Radar de abertura sintética aplicado ao mapeamento e reconhecimento de zonas úmidas costeiras

Abstract

The use of SAR imaging to identify coastal wetlands under different acquisition configurations and different environmental conditions is a research topic at the forefront of knowledge in the field of remote sensing. In order to evaluate the potential of the use of SAR images in tropical coastal environments, an Environmental Protection Area of Baixada Maranhense was chosen to be investigated, given the state of preservation of the study area, where one can observe, mainly, mangrove ecosystems, fields salt marshes and fields flooded by fresh water. RADARSAT and ALOS PALSAR images in different imaging modes were acquired under different environmental conditions. In RADARSAT-1 Wide 1 images, it was possible to separate the mangroves and floodplains, in the dry and rainy seasons, especially when the images were acquired in conditions of zero precipitation, since with humidity in the system, the separation between these environments tends to decrease by the increase in backscattering. The influence of the high tide condition on the backscatter values of the RADARSAT-1 images was not verified. Contextual automatic classification based on frequency was able to separate mangroves from floodplains and water in both seasons from floodplains and water in both seasons. The RADARSAT-1 Wide and Fine images, acquired during the rainy season, under high rainfall, different tidal conditions and different incidence angles (20º - 47.8º) were transformed by the principal component analysis method, which attested that they are highly correlated. The backscatter analysis of these images showed a low separability for smaller incidence angles (20º to 31º), while in the images acquired at larger angles (41º to 47º) the separability is a little higher, when analyzing the mangrove environments and floodplains. In multipolarized RADARSAT-2 and PALSAR-HH images, acquired in the dry season, in areas of floodplain, saline and pasture grasslands, it was verified that the microtopography parameter HRMS, of the grassland areas, influenced about 49% of the values of σ0, of the image PALSAR-HH. For the RADARSAT-2 images, in the HV and VH polarizations, this influence was low, around 24% and 26%, respectively. In the parallel polarizations, the microtopography did not influence the zero sigma variations. Therefore, SAR images can be considered an excellent source of data for the recognition and monitoring of tropical coastal wetlands.