In January 2016, the Paraná River caused flooding in the city of Santa Fe, Argentina, affecting areas bordered by the Colastiné River and Setúbal Lagoon, both located within the alluvial valley of the Paraná. The gauges registered the peak flood during this event; however, a precise spatial delineation of the flood extent was not captured. Therefore, this study aimed to estimate the flood extent and identify the affected areas. In this manner, Sentinel-2 imagery was used, and a high-pass filter was applied to enhance the resolution of the short-wave infrared bands from 20 m to 10 m. The fusion was evaluated using the quality-without-reference index, yielding an accuracy above 81%. Several spectral indices, including the automated water extraction index (AWEI), desert flood index, modified normalised difference water index, new water index, normalised difference water index, and water ratio index, were compared. AWEI demonstrated the highest overall accuracy (97.4%) and Kappa coefficient (94.9%), indicating excellent agreement with ground-truth data. Subsequently, the floodline was overlaid on a 5 m resolution digital elevation model to extract the lowest elevations within the affected areas, which were hypothesised as the initial water ingress locations. The flood impacted areas in the La Costa district and the NW of Setúbal Lagoon, all situated outside the levee system. The results of this study could enable local authorities to anticipate the impacts of future flood events.
Estimation of the 2016 Paraná River (Argentina, South America) flood extent using satellite imagery, band fusion, and spectral indices
Abstract:
