Spaceborne SAR imagery for monitoring the inundation in the Doñana wetlands
- Autores: Belén Martí Cardona
- Directores de la Tesis: Carlos López Martínez (dir. tes.), Josep Dolz Ripollès (dir. tes.)
- Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2014
- Idioma: inglés
- Tribunal Calificador de la Tesis: Emilio Custodio Gimena (presid.), Ernest Bladé Castellet (secret.), Juan Francisco Saura Martínez (voc.), Juan Manuel López Sánchez (voc.), Juan José Negro Balmaseda (voc.)
- Materias:
- Enlaces
- Tesis en acceso abierto en: TDX
- Resumen
The Doñana wetlands, in Southwest Spain, undergo yearly cycles of inundation in fall and drying out during the spring season. As any wetland ecosystem, Doñana depends critically on its water resources. Important monitoring and modeling efforts have been made since the 1990's in order to assess the ecosystem health, partially restore Doñana¿s natural hydrologic scheme and design climate change adaptation strategies for ensuring its sustainability. The present research work inscribes within those efforts. Satellite images of Doñana marshes were acquired between 2006 and 2010, in order to obtain regional, synoptic observations of the inundation evolution, with the final aim to further current knowledge on the wetland's hydrology and to calibrate the existing two-dimensional hydraulic model of the marshes. The images were acquired by the C-band radar sensor ASAR, on board the Envisat satellite. ASAR's all-weather observation capability was particularly suitable to observe the flooding process, since this is associated to overcast weather. The images were ordered at seven different incidence angles in order to increase the observation frequency, and most of them in HH and VV polarizations. A detailed observation of the wetland filling up was achieved during the 2006-2007 hydrological cycle. Backscattering temporal signatures of the main land cover types in Doñana were derived for the different incidence angles and polarizations. The signatures were analyzed with the aid of ground-truth data in order to identify the effect of the flooding on the backscattering coefficients. Conclusions on the feasibility to discriminate emerged versus flooded land were drawn for the different incidence angles, land cover types and phenological stages: Intermediate incidence angles (ASAR IS3 and IS4) came up as the most appropriate single swaths to discriminate open water surface from smooth bare soil in the marshland deepest areas. Flood mapping in pasture lands, the most elevated regions, is feasible at steep to mid incidence angles (ASAR IS1 to IS4). In the medium elevation zones, colonized by large helophytes, shallow incidence angles (ASAR IS6 and IS7) enable more accurate flood delineation during the vegetation growing phase up to approximately mid-April, when the penetration capacity at these swaths might become insufficient. Due to the complex casuistic of the Doñana covers' backscattering, flood mapping from single ASAR acquisitions often resulted unfeasible. Flood mapping was then tackled by the synergistic use of the ASAR scenes with the Doñana¿s digital terrain model and vegetation map. The use of irregular filtering neighborhoods adapted to the topography drastically improved the ASAR image filtering, since natural edges tend to follow terrain contours. Clustering and classification algorithms were design to operate on individual sub-basins, as the pixels elevation is more accurately related to the cover classes within them. Vegetation and elevation maps plus knowledge of Doñana's backscattering characteristics from preceding studies were initially used to select seed pixels with high confidence on their class membership. Next, a region growing algorithm extends the seed regions with new pixels based on their planimmetric adjacency and backscattering distance to the seeds. Frequent flood mapping from the ASAR scenes enabled the identification of several events for the calibration of the marshes hydraulic model. This research work also revealed other target information enclosed in the radar backscattering, such as the helophyte vegetation developmental degree or the potential to discriminate among plant communities. Learnings from this research work are expected to be applied and furthered with upcoming data from the C-band Sentinel-1 constellation and Radarsat-2 satellite, the L-band ALOS-2/PALSAR-2 system and the X-band future Spanish mission PAZ.
