Spatial Data Constructs for Multi-Dimensional Transportation GIS Applications

This paper presents the critical spatial data constructs necessary for transportation applications that use and share data expressed in one to four dimensions. These spatial data constructs include spatial measurement and storage, spatial referencing and metadata, error propagation and transformation. The theoretical foundation and the characteristics for these constructs are provided along with solutions through a transportation-based multi-dimensional data model. The Multi-Dimensional Location Referencing System data model manages spatial and temporal data thereby allowing organizations to implement improved solutions for transportation systems using advanced spatial technologies. Koncz, Adams and Vonderohe 3 INTRODUCTION As a result of increased efficiency in data collection efforts, transportation agencies are collecting transportation-based data such as pavement conditions and incidents using spatial technologies via videos, digital cameras, on-board computers and GPS receivers. Prior to these technologies, these agencies collected, and still do, business data based on linear location referencing methods. A linear referencing method is a mechanism for finding and stating the location of an unknown point along a network by referencing it to a known point (1). Examples of linear referencing methods include milepoint, reference post, and engineering stationing. The increased use of these spatial technologies in transportation agencies has resulted in dimensional integration issues. The availability of highly-accurate, three-dimensional measurements, through technologies such as GPS, makes it possible to calculate locations and distances more easily than with some linear location referencing methods currently in use. In fact, several researchers (e.g. (2)) argue that with the advent of GPS and with the accuracies that GPS is beginning to offer, three-dimensional GPS will replace the need for linear location referencing. However, once three-dimensional data are collected, transportation agencies are left with determining relationships of these data to existing linear business data referenced to associated cartographic centerlines in their spatial databases. Moreover, much of the analysis that supports transportation business functions is performed in the linear domain. Issues regarding storage of multi-dimensional data, inherent error in both GPS and linear measurements, and transformation of data from linear and non-linear coordinate systems become valid with these spatial technologies. Given that many states are recognizing the practical applications in using these spatial technologies in the field, and many state experts are exploring these options daily, a spatial data framework is needed that will allow the integration of multi-dimensional transportation-based data. National Cooperative Highway Research Program Project 20-27(3) was initiated to develop a framework that encompasses all elements and abstractions for the multidimensional representation and integration of transportation-based entities. One of the goals of this project was the development of an improved transportation-based multi-dimensional location referencing system data model based on consensus derived functional requirements. The consensus process involved identifying the functional needs of transportation stakeholders, breaking these needs down to spatial and temporal characteristics, preparing functional specifications based on these characteristics and needs, and providing implementation strategies. This paper shows how a Multi-Dimensional Multi-Modal Location Referencing System (MDLRS) data model provides the spatial data framework necessary for integration of multidimensional data. This paper presents foundational functional requirements, a discussion of spatial data constructs needed for these functional requirements, and how a MDLRS data model satisfies those constructs. MDLRS FUNCTIONAL REQUIREMENTS Ten core functional requirements were synthesized from the results of the NCHRP 20-27(3) Workshop on Functional Specifications for Multimodal, Multidimensional Transportation Location Referencing Systems held in December 1998. These core functional requirements form Koncz, Adams and Vonderohe 4 the essence of a data model that accommodates a comprehensive transportation location referencing system. The ten core functional requirements are (3): 1. Spatial/Temporal Referencing Methods: Support for locate, place and position processes for objects and events in three dimensions and time. 2. Temporal Referencing System/Temporal Datum: Accommodation for a temporal datum that relates the database representation to the real world and provides the domain for transformations among temporal referencing methods. 3. Temporal Topology / Latency: Support for temporal relationships among objects and events and latency of events (i.e. difference in time between scheduled and actual events occurring at a particular location). 4. Historical Databases: Support for regeneration of object and network states over time, and maintenance of network event history. 5. Dynamics: Support for navigation of objects, in near real-time and contingent upon various criteria, along a traversal in a transportation network. 6. Transformation of Data Sets: Support for transformation between linear, nonlinear and temporal referencing methods without loss of spatial / temporal accuracy, precision and resolution. 7. Multiple Cartographic / Spatial Topological Representations: Support for multiple cartographic and topological representations at both the same and varying levels of generalization of transportation objects. 8. Resolution: Support for display and analysis of objects and events at multiple spatial and temporal resolutions. 