Mapping the aerodynamic roughness length using SAR images
Résumé
The aerodynamic roughness length characterizes the way the air flows over a surface, and more exactly the thickness of the layer above the surface wherein the airflow is influenced by the surface. This parameter is an input to several numerical codes that model dynamics of air flow. It is of interest in several domains, including air quality in cities and electricity production from windmills. Satellite data allow the mapping of the aerodynamic roughness, through land cover and land use mapping. Usually, classification of optical data is used for land use mapping and attribution of roughness length to classes. Some studies have demonstrated the potential of Synthetic Aperture Radar for roughness mapping. This paper exploits these early results in order to establish an operational procedure. We have designed a classification process based on a multi-resolution approach. The first step is the combination of a set of SAR images in order to reduce the influence of atmospheric conditions on the representation of structures. This multi-temporal image is processed in order to model the structures and to enhance their properties. Then a classification process is applied to the data set comprising the multi-temporal SAR image, and the modeling of the structures. The classes determined through the supervised classification process are affected to roughness length, based on the corresponding tables determined from the European Wind Atlas. This paper presents this approach, its principle, and mathematical aspects. An illustration of the methodology is provided.