LAND COVER MAPPING IN THE VICINITY OF DĘBLIN AERODROME USING ESA’S SENTINEL-1A/B SATELLITE IMAGERY

ABSTRACT

Land use and land cover (LULC) in the vicinity of airports are the key elements in spatial planning and aerodromes development. In many geographical contexts there are cartographic databases or geographic information systems providing necessary and updated information on LULC. On the other hand where the geodatabases are not available or outdated the satellite images can be used for mapping, especially weatherindependent radar imageries. Synthetic Aperture Radars (SAR) operating in microwave region (L-, C- or X-band) are well suited for detection of buildings and other man-made structures. These objects are called the strong scatterers because they strongly reflect microwaves due to their shape and material of which they are constructed. The strong scatterers can be easily distinguished from so called extended objects representing agricultural areas, surface waters and forests. The extended objects themselves representing different land cover classes can also be separated each from other based on their distinct surface roughness and dielectric properties. The Sentinels-1A/B C-band sensors are freely available and provide high resolution (about 10x10m) images at VV and VH polarizations on a regular basis. The results of LULC mapping with SAR depend on the dates of images acquisitions, polarization, incidence angle and orbit orientation towards the linear topographic objects. Two time series of Sentinel-1 Interferometric Wide swath mode images were used: 8 images from descending and 8 images from ascending orbits. The time series were also split into spring and autumn series in order to exploit the radiometric contrast between extended objects linked to vegetation absence and changing soil roughness. The native data was transformed into sigma nought values and four time series were analyzed regarding: I) polarization VV or VH, II) orbit DESC or ASC. No speckle filtering has been applied except images summation in order to preserve the native resolution. The results showed that combining of descending and ascending orbit is necessary for railways mapping. Numerous sparse buildings forming villages were correctly mapped as strong scatterers on VV images. The VH polarization turned out to be useful for water surface mapping but the aerodrome runway was too similar to the surface of calm water.

Keywords: land cover mapping, Sentinel-1, Deblin aerodrome