Introduction: In the recent decades, raster based cellular automata (CA) models have been used increasingly in simulating land-use changes and urban growth because of simplicity of the computation and their conformity with remote sensing data. Current researches have introduced vector based cellular automata to tackle with some limitations that raster based conventional cellular automata models are faced with, such as sensitivity to spatial scale. The purpose of this study is modelling land use suitability to simulating land use changes through vector based cellular automata with application of cadastral parcels instead of pixel.

Materials and Methods: In the proposed model, urban land use suitability is evaluated based on physical suitability, accessibility and neighborhood effect. Two parameters of altitude and slope are combined together by weighted linear combination (WLC) method and are representative physical suitability and Euclidean distance of each parcel from the closest point of path network is utilized to measure accessibility factor. Spatial externalities of neighborhood land uses are important component in simulating land use changes and using cellular automata is justifiable if neighborhood rules, as heart of the cellular automata models, are defined correctly. One of the limitations of neighborhood rules is the lack of theoretical foundation and empirical validation. So, CA urban models are technology driven rather than theory, due to considering hypothetical ideas about urban dynamics. Therefore, neighborhood rules are often defined by trial and error. In this study, the neighborhood effect is defined based on a function of three components including compactness, dependency and compatibility and the way of land use interactions is studied in three service levels of local, district and regional. Compactness is the tendency toward creating a kind of land-use adjacent to another similar one dependency is depending one land-use on the other land-uses for supplying its needs and compatibility is defined as the arrangement of two or more land-uses together without any significant negative effects. The neighborhood effect of each parcel for each land-use is calculated from total effects of all adjacent parcels in the form of three classes of compactness, dependency and compatibility. It should be considered that each of factors has various impacts in land-use interactions and can be combined together with different weights, regarding the subject land-use. Therefore, relative importance of three factors is defined using expert knowledge captured based on AHP method and then the neighborhood effect of each parcel is computed using to WLC method.

Results and Conclusions: This model is implemented and assessed using data of Tehran 22 municipal region, in two scenarios: land use preference and maximum suitability. The results are illustrated high ability of the model in simulating land use interactions and presenting urban growth in an acceptable level of consistency among the urban land uses. Evaluating the results in the case study area shows an average score 0.77 and 0.55 in Scenario 1 and 0.75 and 0.47 in scenario 2 (from 0 to 1) for compatibility and dependency of new allocated parcels, respectively. The proposed model can be utilized by urban planners to assess different urban scenarios for better decision making and urban planning.