Science Service System

Summary of Proposal HYD3110

TitleIntegrating RADAR images and flooding hydrological modeling
Investigator Maltese, Antonino - University of Palermo, DICAM
Team Member
Prof. Ciraolo, Giuseppe - University of Palermo, DICAM
Eng. Capodici, Fulvio - University of Palermo, DICAM
Prof. La Loggia, Goffredo - University of Palermo, DICAM
SummaryObjectives: The most of literature on remote sensing of floods has focused on rural case studies. Actually, it would be more important developing methods able to map urban floodings because of the increased urban risks (mainly due higher and higher population density). The main issue related to active remote sensing detection through images is the presence of shadows, especially near the buildings. Shadowed areas appear dark and maybe misclassified as water even if it is dry (false positive). In contrast, a flooded area close to a building wall may be allocated to the same range bin as the wall itself, causing a layover whichgenerally results in a strong return and a probable misclassification a sun-flooded (false negative). General aim of the project is to merge information provided by a flooding hydrodynamic model and radar imagery to reduce the uncertainty of flood mapped areas. The process should be as much automatic as possible, computationally efficient, independent of the content of local geo-information databases and capable of providing accurate and reliable results. Method: The starting of our study is linked with the results of the research presented by Giustarini et al. (2013). This paper reports a study on the Flood Mapping in Urban Area using TerraSAR, focusing on the problem of the shadow’s images. Over and under -detection of flooded areas are issues still to be solved. According to the above-mentioned study our proposal will develop an algorithm able to find pixel or regions misclassified into flooded and non-flooded regions. This algorithm, shared by the team, will be based on the processing of the SAR images to extract region pixels. Factors able to influence the backscattering process will be analyzed.These factors will be included in our methodology (wind, rain, roughnesschange). Furthermore the approaches proposed by R. M. Haralick and L. G.Shapiro (1985), will be employed to classify the backscattering images. In addition the hydrodynamic modeling will be used in a "feedback" way in order to cooperate in false positive/negative location. Furthermore a very accurate DEM obtained by airborne LIDAR, will be used to map the water depth of the flooded areas. These dataset will be used to calibrate/validate the hydrodynamic model. Data requirements: For flooding in built-in areas it appear necessary to use a high resolution radar images. ASAR images allows distinguishish structures with 3x3 meter dimensions,for example streets, buildings, cars and trees. For that reason, it’s required working with TerraSAR-X Stripmap images or High Resolution SpotLight images that show flooded urban areas. Similarly it is crucial to have a very high resolution DEM to support hydro-dynamic modeling and evaluation of waterflooding volume. Deliverables: By applying a 2-D flood inundation modeling the urban areas by constraining the friction parameters, modeled urban flood extent will be compared to remote sensing retrievals. Deliverables include reports on the capabilities of radar SAR acquisitions coupled with flood inundation model to detected urban flood region; maps of floods in open areas and urban areas and maps of flood depth estimation based on a high resolution DEM.

Back to list of proposals

© DLR 2004-2016