There is an unfortunate divergence in data modeling strategies between the spatial-centric approach in global information systems (GIS) and the business-centric approach adopted by database applications following Public Petroleum Data Model (PPDM) or similar recommendations.

The divergence can exist since there are many different ways of describing the spatial characteristics of an object. It emerges through the varying technical backgrounds of people involved in spatial data management. It is reinforced by constraints in the way that spatial database technology has commonly been used.

Current spatial database technology can now be used to reduce the separation between spatial- and business-centric approaches to data management. This allows the benefits of structured querying, singular data storage and spatial analysis to be shared across data ranging from high-volume, low-detail traded information through highly detailed information on objects for which a site has an operating responsibility.

Spatial vs. business-centric data models

Practical implementations of exploration and production (E&P) databases incorporating spatial data often have two different "faces" - one spatially and the other business rule-focused - reflecting distinct and poorly connected underlying data.

Spatial and business-focused data stores usually differ in that spatial-centric data models originate in geometric objects, whereas business-focused data stores use models that originate in business objects. Which model is "best" depends on the target environment in terms of business needs, available data and data management resources.

In a typical GIS data model, the primary database table structures are tables that collect data associated with a particular type of geometry such as lines, points or polygons.

Business object data is then connected to these geometric objects. This facilitates searches that start with the spatial data and across different types of business object at the expense of flexibility in drilling down into the business objects.

The complexity in this type of database often evolves in geometric rather than business object areas. For example, tables for polygon data may be segmented to differentiate high resolution polygons for fine-scale mapping from coarse polygons used for regional maps.

In a business-centric model, the primary database table structures are tables that collect data associated with particular E&P objects such as wells, surveys or leases.

Spatial information is stored with each explicit data type. This facilitates querying along the lines of the business objects but makes it harder to do spatial queries across all types of business object.

The complexity in this type of database always evolves along business lines. For example, singular relationships become multiple ones and "current" data becomes time-variant.

Simple spatial stores

Business-centric data models are ideal for systematic long-term automation of business processes, as they provide a well-defined and integral knowledge base. Unfortunately, it is not always practical to thoroughly implement a business-centric data model across all of a company's E&P information.

A business-centric data model needs to evolve to more complexity to meet higher levels of automation. With each increase in complexity the requirements in quality of input are made more stringent, and the accessibility of the information to simple ad hoc querying is reduced. This increases the cost of keeping data in the model.

Most companies will endeavor to keep well-structured information on the E&P activities that they operate, for example in a detailed well database for company-operated wells. It would be uncommon, however, to find them willing to invest the same effort into the management of purchased well data for historical information on blocks that they have an incidental interest in.

A simpler data store is therefore required for such "non-critical" data which covers a wide range of information, often acquired quickly, and for which detailed data analysis would be too expensive.

Spatially enabled databases

Disparate data management for spatial- and business-centric models would not be necessary if we had good implementations of spatially enabled databases.

Most existing spatial data management projects have their origins 5 or more years ago and have not had access to the level of software refinement nor to the computational power that is available today. The way in which we can harness these technologies now allows us to revisit the idea of a closer link between spatial- and business-centric data management strategies.

PPDM well header

To illustrate the application of spatial database technology in the business-centric PPDM data model, we can look at an extension of the PPDM data model that replicates bottomhole and surface locations in a spatially indexed geometry table WELL_GEOMS. This is fully synchronized with the master location stored in WELL_NODE through PL/SQL triggers and procedures:

This business-related structure has a number of advantages over its spatially centric equivalent. In particular, we can accommodate as many different geometries associated with a well on the well geometry row for the well, not only to allow for surface and bottomhole but also to potentially allow "native and database coordinate reference systems.

Practical combination

Having spatially enabled well data in the business side of a database will provide advantages in applications that understand and use the underlying Oracle/Spatial technology but will have little benefit to the established GIS applications and workflows in an organization.

With current ArcSDE and Oracle technology we can now bridge the gap by hosting ESRI's ArcSDE layers in an Oracle Spatial context that is understood as a spatially indexed layer by both ESRI and ORACLE products.

Starting with the WELL_GEOM table added to our well header model, we create individual materialized views of the well for each geometry type required (for example surface hole, bottomhole) and register those views as layers in ArcSDE.

This allows us to see the well location information from the business-centric PPDM model in the spatially elegant ArcSDE World, along with any additional information that we might chose to include in the materialized view such as 'WELL_SDO_SH.'

As a result, we can do spatial queries using ArcSDE technology between the well location and existing ArcSDE hosted polygon data, and more complex structured queries from the well geometries through the full PPDM data model.

We can extend this concept to host other ArcSDE layers in an Oracle spatial environment. When we do that, the spatial data from the ArcSDE layers becomes visible to Oracle spatial technology, allowing spatial queries such as proximities and enclosures to be undertaken either in the Oracle or the ArcSDE realm.

Reaping the benefits

By enabling the hosting of ArcSDE information in Oracle spatial, we open up a path for better integration of spatial and structured approaches to data management. Most importantly, we are more likely to be able to maintain spatial data in fewer versions, and hence with more confidence. In addition, we achieve the sought - after functionality of being able to apply structured queries on spatial data, and spatial queries on structured data.

We'll be able to answer both the structurally dependent query - "Find the wells intersecting the TA-5 sand with more than 10 ft (3.05 m) of pay" - and the spatially dependent query - "Report the distance of seismic lines shot over the agricultural territory controlled by CowCorp Ltd." - from a single consistent data suite.

In this example a Petrosys map view is displaying ArcSDE lease information and Oracle Spatial PPDM well locations, with a spatial query used to highlight wells in a specific lease.

Conclusions

Spatial knowledge doesn't have an absolute and unique representation, and there is a gap in the spatial- and business-centric ways of dealing with this. There are sound business reasons for following a diverse data management path that incorporates both generic spatial and detailed business-oriented data models. Recent improvements in spatial database technology have enabled the construction of databases that effectively provide the advantages of both spatial and business-centric access to the same information.