Use of Scattergrams for Defining Network Subregions

It is often necessary to identify the links and / or the nodes of a subnetwork that corresponds to a well defined geographical area. Examples might be:

• create link types according to geographical subdivision;
• extract a subarea for detailed analysis such as link select and traversal studies;
• systematic change of the speed limit within a community.

As long as the considered subnetwork is identifiable by existing network attributes, it can easily be selected via the usual ``Enter: Link types or attributes'' dialog.

If no such link or node attribute exists, or if the purpose is precisely to create such an attribute, only the graphical information is available. How can we then use the polygon defining the considered region within the network plane in order to perform a systematic operation on only those nodes or links inside (or outside) the region?

The answer is rather simple and underlines once again the strength of the tool approach used within EMME/2: Perform a marking operation on a scattergram using X and Y network coordinates (variables XI, YI, or XJ, YJ). This kind of scattergram is almost a network plot, it shows the nodes or links at their correct location in the network plane. Since the scattergram provides a set of polygon operations for counting, eliminating and marking, these can now be used to mark all nodes or links within (or outside) a given region. Refer to section IV-2.43 of the users manual for details on these operation.

When working with this kind of ``network'' scattergrams, keep the following points in mind:

• Make sure that the X and Y range correspond to an aspect ratio of 4X:3Y, otherwise the network will look distorted. A simple way to achieve this is to directly use the window coordinates of a network plot as the X/Y value range.
• Do not use the linear regression feature, since this is not a statisti- cal analysis.
• Keep in mind that in the scattergram, all elements are represented as points in the X/Y plane. This means that links will be located at one of there endpoints, depending on whether XI,YI or XJ,YJ have been chosen as variables to plot against.
• Remember that the marking value can be changed interactively using the "v" command. This allows to mark more than one region on the same scattergram.

The example below shows how the subregion of the Winnipeg network was prepared for the computation of the example traversal matrix, which was shown on page 15ff of the last issue of EMME/2 News.

The region was outlined on a XI/YI node scattergram, using the "Polygon Mark" command sequence "P ... M", thus marking the nodes contained in the subregion by setting to 1 the node user data UI1 (which contained all zeros initially). The links may now be classified easily by accessing the normal subnetwork selection dialog, using the following criteria:

Links outside the subregion: ui1=0 and uj1=0

In-gate links: ui1=0 and uj1=1

Out-gate links: ui1=1 and uj1=0

Links inside the subregion: ui1=1 and uj1=1

The above selection criteria have then been used in the various modules, to compute the link select attribute, enter the gate information and punch the subarea network used for building the subarea data base.