Strategically Adaptive Sustainable Mobility Systems

Greater Boston Land Use and Transport Model

Our team has developed, calibrated and partially validated an integrated land use-transport (LUT) model for the Boston metropolitan area and run the model under alternative scenarios up to 2030.

The land and transport models are linked by accessibility measures, household and firm locations, and other zonal variables derived from number of agents and travel skims, updated during each model iteration (Figure 1). The models are integrated via python scripting which automates the full model system running including the necessary translation between the different zonal structures and agent types across the land use and transport models and updating of the endogenous variables (e.g. accessibility) based on each model's outputs.

Figure 1. Structure of Integrated Land Use and Transport Model for Boston Metro Area.

In the available computational environment, each complete LUT loop required four hours and we ran 5 loops per scenario, meaning each model run required 20 hours (the majority of computational consumption came from the network assignment stage).

Transport Model

On the transport side of the LUT model, we developed an enhanced four-step transport model for Boston Metropolitan Area. Compared to conventional practices, we introduced more complexity and flexibility to the model, which provides a richer representation of reality, and the ability to run uncertainty simulations (including alternative models) inside a four-step model.

For more details, see Transport Model Specification and Estimation and Transport Model Forecast and Validation for 1990 and 2010.

Participants: Yafei Han, Michael Dowd, and Shenhao Wang.
Supervisors: Prof. Christopher Zegras, CEE Research Associate Mikel Murga

Land Use Model

The land use model is implemented in Cube Land, the commercial version of MUSSA, an operational land use model based on auction theory. The market equilibrium model consists of three sub-models: Demand, Supply and Rent model.

Our LU model version allocates 12 types of households (defined by age, income, size), and 11 types of firms (defined by industry) to 12 kinds of real estate units (e.g., single family, large apartment) in each traffic analysis zone (TAZ).

Participants: Victor Rocco, Pablo Posada, Menghan Li
Supervisors: Prof. Christopher Zegras