Calibration and valIdation of transport-land use models – CITiES

To reach our goal, we (a) define a calibration methodology and developing relevant and efficient algorithms to facilitate the parameter estimation of LUTI models. (b) We define a validation methodology, in both the historical and urban economics senses, and developing the related algorithms. In both cases, analyzing the uncertainty that may arise from either the data or the underlying equations, quantifying how these uncertainties propagate in the model, and performing sensitivity analysis to determine the relevance of the various data and model parameters are of major importance. Completing these various tasks would make LUTI models easier to implement, and greatly enhance the confidence in their results.

Three LUTI models will serve as sample to test the methodologies that will be developed in the CITiES project: Tranus, UrbanSim, and Pirandello. They are quite representative of LUTI models, with two equilibrium models and one microsimulation model. Tranus and UrbanSim are Open Source and also the most largely applied models worldwide, while Pirandello is the only operating model developed in France. The study areas are the Grenoble region for Tranus, the Lyon region for UrbanSim and the Paris region for Pirandello, which offer different local contexts.
The consortium also intends to support the dissemination of LUTI models and CITiES results through significant interactions with policy-makers and end users, in the form of workshops and information meetings.
Besides the scientific difficulties raised by the development of adequate methodologies, another important issue lies in the significant amount of data that is needed to perform the historical validation steps for the case studies. Data acquisition and pre-processing will therefore be a task in itself.

Results will be presented ASAP (The project has started in January 2013)

Perspectives will be presented at the end of the project (The project will end in 2016)

Publications will be presented with the stream. The project has started in January 2013.

In a context of climate change, dwindling fossil resources and mild economic growth, urban sustainability has become a key policy issue. Given the complexity of modern urban areas, designing sustainable policies calls for more than sheer expert knowledge. This is especially true of transport or land use policies, because of the strong interplay between the land use and the transportation systems. Policies that seem perfectly sound intuitively may ultimately yield undesirable effects because of this connection, which is extremely hard to apprehend without the help of numerical simulations. Land use and transport integrated (LUTI) modeling thus offers invaluable analysis tools for planners working on transportation and urban projects. Yet, very few local authorities in charge of planning make use of these strategic models. The explanation lies first in the difficulty to calibrate these models, second in the lack of confidence in their results, which itself stems from the absence of any well-defined validation procedure.
This proposal aims to foster the use of LUTI models for the design and evaluation of land use and transport policies by addressing these two impediments. This involves: (a) defining a calibration methodology and developing relevant and efficient algorithms to facilitate the parameter estimation of LUTI models; (b) defining a validation methodology, in both the historical and urban economics senses, and developing the related algorithms. In both cases, analyzing the uncertainty that may arise from either the data or the underlying equations, quantifying how these uncertainties propagate in the model, and performing sensitivity analysis to determine the relevance of the various data and model parameters are of major importance. Completing these various tasks would make LUTI models easier to implement, and greatly enhance the confidence in their results.
Three LUTI models will serve as sample to test the methodologies that will be developed in the CITiES project: TRANUS, UrbanSim, and PIRANDELLO. They are quite representative of LUTI models, with two equilibrium models and one activity-based model (or transition model). TRANUS and UrbanSim are open source and also the most largely applied models worldwide, while PIRANDELLO is the only operating model developed in France. The study areas are the Grenoble region for TRANUS, the Lyon region for UrbanSim and the Paris region for PIRANDELLO, which offer different local contexts.
The consortium also intends to support the dissemination of LUTI models and CITiES results through significant interactions with policy-makers and end users, in the form of workshops and information meetings.
Besides the scientific difficulties raised by the development of adequate methodologies, another important issue lies in the significant amount of data that is needed to perform the historical validation steps for the case studies. Data acquisition and post treatment will therefore be a task in itself.
To successfully carry out the CITiES project, the consortium can draw on two main strengths:
• its unique panel of experts in urban modeling, urban planning, mathematics and computer science, which is to be the cornerstone of the development of innovative tools for the calibration and the management of uncertainties in LUTI models;
• the significant experience of most partners with LUTI modeling.
The partnership with the IAU îdF and the AURG will also prove extremely fruitful, considering both the data issues and the firm intent to interact with end users.
If the pluridisciplinary character of the project is clearly an asset, it may also constitute a more important difficulty in terms of communication between team members; particular care will be paid in the project management to ensure smooth exchanges of information and knowledge and mutual understanding between the different scientific communities involved in the project.