|
Titel |
Automatic pre-processing for an object-oriented distributed hydrological model using GRASS-GIS |
VerfasserIn |
P. Sanzana, S. Jankowfsky, F. Branger, I. Braud, X. Vargas, N. Hitschfeld |
Konferenz |
EGU General Assembly 2012
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 14 (2012) |
Datensatznummer |
250058458
|
|
|
|
Zusammenfassung |
Landscapes are very heterogeneous, which impact the hydrological processes occurring in the
catchments, especially in the modeling of peri-urban catchments. The Hydrological Response
Units (HRUs), resulting from the intersection of different maps, such as land use, soil types
and geology, and flow networks, allow the representation of these elements in an
explicit way, preserving natural and artificial contours of the different layers. These
HRUs are used as model mesh in some distributed object-oriented hydrological
models, allowing the application of a topological oriented approach. The connectivity
between polygons and polylines provides a detailed representation of the water
balance and overland flow in these distributed hydrological models, based on irregular
hydro-landscape units. When computing fluxes between these HRUs, the geometrical
parameters, such as the distance between the centroid of gravity of the HRUs and the river
network, and the length of the perimeter, can impact the realism of the calculated
overland, sub-surface and groundwater fluxes. Therefore, it is necessary to process the
original model mesh in order to avoid these numerical problems. We present an
automatic pre-processing implemented in the open source GRASS-GIS software,
for which several Python scripts or some algorithms already available were used,
such as the Triangle software. First, some scripts were developed to improve the
topology of the various elements, such as snapping of the river network to the closest
contours. When data are derived with remote sensing, such as vegetation areas,
their perimeter has lots of right angles that were smoothed. Second, the algorithms
more particularly address bad-shaped elements of the model mesh such as polygons
with narrow shapes, marked irregular contours and/or the centroid outside of the
polygons. To identify these elements we used shape descriptors. The convexity
index was considered the best descriptor to identify them with a threshold of 0.75.
Segmentation procedures were implemented and applied with criteria of homogeneous
slope, convexity of the elements and maximum area of the HRUs. These tasks were
implemented using a triangulation approach, applying the Triangle software, in order
to dissolve the polygons according to the convexity index criteria. The automatic
pre-processing was applied to two peri-urban French catchment, the Mercier and Chaudanne
catchments, with 7.3 km2 and 4.1 km2 respectively. We show that the optimized
mesh allows a substantial improvement of the overland flow pathways, because
the segmentation procedure gives a more realistic representation of the drainage
network.
KEYWORDS: GRASS-GIS, Hydrological Response Units, Automatic processing,
Peri-urban catchments, Geometrical Algorithms |
|
|
|
|
|