11. Object-Based Profile

Intended Use --

An object-based profile to SDTS has been suggested as a way to improve SDTS, and other exchange standards, through the incorporation of object-oriented and feature-based concepts. An object-based profile would provide better support for exchanging and archiving geospatial data that conforms to advanced data models that are feature based, such as the data models of CEN GDF, DIGEST-VPF, NTF, IHO S-57, USGS DLG-F, SmallWorld GIS, Argus, Laser-Scan Gothic, Universal Systems CARIS++, and OGC.

A feature-based data model uses a semantic feature level that is distinct from the spatial primitive level. Thus, nonplanar real-world features can be represented even while using a planar graph at the spatial primitive level (Hickman, 1995). According to Arctur (1996)

- "Any given line feature such as a road or bridge can consist of one or more edge primitives (unique polyline segments).

- "Any given edge primitive can be used in one or more line features or area features."

The current SDTS Topological Vector Profile (TVP) can accommodate features and feature-to-feature relationships using composite objects, but some of the feature-based advantages may be lost or de-emphasized. It is possible that the geometric primitives for feature-based data could be raster as well as vector.

This profile may also be used to accommodate data that is seamless, is updated incrementally and historically by transactions, has dynamic and self-defining content schema and semantics, is supported in O-O DBMS's or extended-relational DBMS's, has multiple (multi-scale) geometric representations, has immutable ID's for features, and meets FGDC Framework and OpenGIS requirements.

Participants in the Open GIS effort have expressed interest in an object-based profile to SDTS. Success of interoperability through OpenGIS may compliment the use of SDTS as a bulk data exchange standard in many circumstances, and may emphasize the use of SDTS as a data archive standard.

Primary Sponsors, Interested Groups, and Contacts --

USGS DLG-F developers, FGDC Framework planners, O-O GIS vendors, and several participants in the Open GIS activity are interested in a possible object-based profile to SDTS. Information from David Arctur of Laser-Scan on O-O GIS and object-based suggestions for SDTS is available at

http://www.lsl.co.uk/~arctur/

The Open GIS web site is

http://www.opengis.org/

See also "Just What is SDTS and What Does it Really Need?" (Arctur & Fegeas, 1997) and "OGIS: Building on SDTS" (Fegeas, 1995).

Future Plans --

The USGS-NMD vector program is moving toward a flexible, feature-based data model to support FGDC Framework characteristics. Requirements from this feature-based approach will accelerate the need for an object-based profile to SDTS.

A feature-based model is also used by DIGEST-VPF, S-57, and GDF. The need to harmonize with these other standards as part of ISO approval may be related to an object-based profile of SDTS. One of the recommended characteristics of this profile is support for updates by transaction, and Version 3 of IHO S-57 now includes "incremental transfers" that allow the exchange of only information that has changed. The multi-scale representation capability needed in an object-profile is now available in CEN GDF. Sophisticated indexing, subtiling, and access capabilities are suggestions for the object-based profile, and these capabilities are now used in VPF. Support for multiple levels of topology and multiple data models (including raster) within one data transfer is a suggestion for the object-based profile and is similar to current DIGEST capabilities. Support for a content schema that is dynamic and not "hard coded" to a specific product or theme is a characteristic of DLG-F that has been recommended for the object-profile.

"When used to its full potential, SDTS will allow an extremely wide range of data models and user-defined geographic data to be encoded and transferred without information loss. Finally, the 'umbrella' and modular manner in which the SDTS is specified, and the methodology of defining profiles of SDTS, affords the means for the evolution of SDTS as requirements and technology change." (Arctur and Fegeas, 1997). More information from Arctur and Fegeas should be available at, and after, the SDTS workshop in September 1997.

As stated by Dr. Arctur in "Spatial Data Transfer Standard (SDTS): A GIS Vendor's Perspective" 10/29/96 from

http://www.lsl.co.uk/~arctur/

"[Vision for the next profile and content spec: dynamic and self-defining semantics] My impression of current R&D at USGS on DLG-F and a feature-based profile is that these could be pioneering a very smart way to approach the whole thing. They could constitute a framework for decomposing ANY data model's structure & semantics into parameter values that are stored AS DATA in the dataset, in addition to the real-world data itself. With this approach, then a single generic SDTS encoder/decoder should be able to understand ANY data model, and my concerns about reading & writing datasets for use by other GIS vendors goes away."