Mapping Borehole observations using a true 1-D CRS
By Florence Tan with Ben Caradoc-Davies and Simon Cox
For a long time, the use of spatial representations other than a conventional 2D cartographic view has been a rather dark corner of the GIS world. However, some recently adopted OGC standards promise to allow communities that need these, such as facilities and engineering, urban planning and design, hydro-geology, and the solid-earth geosciences, to integrate their world with the view normally presented in the geospatial community. These include
- CityGML 2.0 [1] – a GML 3.2-compliant upgrade to OGC’s first community application model for representing urban objects and landscapes in 3D
- GML 3.3 [2], which includes an XML representation of (spatial) Linear Referencing, as defined in OGC Abstract Specification Topic 19 [3]
Boreholes are features appearing in several environmental science disciplines, particularly to support observations and monitoring of subsurface phenomena. Observations are made in boreholes using down-hole logging instruments, with data indexed by distance along the hole [8]. Thus a 1D spatial reference system is the natural frame for borehole observations, allowing them to be correlated and easily visualized together for analysis [4].
These developments have been used in a “Secret Friday-morning Project” in Commonwealth Scientific and Industrial Research Organisation (CSIRO). A small question from one of the developers to one of the scientists – “How to manage position in boreholes?” – has lead to the use of a Linear Coordinate Reference System for indexing borehole data in production systems, using OGC standards.
In practice the use of a 1D spatial representation may require some workarounds. In particular, as popular spatially enabled databases such as PostGIS only support 2D coordinates, positions along the borehole must be represented as 2D coordinates in a ‘Lateral Offset Linear Coordinate Reference System’ [5], with the second coordinate set to 0.0 so that the position coincides with the ‘linear’ element.
The theory was tested using the OpenGeo GeoServer software [6]. The test turned out to be almost trivial to execute, thanks to recent upgrades to GeoServer that added support for (a) GML 3.2 (b) complex ‘application schemas’ based on GML 3.2. Since the new schema elements in GML 3.3, such as gmllro:LateralOffsetLinearSRS, are formalized as ‘applications’ of GML 3.2, GeoServer can be configured for GML 3.3 support merely by pointing it at the published schema [7].
The experiment vindicated a significant investment made by CSIRO into the OpenGeo GeoServer software, and also validated the OGC GML 3.3 standardized implementation of linear referencing. A challenge is now to the developers of spatially-enabled datastores to generalize their spatial support to topological dimensions other than 2D.
[1] http://www.opengeospatial.org/standards/citygml
[2] http://www.opengeospatial.org/standards/gml
[3] http://www.opengeospatial.org/standards/as
[4] https://www.seegrid.csiro.au/wiki/CGIModel/BoreHolesAndObservation#Linear_Referencing_40Borehole_41
[5] http://www.opengis.net/doc/gml/3.3#clause-9.5
[6] http://geoserver.org/display/GEOS/What+is+Geoserver
The Geoserver version used for testing was GeoServer Trunk Rev16340 http://files.ivec.org/geoserver/geoserver-trunk/2011-09-25/
[7] http://schemas.opengis.net/gml/3.3/linearRefOffset.xsd
Comments
Peter Baumann (not verified)
Wednesday, 2012-08-22
Permalink
This is very relevant work
This is very relevant work indeed - exciting to see how we gradually break free from the traditional view that there is lat and long, "and nothing else matters". OGC has taken a lead in establishing a CRS universe that can deal with truly multi-dimensional CRSs (and 1-D is a special case, after all). There is a Candidate Best Practice document [1], already resolved by the Web Coverage Service (WCS) group which, due to the nature of their data, is a forerunner in this field. This document establishes rules for identifying CRSs, and even allows to create ad-hoc combinations of existing CRSs. For example, a lat/lon/t CRS for satellite image timeseries might be composed from WGS84 and ISO 88601, which describes temporal CRSs.
Of course, using the coordinate facilities for storing points with a total different meaning - as done in the project reported - is a hack of the tools. But as we have no choice with current tools, given their lack of support beyond lat/lon, this endeavor constitutes an interesting experiment, and hopefully an incentive for GIS developers to boldly go into new dimensions. We are living in exciting times...
This is very relevant work indeed - exciting to see how we gradually break free from the traditional view that there is lat and long, "and nothing else matters". OGC has taken a lead in establishing a CRS universe that can deal with truly multi-dimensional CRSs (and 1-D is a special case, after all). There is a Candidate Best Practice document [1], already resolved by the Web Coverage Service (WCS) group which, due to the nature of their data, is a forerunner in this field. This document establishes rules for identifying CRSs, and even allows to create ad-hoc combinations of existing CRSs. For example, a lat/lon/t CRS for satellite image timeseries might be composed from WGS84 and ISO 8601, which describes temporal CRSs. An open-source implementation handling such CRS identifiers has been completed and is available meantime.
Of course, using the coordinate facilities for storing points with a total different meaning - as done in the project reported - is a hack of the tools. But as we have no choice with current tools, given their lack of support beyond lat/lon, this endeavor constitutes an interesting experiment, and hopefully an incentive for GIS developers to boldly go into new dimensions. We are living in exciting times...
[1] http://www.ogcnetwork.net/system/files/11-135_OGC-NA_Name-Type-Specifica...
Peter Parslow (not verified)
Wednesday, 2013-01-16
Permalink
A bit pedantic, perhaps, but
A bit pedantic, perhaps, but I'm sure that CityGML 2 is a GML 3.1.1 application schema, not GML 3.2.