Browsing by Subject "hierarchical classification approach"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Publication
    Open Access
    Morfología Fluvial.-Intrinsic properties of channel network structure and the hierarchical classification approach for stream-limits delineation
    Afana, A.; Del Barrio, G.; Universidad de Murcia
    ABSTRACT Delineation of drainage networks is an essential task in hydrological and geomorphological analysis. Manual channel definition depends on topographic contrast and is highly subjective, leading to important errors at high resolutions. Different automatic methods have proposed the use of a constant threshold of upslope contributing area to define channel initiation. Actually, these are the most commonly used for the automatic-channel network extraction from Digital Elevation Models (DEMs). However, these methods fail to detect an appropriate threshold when the basin is made up of heterogeneous sub-zones, as they only work either lumped or locally. In this study, the critical threshold area for channel delineation has been defined through the analysis of dominant geometric and topologic properties of stream network formations. In addition, a hierarchical classification approach has been integrated in order to verify landscape units in relation to dominant intrinsic properties. Such approach provides various critical thresholds as much as necessary in relation to DEM-data resolution as well as homogeneity or heterogeneity of the dominant landforms. The study has been carried out in two controlled drainage basins with different lithology and geomorphic processes. High-resolution DEMs (1m) were used to obtain the best detailed drainage network that the algorithm can generate in a homogeneous landscape, whereas a mediumresolution DEM (30m) was applied in a heterogeneous complex landscape. Validation results revealed that the above approach is adequate for describing terrain dissection, since its function depends on intrinsic properties of the drainage network, being at the same time objective and easy to implement. Likewise, it provides an enhanced approximation to empirical geomorphometric parameters used to describe stream network dimensions.