Observations & Measurements - GML Simple Features Profile
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
omsf-gml
omsf-json
presentations
.gitignore
OMSF_ImplementationModel.png
README.md

README.md

OGC Observations and Measurements - Simple Feature Encodings (OMSF)


NOTE: These encodings are work-in-progress, and at this point, has not been endorsed by the OGC or any other standards organization. They may (and probably will) change in a backwards incompatible way during the drafting process.

There is an on-going discussion in leveraging the OMSF encodings in the INSPIRE alternative encodings action (MIWP 2017.2) for a simplified encoding of the INSPIRE O&M datasets.

The namespaces http://www.opengis.net/omsf-gml/1.0 and http://www.opengis.net/omsf-json/1.0 have not (yet) been approved by the OGC Naming Authority, and thus may also change. Consider yourself warned.


Handling complex feature structure of the O&M XML Implementation (as in OGC 10-025r1) is only possible for many WFS server and client software with a considerable implementation cost, added code complexity and lower performance. The purpose of this activity is to define simple encodings for the most used O&M Observation types, and thus enable interoperable O&M data exchange between existing software applications, servers and clients limited to using simple (non-complex) GML features and/or GeoJSON.

OMSF Implementation model

OMSF Implementation model

All OMSF encodings are based on the same implementation model profile derived from the O&M Observation Core and Specialized Observations conceptual models defined in Observations and Measurements v2.0 (OGC Document 10-004r3, also published as ISO 19156:2011, Geographic information — Observations and Measurements). Although the implementation model is encoding-independent, the feature structure and property types have been intentionally chosen to be easily encodable as simple features according to the requirements of the GML Simple Features Profile version 2.0 (OGC Document 10-100r3.

This profile contains implementation model classes only for the following O&M Core and Specialized Observation UML classes:

O&M v2.0 class OGC name OMSF feature
OM_CategoryObservation http://www.opengis.net/def/observationType/OGC-OM/2.0/OM_CategoryObservation omsf:CategoryObservation
OM_CountObservation http://www.opengis.net/def/observationType/OGC-OM/2.0/OM_CountObservation omsf:CountObservation
OM_Measurement http://www.opengis.net/def/observationType/OGC-OM/2.0/OM_Measurement omsf:MeasureObservation
OM_Observation http://www.opengis.net/def/observationType/OGC-OM/2.0/OM_Observation omsf:GenericObservation
OM_TimeSeriesObservation http://www.opengis.net/def/observationType/OGC-OM/2.0/OM_TimeSeriesObservation omsf:MeasureTimeseriesObservation
OM_TruthObservation http://www.opengis.net/def/observationType/OGC-OM/2.0/OM_TruthObservation omsf:TruthObservation

Implementation model classes for the O&M UML classes OM_ComplexObservation, OM_DiscreteCoverageObservation, OM_GeometryObservation, OM_PointCoverageObservation, and OM_TemporalObservation are not included. Thus this implementation model is intentionally more limited than the one used for the Observations and Measurements - XML Implementation (OGC Document 10-025r1) and OGC Observations and Measurements – JSON implementation (OGC Document 15-100r1) which are able to express the full O&M abstract model. However, expressing all these is possible by using the omsf:GenericObservation feature type with a remote reference to the Observation result.

This profile does not provide encodings for the sampling feature data, as the feature of interest is only presented by it's geometry, optionally by it's name and, also optionally, by a remote reference to the description of the complete feature of interest.

