Oregon Metro’s regional traffic count program underwent a big change in 2024. AI processing of videos collected by our vendor resulted in a price decrease. This allowed us to class count, determining truck volumes, at every count location. It also allowed us, for a small additional charge, to order bike counts, for which the vendor has an AI procedure. But we noticed differences compared to other data sources, so we sought raw video for select time periods and locations to validate the data for use in model validation. This presentation will share the lessons we learned for those who may consider a similar approach.

The General Modeling Network Specification (GMNS) [1] defines a new data format for routable roadway network files. This research describes findings while adapting MATSim networks to this specification.

The GMNS spec is thoughtful and mostly complete. A MATSim-to-GMNS converter was written along with a web-based GMNS viewer. Issues did come up while applying this new data format, and the pros and cons found in the GMNS specification will be discussed.

There has been significant work done in the space of travel demand model network development including the use of open-source data for spatial alignments and leveraging data from state linear reference systems over the past decade. This body of work has in a common is a singular focus or application for that development effort even when they may share geographic coverage. Further models that are used for multiresolution model often have networks developed independently. For example, in many areas, the statewide model will have an independent network system and project coding from an MPO that is included in the statewide model aera. This independent approach creates redundant work as well as the potential for inconsistent project assumptions when evaluating investments. This presentation will discuss the benefits of developing a statewide multimodal transportation network that harmonizes the network features and characteristics of Ohio’s MPO models and statewide model network into a single database and highlight the opportunities this creates for efficiency, accuracy, and consistency. Currently Ohio Department of Transportation maintains a statewide model (OSWM) and seventeen MPO models. The state has made significant investments to its modeling tools for the past several decades. It has one of the most robust tour-based multimodal statewide model platforms, and its MPO models range in complexity from the activity-based 3C models to trip based models deployed in the smaller areas. The challenge Ohio DOT and its MPOs face is diminishing staff resources to adequately maintain the necessary input datasets to the high standard that is required for model applications. These include long range planning for the state and regions, project prioritization and project forecasting.

 With the use of a single network database that both supports the SWM as well as generate MPO model input networks, the state can better manage model application uses while ensuring project consistency and network accuracy within the constraints of less local staff resources at the MPOs. Such an integrated approach to a network database allows the stakeholders across the state to maintain and update their own model specific datasets from current up-to-date network data and provide them the opportunity to publish their own project coding back to the integrated network. This has the potential to not only facilitate local and statewide transportation planning and modeling, but to also facilitate the prioritization process by enhancing project coding consistency in Ohio.

This presentation will not only discuss benefits such as enhanced scalability of networks, alleviating the burden on staff to code networks, and increasing consistency across the state and its MPOs, but also address challenges such as dealing with different attribute naming conventions and needs, variations in the required level-of-detail of the network, and developing an editing environment that is accessible to the user base.

This presentation addresses the critical challenge of network connectivity issues in travel demand model (TDM) development, where discrepancies between modeled traffic volumes and observed traffic counts often indicate underlying network deficiencies rather than simply requiring volume adjustments. We present practical techniques for identifying and resolving complex network connectivity problems within TDMs, focusing on both the demand side (traffic analysis zone systems) and the supply side (network link configuration). 

Our methodology leverages established principles from the trip assignment stage of the four-step TDM process, utilizing assigned traffic volumes to diagnose and rectify network inadequacies. Specifically, we analyze network links with zero assigned volumes to identify underlying issues, including an insufficient number of zone connectors, inaccurate link representation, and incorrect network link directionality. We propose innovative solutions for reconciling discrepancies between the network representation and real-world infrastructure, including techniques for updating zone connectors, adjusting link length attributes, and ensuring accurate network topology.

The Arizona TDM network, constructed using the linear referencing system (LRS), serves as a compelling case study. This network provides a valuable platform for demonstrating the inherent differences between true shape GIS networks (typically used for visualization and spatial analysis) and hybrid TDM networks (optimized for transportation modeling). By examining these differences, we underscore the importance of rigorous network coding and validation (based on understanding of trip assignment theory) for generating realistic and reliable travel demand forecasts. The presentation will showcase the TDM network analysis results, including visualizations of network connectivity issues, and network performance metrics, demonstrating the impact of these issues on the traffic flow assignment process. We will also present the outcomes of applying our proposed solutions, illustrating improvements in network flexibility and efficiency.