too restrictive? What would be lost if the domain is left unspecified?
2. There are multiple candidates for specifying that the domain of this property can be a dataset: dcterms:Dataset, void:Dataset or dcat:Dataset. Of these, dcterms:Dataset has the broadest definition, it can also be used for datasets that are not Linked Data. Both void:Dataset and dcat:Dataset are subclasses of dcat:Dataset. So is dcterms:Dataset the best choice?

spatialResolution

This property is proposed to allow associating a geometry or a dataset containing geometries with a measure of spatial resolution. This value can convey the level of detail of a geometry or a dataset, which can be an important factor in deciding if and how the data will be used by a consumer (similarly to a scale number on a paper map). The value can also be used to get an idea of positional accuracy of data, which can be important the data are interpreted or used in calculations, if positional accuracy is not otherwise described. If geometries have undergone generalisation with a certain generalisation distance (tolerance), that distance can be used as the spatial resolution.

Definition

Type

Property

Qname

locn:spatialResolution

Description

The minimal distance in meters beyond which two neighbouring points are indiscernible. This property can apply to a geometry or a dataset containing geometries.

xsd:decimal

Domain

locn:Geometry or dcterms:Dataset

Questions

1. This proposal implies that spatial resolution is the same in all directions. For some carefully measured positions, positional accuracy can vary in different directions, for instance the X, Y and Z directions. The proposed property does now allow such more detailed expressions. Is that a problem?
2. The current definition applies to vector data. Can or should the definition be made applicable to raster data? In that case the spatial resolution could be the diagonal distance of a raster cell, for instance.

- there is a lot of prior art relating to this, in the remote sensing community. Lets make sure to align, rather than re-inventing (and risking getting it wrong).

1. This property can be applied to different things: a geometry or a dataset. Perhaps the domain specification is too restrictive? What would be lost if the domain is left unspecified?
2. There are multiple candidates for specifying that the domain of this property can be a dataset: dcterms:Dataset, void:Dataset or dcat:Dataset. Of these, dcterms:Dataset has the broadest definition, it can also be used for datasets that are not Linked Data. Both void:Dataset and dcat:Dataset are subclasses of dcat:Dataset. So is dcterms:Dataset the best choice?

Definition in RDF (Turtle notation)

@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix dc: <http://purl.org/dc/elements/1.1/> .
@prefix dcterms: <http://purl.org/dc/terms/> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
@prefix locn: <http://www.w3.org/ns/locn#> .
@prefix vs: <http://www.w3.org/2003/06/sw-vocab-status/ns#> .

locn:CRS
	a rdfs:Class ;
	rdfs:label "coordinate eference system (CRS)"@en ;
	dcterms:identifier "locn:CRS"^^xsd:string ;
	rdfs:comment "An identification of a reference system that is used for the coordinates of a geometry, or for a coordinate reference system that is used in a dataset containing geometrical geographic data."@en ;
	rdfs:isDefinedBy <http://www.w3.org/ns/locn> ;
	vs:term_status "unstable"@en 
	.
locn:crs
	a rdf:Property ;
	rdfs:label "coordinate reference system (CRS) used"@en ;
	dcterms:identifier "locn:crs"^^xsd:string ;
	rdfs:comment "identifies the coordinate eference system (CRS) used."@en ;
	rdfs:isDefinedBy <http://www.w3.org/ns/locn> ;
	rdfs:range locn:CRS ;
	vs:term_status "unstable"@en 
	.
locn:spatialResolution
	a rdf:Property ;
	rdfs:label "spatial resolution"@en ;
	dcterms:identifier "locn:spatialResolution"^^xsd:string ;
	rdfs:comment "The minimal distance in meters beyond which two neighbouring points are indiscernible."@en ;
	rdfs:isDefinedBy <http://www.w3.org/ns/locn> ;
	rdfs:range xsd:decimal ;
	vs:term_status "unstable"@en 
	.

References

1. Discussion on public-lod@w3.org: http://lists.w3.org/Archives/Public/public-locadd/2014Sep/0000.html