Etalement urbain et géoprospective: Apports et limites des modèles de spatialisation Application aux modèles SLEUTH, LCM et NEDUM-2D

The aim of this paper is to assess the adv antages and drawbacks of using the SLEUTH, LCM and NEDUM-2 D m odels according to their ability to translate narrativ e userdefined urban planning scenarios into quantitativ e land use and land cov er change sim ulations. SLEUTH and LCM, which are land use and cov er change m odels, are spatially explicit and inductiv e pattern-based m odels while NEDUM-2 D is an econom ic processes-based m odel. Applied to the Toulouse m etropolitan area (France), SLEUTH and LCM are calibrated based on past urban growth dy nam ics observ ed between 1 990 and 2 006. After a detailed description of the m ethodology utilized to build contrasted, coherent and plausible urban development scenarios, the m odeling fram ework and hy pothesis of each m odel are presented. Then, a com parativ e analy sis of the SLEUTH and LCM outcom es is conducted based on a baseline scenario for 2 02 5 constructed using past trend extrapolation. The results show important differences between the LCM and SLEUTH models' outcomes. The influence of models on the sim ulation results concerns both the am ount of change and its spatial allocation. The findings also highlight that, based on a scenario-oriented approach, none of these empirical m odels is suitable by itself to deal with contrasted and anticipatory scenarios assum ing the absence of path-dependency . Based on this analysis, we suggest developing a new hybrid and dynamic spatially explicit economic model that should be used in a fully controlled forecasting mode and combine econom ic processes and spatial patterns of urban growth.