Property mapping

The implementation model of the OMSF is a simplified version of the Observation class as defined in the ISO 19156 standard. The following table summarises the simplification decisions applied:

O&M attribute/association O&M type O&M Multiplicity OMSF property OMSF type
featureOfInterest association with GFI_Feature 1 geometry GM_Object
featureOfInterest association with GFI_Feature 1 samplingFeature Reference to an external resource
featureOfInterest association with GFI_Feature 1 ultimateFeatureOfInterest Reference to an external resource
metadata association with MD_Metadata 0..1 metadata Reference to an external resource
observedProperty association with GF_PropertyType 1 observedProperty Reference to an external code list
parameter NamedValue 0..n n/a n/a
phenomenonTime TM_Object 1 phenomenonTime TM_Object
procedure association with OM_Process 1 usedProcedure Reference to an external resource
procedure association with OM_Process 1 madeBySensor Reference to an external resource
relatedObservation association with self 0..n n/a n/a
result Any 1 result varied
result Any 1 timeStep TM_Instant
result Any 1 unitOfMeasure Reference to an external code list
resultQuality DQ_Element 0..n n/a n/a
resultTime TM_Instant 1 resultTime TM_Instant
validTime TM_Period 0..1 validTime TM_Period

Rationale for the not included (n/a) properties:

  • parameter: 0..n multiplicity of name-value pairs (as user defined types) would require compliance level SF-1, or a specially encoded list type. Trade-off between completeness and simplicity.
  • relatedObservation: not a problem to include technically (as a reference), but rarely used in practice.
  • resultQuality: embedded quality info would a user-defined type and thus SF-1 level. Rarely used in practice, trade-off between completeness and simplicity.

Design considerations

The following primary design goals have been followed (in priority order):

  1. The defined feature types must be suitable for encoding using GML Simple Features Profile level SF-0 or SF-1.
  2. Each defined feature type must have a relevant geometry property for spatial processing and map visualization purposes.
  3. The defined feature types must follow the O&M model structure and property naming as long as it does not conflict with higher priority design goals.
  4. The defined GML feature types should be as simple as possible, but not simpler (so called Einstein's razor).

Feature of interest of the described Observation is represented as separate parts for the sampling feature and the ultimate feature of interest as suggested by W3C Extensions to the Semantic Sensor Network Ontology proposal. Also to align with the W3C Semantic Sensor Network Ontology specification, the method and the implementation of the measurement procedure has been split into two separate properties: usedProcedure and madeBySensor.

As the GML Simple Features Profile requirements for level SF-0 does not allow for including the structural (object oriented) presentation of the feature of interest of the Observation, the sampling feature is encoded by one mandatory property geometry and an optional property samplingFeature. In a similar fashion the ultimate feature of interest (the model of the observed real-world object) is encoded using an optional property: ultimateFeatureOfInterest. The geometry property of a OMSF Observation feature is the geometry of the sampling feature of the observation, or if no sampling feature was used, a representative geometry of the ultimate feature of interest.

In contrast to the OMXML (complex feature) implementation model, hard-typing is used for the different Observation types: GenericObservation, MeasureObservation, CategoryObservation, CountObservation, TruthObservation and MeasureTimeseriesObservation are each defined as separate feature types with fixed result value types. This is an intentional trade-off between simplicity and flexibility.

Observations with complicated results, such as coverages, have been considered out-of-scope of this profile. However, it's possible to encode these using the GenericObservation feature type with a reference to the remotely provided result. The observation level metadata and the definition of the observed property are always externally linked from the OMSF Observations.

Timeseries data

OMSF implementation model contains MeasureTimeseriesObservation feature type for encoding simple, double-valued time series data. Time series with result values for several points in time does not fit with the GML Simple Features Profile compliance level SF-0 without mild violence, since repeated elements are not allowed. Technically the time series values (and even time instances) could be encoded inside a single element using list type, but encoding and decoding would require special processing, which would at least partly defeat the gains of restricting the feature type to SF-0. So the MeasureTimeseriesObservation contant repeated timeStep and result properties which require using SF-1 compliance level when encoding into GML Simple Features.

