To summarize the panel's comments in the form of recommendations for WSDOT's statewide model development process, this section will discuss the following topics for consideration:
The panel was given several funding scenarios including $500,000, $1 million, $2 million, $3 million, and $5 million or more, for developing a statewide model. The panelist identified $1 million as a good starting figure for a planning-level model to meet various policy and investment scenario analysis needs while also meeting substantive validation requirements. With this budget the model could include: a solid highway component and a transit network focusing on high capacity, intercity passenger travel. It was also noted that this $1 million estimate did not include data acquisition and also that $1 million be added to for the incorporation of a full truck component.
The panelists agreed that the first step in the model development process is creation of a list of statewide analytical needs and prioritization of this list based on which needs are immediate versus those that can be incorporated further down the line. Secondly, the panelists recommended that WSDOT identify a specific budget, schedule, and staff for the model development effort.
Oregon noted currently one FTE is dedicated to statewide model development and application. Ms. Knudson noted that this level of staff support is insufficient; indicating that staff budgeting might need to be higher than WSDOT currently anticipates.
Once budget, schedule, and staff have been clearly identified by WSDOT, the panelists suggested that WSDOT conduct a complete data inventory to identify all available data for application in the model and then to cross reference the available data to the identified analysis needs. After reviewing what data is available in relation to potential modeling needs, WSDOT can scope model development in phases relating to existing data and secondary data.
One panelist recommended that the initial phase of the model only take into account existing data sources, and then a second phase of the model incorporate secondary data. Another panelist suggested that WSDOT proceed with the initial model development without all of the necessary data but scope the data projects needed into the process.
The panelists agreed that properly phasing the model development process would be imperative to building a successful and logical statewide model. WSDOT should focus on the basic components of the model first, and then incrementally approach model advancements based on statewide priorities. It was noted that the state model should focus on creating a routable highway network that can assess 'high-level,' larger-scale movements, including intercity traffic. Hiring a consultant to execute a preliminary needs assessment and model planning exercise was also suggested.
The panelists also reiterated the importance that WSDOT identifies the following aspects of the model at the beginning of the model development process keeping the budget, schedule, and resources close in mind: software platform, resolution, and user expectations. Panelists agreed that managing expectations from statewide and regional planners will be critical to ensuring usable model results for all stakeholders.
Panelists also emphasized the importance of coordination with MPOs in the model development process. MPOs often rely on statewide models for external trip information. Additionally, MPO and statewide models often collaborate on land use and economic data projections. Maintaining communication with regional agencies will be critical in this process.
Finally, panelists noted the importance of recognizing the return on investment in model processes. Prior to incorporating more modules or functions to their model, the agency should first consider what they will get from these model modifications and if these adjustments are worth their cost. It was noted that simplicity is key in large-scale models, as is maintaining reasonable expectations and understanding the aspect of uncertainty in all forecasts.
The panelists recommend that model structure and features directly reflect the needs of the state in relation to available data and resources of the agency. Therefore, the panel emphasized the importance of the development of a routable highway network. Once this network is in place, WSDOT can expand their model scope. The following bullets describe the specific modeling recommendations:
Panelists stressed the importance of testing the model's sensitivity, particularly for extreme scenarios, in the development process to assess how critical assumptions drive forecast results.
Many existing data from national sources and the data collected by other states and MPOs within the state can be used to develop the initial version of the WSDOT statewide model. As identified in Section 6.2, the panelists recommended that WSDOT conduct an all-inclusive data assessment to determine what data is readily available to WSDOT, and from there the agency can identify any data gaps for potential further investment.
For highway side data, panelists recommended that WSDOT review all existing counts, for example the HPMS source counts, Automatic Traffic Recorders (ATRs), spot counts, and project counts, and prepare the budget to clean count data.
Mr. Cervenka of the Federal Transit Administration (FTA) recommended the use of the FHWA HERE data, a purchased national data set of roadway travel speeds based on probe vehicles,[10] for problem identification analysis, accessibility analysis, and model calibration/validation. Mr. Mahapatra suggested that WSDOT investigate the potential use of commercial data products such as the origin-destination tables available from cell phone providers.
Panelists suggested that long distance behavioral data would pose a challenge both in data and in trip distribution, but to look at existing HHTSs and possibly high speed rail studies for more information. If WSDOT would like to implement a HHTS, it was suggested that a rolling approach to conducting a HHTS may be appropriate for WSDOT, given this type of survey would allow WSDOT to create customized add-ons and prioritize the questionnaire. A household location choice survey was also recognized as a possibility for WSDOT application. It was determined that the initial statewide model can be developed without conducting surveys, but local surveys could be conducted later in the model development process.
Panelists identified passive data sources, like AirSage, transit card, Bluetooth, and other data sources for review for applicability. These sources may often be noisy and difficult to tell a story with, but they may provide useful data outlets once reviewed.
WSDOT should next identify and prioritize their agency needs related to statewide policy analysis. WSDOT should also identify a target budget, schedule, and resources for the statewide model, with a separate resource list for data. Using this information, WSDOT can develop the model in phases accordingly. A small, upfront consultant contract to design a blueprint for statewide model has proven successful for various states and was recommended as a possibility for WSDOT.
Sustained executive management support will be critical to the statewide model's success. Staff should be honest with management about what can be accomplished with a statewide model. MPO and regional transportation organizations should be involved throughout the entire model development process, as should state forces, consultants and universities and research institutions, as appropriate. State force needs to be involved every step of the way to ensure clear understanding of the model construct for ongoing model applications and continuous model improvements.
National Academy of Sciences. SHRP 2 Capacity Project C05: Understanding the Contributions of Operations, Technology and Design to Meeting Highway Capacity Needs. 2012. Available at: http://www.trb.org/Main/Blurbs/166939.aspx
[10] See http://ops.fhwa.dot.gov/freight/freight_analysis/perform_meas/
