Strategically Adaptive Sustainable Mobility Systems
We conducted a series of analysis of exogenous uncertainty, model uncertainty and behavioral uncertainty using our Boston LUT model.
Exogenous Uncertainty (Sea Level Rise)
Visualizing Inundation Impacts on Transportation Networks (by Michael Dowd) Dowd, M.(2015) Modeling Inundation Impacts on Transportation Network Performance: A GIS and Four-Step Transportation Modeling Analysis. Thesis, MST & MCP, Massachusetts Institute of Technology, Department of Urban Studies and Planning, June.Dowd (2015) developed a method (Inundation Impact Assessment) for quantifying transport network impacts under six different inundation levels, one-foot to six-feet. A visualization of the SLR scenario and impacts can be found here. Dowd (2015) also demonstrated how four-step transport models can be used to plan for SLR by modeling two different demographic scenarios for the year 2030 with two different public transport infrastructure alignments.
Han, Y. Zegras, C., Rocco, V., Dowd, M., Murga, M. (2017) When the Tides Come, Where Will We Go? Modeling the Impacts of Sea-level Rise on Greater Boston's Transport and Land Use System. Transportation Research Record 2653, pp. 54-64.
We demonstrate how a land use-transport (LUT) model can be used to forecast the short and longer term impacts of potential 4-foot sea level rise in Greater Boston by 2030 (Han et al., 2017). The short-term scenario represents the immediate transport system response to inundation, which provides a measure of resiliency in the case of an extreme event, such as a storm surge. The longer term scenario aims to predict how households and firms would prefer to relocate in the “new equilibrium” where over ten square miles of land disappear and the transport network inundations become permanent.
Behavior and Model Uncertainty
Han, Y. (2015) Temporal transferability assessments of vehicle ownership models and trip generation models for Boston Metropolitan Area. Thesis, MST & MCP, Massachusetts Institute of Technology, Department of Urban Studies and Planning, June.Han, Y., Zegras, C. (2016) Exploring Model and Behavior Uncertainty: Temporal Transferability Assessment of Vehicle Ownership Models for Boston, Massachusetts, Metropolitan Area. Transportation Research Record 2563, pp. 122-133.
Han (2015) analyzed the temporal transferability of vehicle ownership models and trip generation models for Boston metropolitan area from 1990 to 2010. The statistical tests show significantly changed preferences in household vehicle ownership choice and trip production. The prediction tests suggest that failing to consider preference changes cause significant bias in population demand forecasts. Han and Zegras (2016) further analyze model uncertainty and behavior uncertainty in vehicle ownership modeling.
Posada, P (2015) Location, location, location choice models. Thesis, M.C.P., Massachusetts Institute of Technology, Department of Urban Studies and Planning, June.
In the land use model Posada (2015) examined data-related uncertainty (how model estimation changes with different data sources) as well as temporal transferability (how do preferences change over time). The analysis of data-related uncertainty indicates that the models are sensitive to the specific dataset used in the estimation. The accuracy seems to be correlated to agent category size in the sample data. Preference change uncertainty was examined in the firm location choice model. The location choice models suggest that firms' willingness to pay for clustering has changed from 2000 to 2010.
Uncertainty Propagation
Within the full four-step model system we analyzed the propagation of uncertainty arising from two sources: model uncertainty and behavioral uncertainty, and from individual component (Vehicle Ownership, Trip Generation, Mode Choice). We compared the magnitudes of their impacts on network performances. (See a working report here.)
Our overall finding is that behavior uncertainty has substantial impacts on model forecast results, while sampling uncertainty of parameters leads to a smaller range of variations relative to the point estimates.
Participants: Michael Dowd, Yafei Han, Menghan Li, and Shenhao Wang, Victor Rocco Supervisors: Prof. Christopher Zegras, CEE Research Associate Mikel Murga.