Market Drivers And Maturing Technologies Trigger An Explosion In Geographic Data Acquisition And Analysis.
When you stop to consider it, most information - some 90 percent, according to MapInfo Corp. - has some geographic attributes. Analysts in business and government have long struggled with making effective use of geographic data. In this decade, a number of technologies have converged to make Geographic Information Systems (GIS) effective tools for leveraging information with a geographic content. According to a DataQuest (San Jose, Calif.) analysis reported in ComputerWorld (June 10, 1996), "incr eased corporate use of GIS technology to improve customer service and cut costs is expected to help propel the overall market from $862 million last year to $1.7 billion in 2000." This growth is attracting attention from market leaders in the database an d CAD worlds, as well as from traditional GIS vendors and a slew of startups aiming to capitalize on the incorporation of mapping applications into the expanding component and Web-based development infrastructure.
Geographic-intensive industries such as public utilities, transportation, environmental management, and retail marketing are driving the explosive growth of the industry. Of course, all socio-demographic information has a geographic component, as does in formation about land and its uses. Consider, too, information about production schedules in a manufacturing environment. It may be helpful to take into account the location of component suppliers (specifically their distance from the assembly facility). All information about a business's sales contains data about billing and shipping addresses. Market analysts can tap into this information to plan targeted promotions and make location decisions.
The maturation of a number of technologies is also contributing to the explosion of GIS applications. Of course, the accelerating decrease in the cost of fast computers since the end of the 1980s enables maps to be rendered in realtime. Remote sensing is now one of the primary means of collecting geographic data. Speaking of data, the ready availability of digital demographic data collected in the 1990 census has had a major impact on the industry. Finally, software has matured considerably, in its anal ytical power, ease of use, and in the breadth of applications to which it can be applied.
Traditionally, GIS has focused on engineering applications; there is no question that GIS is the core automation technology in business areas such as resource extraction, facilities management in the utility and telecommunications industries, transportat ion and logistics, and environmental management. With technological advances and increasing market demand for competitive business tools, mapping software is emerging as the next hot business tool. Many businesses are turning to mapping software - tools that plot pertinent data and display it in easy-to-comprehend maps - to improve their ability to analyze pertinent market and corporate data and arrive at fast, effective business decisions. Consider that these goals sound suspiciously like the objective s that data warehouse developers are driving towards.
Market analysis and presentation systems are a giant leap in the evolution of GIS technology. Affordable, easy-to-use GIS products provide sales and marketing professionals with a variety of ways to target qualified customers, optimize sales planning and activities, visually plot key demographic data, and make intelligent marketing decisions. Professionals responsible for market planning and plant site location decisions are quickly discovering the advantages of using GIS. An analyst might map demograph ic characteristics of potential customers within target market areas (see Figure 1), and can plot average distance from concentrations of probable customers to existing business locations. With data on which products are purchased in specific neighborhoods (based on automobile registration or store credit card information, for instance), it becomes easy to identify new retail sites with a high confidence in their success. (See Figure 2.)
Because vector files require more graphic horsepower and map maintenance, they have largely been the format of choice for engineering-oriented products. Desktop PC products used primarily for business analysis have employed the raster format, which requi res less maintenance as well as hardware that is more in line with the standard PC infrastructure. The maintenance burden measures the ease of creating and maintaining digital maps. Most engineering maps have been traced from the analog versions. Digitiz ing maps into vector format has kept a lot of people busy tracing over a digitizing table during the past ten years; scanners capable of dealing with the large size or high resolution of these maps have been prohibitively expensive. Scanning them pieceme al into raster format is more feasible, but that process loses the detail necessary for engineering applications. As you might expect, digitizing maps is an imperfect process. As a result, many of the products we will discuss later include sophisticated error correcting utilities.
Many GIS applications build on an aerial photograph of the subject area. Here in the San Francisco Bay Area, the BADGER (Bay Area Digital GeoResource, at http://badger.parl.com) project is building a geographic data inventory that it will publish over th e Internet. BADGER is starting out by creating a seamless photograph of the area at a resolution of below seven meters. Modern aerial photography is better described as remote sensing because it uses a wider spectrum of the electromagnetic spectrum to cr eate a rich variety of digital images.
