Designing Geodatabases for Transportation. J. Allison Butler

the two pieces of pavement become one. For the ramp and mainline centerlines to intersect at this location, the ramp centerline needs to turn toward the main centerline and intersect at the appropriate milelog location at a right angle. The resulting appearance may look strange, but it ensures that the intersection will occur at the proper location from the perspective of the limited-access highway’s linear measures. If you are also providing carriageways, they can continue along the path of the ramp as it merges with the highway pavement.

      Figure 4.6 Logical centerlines Logical centerlines need to intersect at the location that makes sense from the perspective of connectivity and the linear datum. In this example, the agency rule says that the point of intersection, for which the route measure is defined, is the end of the physical gore; i.e., where the pavement of the ramp merges with that of the mainline. A ramp carriageway need not be constrained by that definition, since the purpose of a carriageway is to provide a good map appearance at a large scale.

       Intersections

      An intersection is normally visualized as the location where two roads cross at grade. You may also want to look at using a number of intersection subtypes as a way of managing behaviors and symbology. For example, an interchange could be either a way to manage a collection of roads and at-grade intersections or just a big intersection with internal turns for pathfinding.

      A railroad grade crossing is another intersection subtype, one between a railroad and a road. You may want to restrict the classification of this kind of intersection to those places where a track crosses a road through topology rules. Such a function can be easier to implement if you create a railroad grade crossing subtype for intersections.

      Figure 4.7 Intersection subtypes Three of the four basic intersection subtypes are shown here. An at-grade intersection is the typical place where two or more routes meet. A railroad grade crossing is where a route crosses a railroad. If there are multiple tracks, you may want to use a multipart point feature class. The third subtype shown here is the access point, which represents a location on a route where a nonmapped facility connects to a mapped facility. Access points include shopping center driveways, recreational trail crossings, and anywhere you have intersection-like elements with characteristics to store, such as traffic signals and crashes.

      There are also places that look like intersections where only one road seems to exist. These are access point intersections, and they occur where trails cross a road or a major driveway provides access to the road system. Like at-grade intersections, you will likely want to retain information about these locations, such as traffic signals and traffic crashes.

      Using intersection subtypes will provide you with the ability to check cardinality to ensure that it meets your expectations. For example, you will need at least two intersecting roads for an at-grade intersection, one road and at least one railroad track for a railroad grade crossing, and only one road for an access point. An interchange will need to include at least one road classified as a limited-access highway.

       Realignment

      The way you segment centerlines will affect future editing workloads. One of the more common issues you will deal with is facility realignments. A typical approach to building a geodatabase for road features is to identify the entities to be included by name and route number. You may additionally subdivide these entities at county lines or district boundaries to reduce the extent of changes or to coincide with your data maintenance organizational structure. This is all fine and good, but once the features are created, you need to break the symmetry of entities and features and treat name or route number as just another attribute. If you try to preserve the symmetry, you will have a lot of extra work when a road is realigned, particularly if you are using a linear location referencing method. The problem is that realignment causes the total length of the facility to change, resulting in a ripple effect through all the downstream locations. You either have to revise all their measure values or create some sort of reconciliation structure in the measure values to deal with the missing or duplicative values within the area of realignment. You also may lose the ability to properly map event data applicable in datasets tied to old measure values.

      A better choice is to treat the new section as a new facility. Add a facility status field to the database and give it one of three values: active, retired, and replaced. (See chapter 6.) You would leave the old alignment in the dataset; just list its status as retired. If you need to display historical data on the former alignment, the feature and its linear measures are still in place to do the job. You also do not need to change any downstream data. All the data about the new section is also new, so there is no lineage issue for the facility identifier and linear measures used by it. If you want to see the present state of the system, just select all the active facilities. You can select by date and status to produce various time-stamped views of the geodatabase.

      Figure 4.8 A strategy for realignments One of the traditionally difficult aspects of a transportation dataset based on a route-milelog linear referencing method (LRM) is facility realignments that change the length of the facility at a midpoint. Such a change, if accomplished through traditional methods, will reshape the affected geometry and recalculate the measure for downstream events. A better way that avoids modifying all the event records is to make a new feature for the realigned portion and abandon the replaced segment. A status attribute identifies active and retired route segments. This approach has the benefit of retaining the original geometry and measures for time-based analyses.

       Segmentation methods

      Several references have been made to the traditional transportation dataset structure of short segments with repeating attributes. This is a perfectly valid way to go if you have a manageable number of segments and relatively few changes to the system other than added facilities, which is the case for most local governments. Once built, roads are rarely removed. Subdivisions add new mileage. Annexation may change jurisdiction for a road, but it does not eliminate the road itself. Most capacity projects will widen a road, not change its location. All in all, this is a relatively stable situation where editing means adding new roads as they are opened to traffic.

      The least demanding way to accommodate the needs of smaller datasets is to break roads and other linear facilities at intersections, but not to actually create intersection features. The resulting “independent” segments work fine for basic mapping needs and geocoding. The viewer will not be able to discern that a given line segment is composed of smaller pieces.

      Figure 4.9 Independent segments The simplest approach to making a transport dataset is to create independent segments representing linear portions of the total system. A typical way to do this is to segment longer facilities at intersections. Each segment gets the same set of attributes so that values extending across segment termini are duplicated for all segments to which the value applies. You can use segment attributes to determine line symbology, or create separate map layers to represent each kind of facility.

      Some database designers may argue that this simple design is inefficient. Although you will have to duplicate much data for contiguous segments of the same facility, this overhead only applies as a facility is added to the dataset. This initial workload may be offset by avoiding more complicated structures that impose their own overhead. You also do not need perfect geometry with no overlaps or gaps unless you plan to build a navigable network.