Document Actions
Survey of Surface Roughness for Hydrodynamic Models
The project assessed benefits for the quality of hydrodynamic models by using high resolution satellite imagery and land cover data. In cooperation with the Hamburg University of Technology, Institute for River and Coastal Engineering, a test area at the River Stör in the Federal State Schleswig-Holstein was selected, where a hydrodynamic model already existed. Therein the necessary surface roughness was first estimated roughly in rasters of large land cover units.
For the test area of 64 km² along the River Stör QuickBird satellite data from September 21st 2005 and CORINE data with four land use types (settlement, field, grassland and woods) as well as the border of the river were allocated.
Through a knowledge-based classification (FRICK, 2007), which used the high resolution satellite data (QuickBird) and the land use data (CORINE), the surface roughness was calculated on the pixel base. The river body was used to retrieve training areas for water, for every other class the CORINE data was used. Because of the minor quantity of land use types of the CORINE data only few water dependent vegetation classes could be extracted (exclusively nearby the river).
For a certain comparability to the roughness classes of the previous model, nine classes according to their characteristic surface roughness (sealed, open soil, mowed grassland, medium grass 0.5-1.5 m, high grass 1.5-2.5 m, intensive fields/meadows, swamps and vegetation, trees/shrubs, water) were combined and were allocated with values of hydraulic roughness. RATH (2006) developed a scheme to aggregate and parameterize pixel based land cover classes to hydraulic roughness classes and to assess the quality of these transformations in discrete meshes. Thus classes that contain sealed surfaces (concrete, asphalt etc.) were combined to one class "sealed", because of there similar hydraulic roughness.
The classification of surface roughness for hydrodynamic models by this method is highly reliable. It exceeds rough aggregations of field surveys or cadastral coarse data sets. Ultimately it provides an up-to-date definition of the land cover.
Literatur:
- Frick, A. (2007): Beiträge höchstauflösender Satellitenfernerkundung zum FFH-Monitoring - Entwicklung eines wissensbasierten Klassifikationsverfahrens und Anwendung in Brandenburg. Technische Universität Berlin, Dissertationen online. URL: http://opus.kobv.de/tuberlin/volltexte/2007/1413/.
- RATH, S. 2006: Model Discretisation in 2D Hydroinformatics based on High Resolution Remote Sensing Data and the Feasibility of Automated Model Parameterization. PhDThesis, Hamburger Wasserbauschriften, Institute of River & Coastal Engineering, Hamburg University of Technology.
