Modelling Coastal and Land Use Evolution Patterns through Neural Network and Cellular Automata Integration

The growth and development that the Municipality of Almada (Portugal) has had, an area with an extensive coastal line, both Atlantic and estuary, translates directly into morpho-functional changes and specific spatial dynamics, with these spaces acquiring more and more complex semantic meanings. The modelling of spatially dynamic and naturally complex phenomena, such as those occurring in the coastal areas, are important for the definition of an innovative strategy of physical planning and environmental management in these coastal areas. In this spirit we have undertaken the development of innovative methods aimed at this modelling, which enable the understanding of the land use and occupation phenomenon in coastal areas subjected to strong urban pressure. The results of which will be discussed. This paper presents a method to simulate the coastline evolution (integrating land use/cover), linking neural networks and cellular automata in a GIS environment. Such alterations simulated solely by means of cellular automata are not suitable, because these models, in their more conventional structure, contain limitations in their spatial parameters and transition rules. In this paper, a neural network is used to calibrate the degree of importance that each prediction variable (probability) has on the geographic constraints (weights). All these variables are handled within a GIS environment.