Development of a Multi-Resolution Network to Support Statewide Modeling and Project Analysis
Corresponding Author: Ashley Tracy, Whitman, Requardt, and Associates
Presented By: Scott Thompson-Graves & Jonathan Avner, Whitman, Requardt & Associates, LLP
Abstract
Ultra-high multiple resolution networks were prepared for both DelDOT and Maryland SHA from GIS Centerline files. The main purpose of the DelDOT multi-resolution network was to provide a fine-grained highly detailed network containing sidewalk and trail information for the introduction of quantitative bicycle and pedestrian forecasts to the Peninsula Travel Demand Model (TDM), as well as allow the calculation of health measures of effectiveness and increases in efficiency for DelDOT staff who utilize the model for traffic impact studies. The purpose of the SHA multi-resolution network is to support activity-based modeling and transit planning studies, and allows for further exploration of uses of the travel demand model.
This presentation will focus on best practices and lessons learned for developing multi-resolution networks, based on experience with DelDOT and Maryland SHA. It will include discussion of the efforts to build a multi-resolution network, appropriate data sources to use, and challenges encountered at each step of the process; such as network routability fixes, the customized GIS Network Builder tool, how to resolve junction control and transit line files, network attribution, composite zone structure and related disaggregated demographic data for each level, centroid connector generation, software limitations, run time implications, and data maintenance.
A network at varying levels of resolution offers planners the flexibility to expand the scope of travel demand model applications, allowing users of the model to generate a hybrid network that balances the ability to generate desired outputs with reasonable run times for a given application. Multi-resolution networks are relatively fresh in practice, and agencies are adapting by developing networks and model processes that should only improve over time as more robust datasets become available.