Another technology developed for defense purposes, Global Positioning Satellites (GPS) provide a way to specify location extremely precisely; field mapping technicians typically employ GPS receivers. A key step in creating a valid digital map is georefer encing, or synchronizing the data to a standard coordinate system. This process is typically a matter of identifying several reference points whose true geographic coordinates (such as latitude or longitude; this is where GPS plays an important role) are known. From known coordinates, scale, and measurements on the map, software can calculate the coordinates of all of the features on the map. Another aspect of mapping that we carry forward from the prehistory before computers is projection of the curved surface of the earth onto a two-dimensional coordinate system. Creating such a two-dimensional projection implies that any map of a large area is bound to be distorted to some extent. Some software products give digital map makers control over where to concentrate the distortion imposed by both projection and the digitization process; a feature called "rubber sheeting" lets the user fix specific positions on the map to geographic coordinates and stretch the balance of the map to fit the area in questio n.
GIS lets you integrate data from numerous sources, including tabular or statistical information, addresses, data from CAD systems, sound, and even video images. Associating attributes of interest with locational information enables people to view them in a way that leads to a higher level of understanding. GIS products vary in the way that they relate map features to attributes. Traditionally these linkages have been proprietary, as have GIS file formats. Devised during the past decade, GIS products emp loyed data structures that incorporated both geographic and attribute data in order to optimize performance on available hardware. Recently, we have observed increasing reliance on using common desktop or RDBMS databases (we'll visit some of the latter b elow) and creating tables with explicit and open geographic keys, which enhances the interoperability of both the data elements and map files.
GIS tools support several relatively standard forms of analysis. Much of the analysis involves overlaying coverages with differing themes to visualize the relationship of one type of attribute with another. Sometimes simply seeing two or three different data layers simultaneously reveals previously hidden relationships. For example, a map of income distribution overlaid on a map of actual customer locations presents customer income distribution in an easy-to-understand visual display. GIS users employ p roximity analysis to measure distance as a straight line or via a networked path, such as a street network. "Contiguity analysis" analyzes adjacency, which is useful in siting and environmental contexts. For instance, you might perform a contiguity analy sis to identify the stores in a shopping center as well as the income levels of potential customers who live nearby. Boundary definition lets the analyst define regions that contain specified populations. For instance, a school district might use this to ol to redefine school assignments so that student populations are distributed to match each school's ideal capacity.
Surface trend analysis maps the change in a dependent variable as distance from a specific location or feature increases (for example, how wildlife populations decrease with distance from a protected habitat). Network analysis addresses routing problems and is used to plan sales routes or compare the efficiency of different delivery methods. Using these analysis mechanisms together lets analysts efficiently evaluate different scenarios and plans, and it provides an effective means of communicating findi ngs in the familiar format of a map.
Oracle is focusing its Oracle7 Spatial Data Option on two problems: how to efficiently store, access, and manage different types of data in a single database, and how to improve performance for these very large databases holding hundreds of gigabytes of spatial data. Oracle markets the Oracle7 Spatial Data Option (SDO) as an option to the Oracle7 Server that is fully compatible with other Oracle7 options, including the procedural and parallel query options. Oracle anticipates that implementers managing very large databases containing geographic content will not typically have had much experience developing geographic applications. As a result, they have designed technology to automate optimization of such data.
Within the context of the Oracle RDBMS, the SDO provides tools to manage spatial data in a way that preserves its inherent spatial organization. To do so, SDO introduces a new RDBMS data type: Helical Hyperspatial CODE (HHCODE). HHCODE employs an encodin g and linearization technique for combining data of two or more dimensions into a single value that represents the intersection of all of the desired dimensions. Once encoded, SDO physically stores data in a predictable manner; the greater the proximal r elationship between data records, the closer together they are stored in the database. Because the data aggregates, or clusters, based on its defined dimensions, this strategy yields rapid access to relevant data by quickly eliminating all data that fall s outside the dimensional boundaries of a locational query. SDO provides HHCODE as an application development building block, supplying SQL and PL/SQL extensions for creating and decoding HHCODE data and performing spatial operations.
