GIS and spatio-temporal modelling for the study of alluvial soil and vegetation evolution.

This research was developed in an interdisciplinary context, as collaboration between the pedology and GIS Institute of the Rural Engineering Departament of the Swiss Federal Institute of Technology (EPFL). Within the pedological domain, this works comes within the scope of environmental research aiming at the analysis of the sensitivity of the alluvial ecosystems to natural and anthropic changes. Within the GIS domain, this research comes within the scope of the GIS design methodologies applied to different thematic domains, and one the other into the research concerning the integration of time into GIS applications. The interest of alluvial zones is the fact that these ecosystems have an large complexity and biological diversity, which surpasses, in general, that from other temperate environments because of their rapid dynamic. This dynamic, running in a human scale, permits reliable historical reconstruction. In spite of their ecological interest (e.g. large diversity of species, land-water interface, flood regulator...). These ecosystems are among the most threatened in Europe because of the high degree of man-induced alteration. Recently, the European alluvial zones have been inventoried with the goal of conservation, protection and restoration. Subsequently, Switzerland gas estabilished an inventory of its alluvial zones (0,25% of the territorial area) and mapped the vegetation. Since 1992, a Federal Edict has regulated the protection of the 169 sites considered to be of national importance, and suggests measures of their management. The management suggested in the Federal Edict must be based on rigorous scientific knowledge about such ecosystems. At the present such knowledge is quite fragmented. The speed of transformations, the vegetation change patterns and direction, as well as the most adequate scales to bring out the spatio-temporal changes in this environment is still questions without answers. Concerning alluvial soils, there is litlle avaiable data, principally that about alluvial soils developed from the recent sediments. The stablishment of a model able to represent in space and time the entities and its change becomes fundamental for the comprehension and management of such as environment. This research has pursued a double aim: On the other hand, it is a matter of improving our knowledge about alluvial environment upon human action, and particularly about the formation, evolution, the present-day variability and spatial distribution of soil. On the other hand, the focus is done in the design of a GIS methodology able to integrate spatio-temporal information and the particularity of environmental research. In oder to fulfil these objectives, this research relies on an integrated approach that combines synchronic and diachronic approaches into a double research methodology: a pedological methodology and a GIS design methodology that integrates the spatio-temporal information. With such an approach we were able to study the present-day state of soils in relationships to the change sequence lead by fluvial dynamics and/or human inventations.

These (Docteur ès Sciences) - École Polytechnique Fédéral de Lausanne, Lausanne.