Overview
A literature review showed that urban transport interaction modelling, the indirect distributional effects of transport and socio-economic assessments of transport are mostly treated in theory and that nothing was found in France on the practical level.
SIMAURIF is a land-use and transport interaction (LUTI) model of the Paris region that was developed by the Urban Planning and Development Institute of Paris Ile-de-France region (IAU Ile de France), in collaboration with the THEMA laboratory of the University of Cergy-Pontoise.
The SIMAURIF model is based on two existing models: OPUS.UrbanSim (a land use model developed by the University of Washington) and Davisum METROPOLIS (a transport model).
This research aimed to develop an operational dynamic and integrated model of the interaction between urbanisation and transport in the region of Ile-de-France and its application to a case study, the tangential North. The model is called SIMAURIF in French (SIMulation de l'interAction Urbanisation-transport en Région d'Ile-de-France), and uses two existing models: Urbansim, a urbanisation model and Davisum-METROPOLIS, a traffic forecasting model.
The first objective was to develop an operational model that could test a transport scheme or simulate transport or land-use policies and show the effects on urban development or on mobility behaviour.
The second objective was to measure the indirect effects of infrastructure (residential attractiveness, economic development).
The methodology was based on the following principles:
- taking into account environmental constraints (flood areas, historic monument areas, Natural Areas of Fauna and Flora Ecological Interest, etc.) and local politics (local development plans);
- a will to give consistency between entries and outputs of different models;
- an adaptation of the data structure of UrbanSim of the study area (i.e. the Ile-de-France region); and
- a will to exploit fully the municipal level data.
The methodology was also built on the construction of comprehensive and disaggregated databases and not synthesised as they were in the United States, a continuing effort to analyse the data descriptively before moving on to modelling, and a unique database to estimate the various choice models. The quality of the models obviously depends on the quality of econometric estimates, and also on the data quality. An unprecedented effort was therefore made to build the database entries and database for model estimates.
As for the model calibration, the methodology was based on models being calibrated independent of each other, then a global calibration, the validation of each model and the validation of all models. It was necessary to estimate for each model independently. To develop a tool, it is better to settle for a simple predictive model rather than a sophisticated explanatory model that is inapplicable in forecasting the current version of UrbanSim. This methodology of independently calibrating each model was time consuming.
There was also a clear distinction between the calibration period and the simulation period.
This chronology reflects the methodology:
- Literature review on traffic model, urbanisation model and integrated models
- Theoretical study of UrbanSim (the four main models and five secondary models not discussed in the study)
- A first grip of UrbanSim on an American case study (Eugene - Springfield)
- Architecture design of the integrated model SIMAURIF
- Model calibration of travel demands of 606 areas from the Global Transport Survey 2001 and for three travel trip reasons
- Establishment of a georeferenced and disaggregated database into 50,000 square cells on the Ile-de-France region
- Creation of loca
Funding
Results
SIMAURIF was used to assess the impacts of the North Tangential project (a tram train project in the North of Paris scheduled to be opened in 2016). The application of the SIMAURIF integrated model on that project case was successful. It yielded interesting results for the horizon year 2026 (12 years after commissioning) on economic and residential attractiveness of the territory served by the infrastructure.
The model could forecast population and spatial distribution of employment at the cell level and could distinguish a difference of repartition between the scenario with and the scenario without the infrastructure. These results confirmed the overall qualitative analysis of planners, although not conclusive on the issue of the real estate.
The SIMAURIF tool could position itself as a new decision support tool as part of a multi-criteria analysis. It assessed the impact of transport infrastructure on the household and businesses relocation under the space availability constraints and under the housing market competition conditions. The model could not simulate the housing price in 2026 at the cell level, however, but only at an aggregate level.
Technical Implications
- The integrated approach consisted in coupling a traffic model and a urbanisation model.
- The cell level approach consisted in studies on the scale of the address, block, district that must be generalised.
- The disaggregated approach consisted in independently representing agents with their characteristics and their probabilities of discrete choice.
- UrbanSim had to be converted to the French context (efficient organisation of space, public transport, etc.) as UrbanSim did not integrate public transport variables.
- The project showed that it was better to build a simple model with few variables (10 maximum) than building a complicated model because it would be impossible to apply it in forecasting as the variables chosen for estimation could not be forecast.
- The project showed that excellent knowledge of the study area was essential.
- The model could not measure the housing prices but could measure some indirect effects such as residential attractiveness and economic development due to infrastructure.
- The project showed that an inter-disciplinary team and adequate resources were necessary as modelling the interaction between transport and land use requires different skills: econometrics, demography, economics, geography, urban planning, complex systems, mobility behaviour, and computer science.
- In conclusion, a wider integration of the model could be interesting, such as connecting the integrated model to an environmental assessment model (pollution, noise, emissions, etc.), in order to calculate sustainable development indicators for example, or to a macro-socio-economic model.
Policy implications
None