The HHCODE strategy optimizes performance by eliminating the dependency on indexes, because the data is the index. According to Oracle, this strategy significantly reduces load and query processing overhead and yields near-linear access performance as sp atial data volume grows. SDO takes advantage of the HHCODE structure to sort and store spatial data in multiple tables, called partitions, that subdivide dynamically. When a table containing data for a specific region becomes too dense to maintain fast a ccess times, it automatically subdivides into multiple partitions based on its spatial organization.
Informix is using Illustra's Object Relational DBMS and its DataBlade technology as a lever for optimizing spatial data management. DataBlade modules define new data structures, functions to manipulate them, and data access methods customized for the dat a type. Illustra presents the fact that the DataBlade technology migrates application intelligence from client libraries to the server as a key benefit. When applied to GIS applications, performing spatial processing on the server (and returning only th e results to the client) represents a fundamental shift from the historically client-centric GIS application architecture. Illustra offers both 2D and 3D Spatial DataBlades. The 2D Spatial DataBlade adds 10 data types that describe common planar geometri c shapes and polygons, together with over 200 functions that allow object creation, comparison, manipulation, and queries. The 3D Spatial DataBlade adds 18 data types such as circles, 3D boxes, lines, polygon mesh, and surfaces, along with over 1,000 fun ctions to manipulate them.
Illustra is keeping abreast of SQL3 standards efforts and already permits formulation of SQL queries that operate on the custom data types, with functions added by a DataBlade. This capability lets relational database application developers leverage thei r existing SQL expertise while gaining the benefits of object oriented technology. When working with spatial data using one of the DataBlades, you can formulate queries that include functions specifying location, distance, and intersection relationships; you can also call Spatial DataBlade functions directly via the server's API. Another benefit of applying DataBlade technology to spatial data is its support for data type specific indexing schemes. Illustra applies its R-Tree indexing scheme, which buil ds a three-dimensional space of index nodes comparable to the two-dimensional space created under a B-Tree scheme, to the problem of obtaining high performance in large geographic database queries.
IBM's GIS offers a range of GIS products and tools. IBM built its geoManager product on DB2 and positions it as a spatial GIS solution for the storage of maps, schematics, documents, and associated attribute information. IBM structured geoManager to opti mize enterprise record-keeping of asset location and event details, optimizing it for facilities management applications. IBM customers can deploy geoManager on their IBM mainframes or RS/6000 workstations. Companion geoInterface packages for OS/2 and Wi ndows workstations provide APIs for construction of GIS applications accessing the data in geoManager. The IBM GIS Database Interface for AIX provides an interface between ESRI's Arc/Info and ArcView products and the IBM DB2 database family. This interfa ce lets the ESRI products use the DB2 tabular data found in GIS systems.
Computer Associates CA-OpenIngres product comes with the Object Management Extension (OME), a technology that is externally similar to the Illustra DataBlade technology. OME supports defining data types, and operations can be performed on either standard or user-defined data types. These functions are compiled and stored in the DBMS server and become client-specific extensions to the SQL syntax. CA has used OME to implement its spatial data support, the Spatial Object Library, which contains geographic data types and new geometric SQL functions.
Although Sybase has not yet incorporated support for geographic data into its SQL Server RDBMS, the Vision International division of Autometric Inc. markets a Spatial Query Server (SQS) built on the Sybase Open Server framework that appears as a tightly integrated feature of the Sybase Server. Applications interact with the SQS by generating requests in GeoSQL, the SQS proprietary extension to Sybase's Transact SQL, which supports spatial data. SQS spatial data types are constructs that describe spatial entities on the earth's surface. Unlike traditional Sybase data types, GeoSQL spatial data type maps to several columns in the Sybase database table and/or to other auxiliary tables. SQS supports queries that include spatial operators such as intersect, inside, outside, beyond, and within as part of the where clause predicate.
Next month, I will tour much of the rest of the GIS industry, starting with a thorough assessment of ESRI and MapInfo's products. I will discuss several development components that make it relatively easy to integrate GIS capabilities into custom applica tions.