9. Object-Level Metadata: Storage and expressions of object-level metadata to guide general data use. 10. Accuracy and Error Propagation: Support for association of error measures with spatial / temporal data at the object-level and propagation of those errors through analytical processes. Functional requirements one, six, eight, nine and ten pertain to spatial aspects of phenomena and their interactions. These spatial-based functional requirements form a natural progression. Multi-dimensional spatial data require a framework for storage in various resolutions and formats. Once multi-dimensional data are stored, information is needed on the data’s background (i.e. metadata), which provides limitations on its use and transfer. One of the critical elements of metadata is positional accuracy. The positional accuracy of a spatial entity is described through error measures and propagation of those measures. With the ability to store data in different dimensions, in different resolutions, in different coordinate systems, along with error measures, intelligent transformation of various types of data can occur. For a review of how the MDLRS data model satisfies the temporal-based functional requirements, the reader is referred to (3). SPATIAL DATA CONSTRUCTS This section provides the motivation for spatial data constructs needed to satisfy each of the spatially based functional requirements. The spatial data constructs required for spatial data storage, spatial metadata, spatial error propagation and spatial data transformation are described below. A review of spatial data representation aspects applicable to transportation stakeholders can be found in (4). Koncz, Adams and Vonderohe 5 Essential Spatial Data Constructs for Storage The spatial data constructs needed for spatial data storage include spatial referencing systems, composed of spatial referencing methods and reference objects, coordinates and measurements. This section describes these data constructs. In using geographic information, it is important to consider spatial measurement concepts. Spatial measurement requires a set of geometric assumptions to create a spatial reference system (SRS). A SRS provides a mechanism to: 1) situate measurements on a geometric body, such as the earth; 2) establish a point of origin and orientation of reference axes, and 3) provide geometric meaning for measurements and units of measure (5). A spatial referencing system is composed of spatial referencing methods, reference objects and a datum. A spatial reference method (SRM) can be thought of as a mechanism for finding and stating the location of an unknown point by referencing it to a known point (adapted from (6)). An example spatial reference method is a state plane grid based on a Transverse Mercator projection. Reference objects (RFO) are objects whose locations are known and from which measurements are made in the real world to determine unknown locations of other objects and provide relationships needed to link a SRM to a SRS datum. One type of SRS is a geodetic spatial referencing system (i.e. a type of horizontal SRS) based on an ellipsoid definition (i.e. datum object), projections (i.e. spatial referencing methods) and control stations (i.e. reference objects) whose locations are known in Latitude / Longitude and the projection coordinates. The locations of objects and events can be expressed in five main categories of spatial referencing systems: vertical, geocentric, horizontal, cadastral, and linear (7). Table 1 presents the Datum Objects and Reference Objects for each SRS and Figure 1 illustrates several of these systems. These systems are described below (8). • Geocentric: a three-dimensional orthogonal coordinate system having a defined orientation and an origin near the center of the earth (e.g. WGS84 (GPS)). • Horizontal: a two-dimensional angular coordinate (Latitude, Longitude) system based on an ellipsoid having a defined orientation, and an origin near the center of the earth or a twodimensional orthogonal coordinate (X, Y or Easting, Northing) system having a defined orientation, and a mathematical (or projection) relationship with the ellipsoid. • Cadastral: a coordinate system based on legal locations using the acre as a basic unit, with an origin on the surface of the earth (9). An example of a cadastral coordinate system is the Public Land Survey System (PLSS). • Linear: a one-dimensional coordinate (identifier, offset) system identifying a location by reference to a segment of a linear geographic feature (such as a roadway) and distance from some point along that segment (10). Examples of linear coordinate systems include mileposts, and civic addresses. • Vertical: a one-dimensional coordinate system, representing the position (elevation) of an entity above or below a surface such as the geoid. Multidimensional spatial (and temporal) coordinates must be able to express a variety of locational representations with dimensional requirements and uncertainty estimates. Functional requirement one, as presented earlier, expresses this need to represent locations in both threeKoncz, Adams and Vonderohe 6 dimensional space and time. Dimensional requirements include locating objects and events to a roadway longitudinally in the proper relative order with respect to other objects and events and to specify laterally the appropriate lane of a highway. While it is common for highway agencies to maintain longitudinal location coordinates (e.g. as illustrated by strip maps), lateral offset measurements are often not maintained. Additionally, vertical locations should be sufficient to place an event or object on the proper feature, thereby allowing functions such as separating planar-coincident facilities (e.g., road-on-bridge versus road-under-bridge), necessary for incident management. In typical geo-standards, points are associated with coordinates either as attributes or as separate objects. In actuality, coordinates are artifacts or derived values based on imperfect measurements to reference objects. Some coordinates are not generated by ground measurements (e.g., digitizing and COGO) but can, nevertheless, contain error. A distinct advantage of storing measurements in addition to coordinates is the ability to recompute coordinates when there is a change in the datum (i.e., if the datum changes and one only has coordinates, errors are introduced when transforming coordinates to the new datum). Essential Constructs for Spatial Data Management Spatial data constructs are needed to provide information on the use and transfer of spatial data. These constructs include metadata and error propagation at the feature level. This section describes these data constructs. Stakeholders are interested in current, reliable, accessible, and understandable data. It is often necessary to rely on data obtained from secondary sources, originally collected for purposes other than present applications. Since various stakeholder data are in demand, they will be traded, compiled, transferred, transformed, bought, sold, reclassified, and reinterpreted. Data may be used for many purposes other than that for which they were originally collected (11). Information that describes content, quality, condition and other appropriate characteristics of data that influence their validity may or may not be conveyed through these transactions (12). Additionally, problems arise in many cases where there are multiple sources for the data, each with a separate error. When individuals review stakeholder data without knowing how the data were generated, they may misinterpret the information or view the data with a false sense of completeness, currency, or numeric accuracy (13). Two key means of communicating the background of spatial data sets to stakeholders are through metadata and error propagation.