Spatial disaggregation and aggregation procedures for characterizing fluvial features at the network-scale: Application to the Rhone basin (France)

Network-scale heterogeneity in fluvial forms and processes has been analyzed for decades through the field-based characterization of a reduced number of sites. Spatially continuous data sets are increasingly available at a national scale (e.g., digital elevation models or aerial orthophotographs) and open up new possibilities to investigate spatial structures in stream networks. The paper introduces a GIS methodological framework to support and automate the large-scale characterization of the fluvial system based on spatial disaggregation and aggregation procedures. The spatial disaggregation consists in the longitudinal discretisation of geographical objects into relatively high resolution spatial units to characterize fluvial features continuously along stream networks. Because rivers are heterogeneous and hierarchically organized, the spatial aggregation aims to delineate statistically pattern or process-based spatial units to answer questions at appropriate scales. The paper develops the GIS implementation of the disaggregation and aggregation procedures for characterizing stream networks in a geomorphic perspective. We illustrate the approach on the Rhone network, focusing on the characterization of three geographical objects commonly used in classification schemes: the streamline, the valley bottom, and the active channel (the latter on a limited area). The spatial aggregation is automated using the threshold test of Pettitt (1979) to detect change points in univariate data series and delineate homogeneous spatial units throughout the stream network. We provide three simple application examples to illustrate the potentialities of the spatial aggregation to investigate spatial heterogeneity in stream networks. Finally, insights, limitations and future challenging issues emerging from the application of the methodological framework are discussed. (C) 2010 Elsevier BM. All rights reserved.