GML encoding example:

<omsf:MeasureTimeseriesObservation gml:id="f-1">
  <omsf:phenomenonTimeStart>2017-08-17T12:00:00Z</omsf:phenomenonTimeStart>
  <omsf:phenomenonTimeEnd>2017-08-17T18:00:00Z</omsf:phenomenonTimeEnd>
  <omsf:resultTime>2017-08-17T12:11:20Z</omsf:resultTime>
  <omsf:usedProcedure xlink:href="http://xml.fmi.fi/process/met-surface-observations" xlink:title="Meteorological surface observations" />
  <omsf:observedProperty xlink:href="http://vocab.nerc.ac.uk/collection/P07/current/CFSN0023/" xlink:title="air_temperature" />
  <omsf:samplingFeatureName>Helsinki Kumpula weather observation station</omsf:samplingFeatureName>
  <omsf:geometry>
      <gml:Point gml:id="p-1" srsName="http://www.opengis.net/def/crs/EPSG/0/4258" srsDimension="2">
          <gml:pos>60.20307 24.96131</gml:pos>
      </gml:Point>
  </omsf:geometry>
  <omsf:ultimateFeatureOfInterestName>Helsinki Kumpula</omsf:ultimateFeatureOfInterestName>
  <omsf:ultimateFeatureOfInterestReference xlink:href="http://sws.geonames.org/843429/about.rdf"/>
  <omsf:timeStep>2017-08-17T12:00:00Z</omsf:timeStep>
  <omsf:timeStep>2017-08-17T13:00:00Z</omsf:timeStep>
  <omsf:timeStep>2017-08-17T14:00:00Z</omsf:timeStep>
  <omsf:timeStep>2017-08-17T15:00:00Z</omsf:timeStep>
  <omsf:timeStep>2017-08-17T16:00:00Z</omsf:timeStep>
  <omsf:timeStep>2017-08-17T17:00:00Z</omsf:timeStep>
  <omsf:timeStep>2017-08-17T18:00:00Z</omsf:timeStep>
  <omsf:unitOfMeasure xlink:href="www.opengis.net/def/uom/UCUM/degC" xlink:title="Degree Celsius"/>
  <omsf:result>12.5</omsf:result>
  <omsf:result>12.0</omsf:result>
  <omsf:result>11.0</omsf:result>
  <omsf:result>13.2</omsf:result>
  <omsf:result>12.5</omsf:result>
  <omsf:result>14.1</omsf:result>
  <omsf:result>14.1</omsf:result>
</omsf:MeasureTimeseriesObservation>

JSON encoding example:

{
  "type": "Feature",
  "id": "f-1",
  "geometry": {
    "type": "Point",
    "coordinates": [ 24.96131, 60.20307 ]
  },
  "properties": {
    "observationType": "MeasureTimeseriesObservation",
    "phenomenonTimeStart": "2017-08-17T12:00:00Z",
    "phenomenonTimeEnd": "2017-08-17T18:00:00Z",
    "resultTime": "2017-08-17T12:11:20Z",
    "usedProcedureName": "Meteorological surface observations",
    "usedProcedureReference": "http://xml.fmi.fi/process/met-surface-observations",
    "observedPropertyName": "air_temperature",
    "observedPropertyReference": "http://vocab.nerc.ac.uk/collection/P07/current/CFSN0023/",
    "samplingFeatureName": "Helsinki Kumpula weather observation station",
    "ultimateFeatureOfInterestName": "Helsinki Kumpula",
    "ultimateFeatureOfInterestReference": "http://sws.geonames.org/843429/about.rdf",
    "timeStep": [
        "2017-08-17T12:00:00Z",
        "2017-08-17T13:00:00Z",
        "2017-08-17T14:00:00Z",
        "2017-08-17T15:00:00Z",
        "2017-08-17T16:00:00Z",
        "2017-08-17T17:00:00Z",
        "2017-08-17T18:00:00Z"
    ],
    "unitOfMeasureName": "Degree Celsius",
    "result": [12.5, 12.0, 11.0, 13.2, 13.5, 14.1, 14.1]
  }
}

The using repeated properties allows client applications to treat both timeStep and result as arrays of simple values, which would not be possible using time-value-pair encoding.

GML and GeoJSON Encodings

OMSF contains both GML and GeoJSON encodings of the common OMSF implementation model defined above. For more details see

Acknowledgements

Work on this application profile was initiated by Ilkka Rinne for the needs of Vaisala and Finnish Meteorological Institute in 2017.