Modelos de simulación espacio-temporal y teledetecciónel método de la segmentación para la cartografía cronológica de usos del suelo

Abstract

The conception of geomatic simulation models applied to environmental dynamics aims to carry out past and present dynamics, to reveal the characteristic spatiotemporal behaviour and, through a rule base, to anticipate future changes, or at least, to build foreseeable scenarios in the short to medium term. Modelling environmental dynamics contributes to understanding and anticipating future developments, forecasting and simulation tools provides a decision support for environmental management.The geomatic simulations are based on spatiotemporal databases (mainly land use maps). In order to build these databases aerial and satellite remote sensing, but also any other tool providing both, analogue and digital, classification has an inescable role. In the framework of several research projects focused on the analysis and the implementation of simulation models applied to environmental dynamics, the usefulness of classification procedures recently implemented in the IDRISI Taïga GIS such as segmentation to obtain chronological land use maps supporting these models has be shown. In practical terms, the segmentation classification approach is being applied to aerial orthophotographs as an alternative to the classic manual photo interpretation. This paper shows the results of the construction of a time series of land use maps in the Eastern French Pyrenees. A purely spectral segmentation has been applied to the BD Ortho 2004 (Institut géographique National, France), a digital aerial photography with three visible RGB bands, and to an older one which had to be scanned and subsequently orthorectified (1980, panchromatic). In the next step, the main class drawn from a supervised classification is assigned to the segments. This paper explains the implementation and discusses the results. The principal conclusion is that the automation made by the presented method allows important savings in time and work during the detection stage of homogeneous spatial units and the allocation of attributes. This represents an effective help for configuring series of chronological databases.