Espace membre

Cet espace est dédié aux étudiants, aux enseignants et au personnel administratif de l'école


Mot de passe oublié?



Doctorate thesis defense of Chayma Chaabani

Doctorate thesis defense on July 16th 2020 at 10H00 ,in Sup’Com Amphitheater Ibn Khaldoun.

Entitled :Flood monitoring using InSAR data

Presented by :Chayma Chaabani


President :

Prof. Sadok El Asmi

Professor at University of Carthage




Referees :

Prof. Hedia Chakroun

Professor at University of Tunis


Prof. Vito Pascazio

Professor at University of Naples "Parthenope"


Inspector :

Prof. Abdelaziz Kallel

Professor at University of Sfax




Thesis Supervisor :

Prof. Riadh Abdelfattah

Professor at University of Carthage


In the context of flood risk management, it is important to provide accurate flood maps in order to ensure an effective guidance for the decision makers. The aim of this thesis is to produce automatic or semi-automatic mapping tools to monitor the soil behavior during and after the inundation events, that is to say, mapping the flooding extent and tracking the ground surface displacements that are triggered by floodwater (post-flood).

Based on the specific requirements of the flood mapping problematic, we used multidisciplinary methods from the RADAR remote sensing imagery, the computer science approaches and the thematic knowledge about hydrology and geology. The Synthetic Aperture RADAR (SAR) data were used since they provide information regardless of the sunlight and the weather conditions. Moreover, we considered the SAR interferometry (InSAR) technique that measures the phase difference between two SAR images in order to evaluate the soil surface changes and the Earth's topography.

We addressed the problem of flood mapping in rural areas without a-priori knowledge about the flooded region. We adapted the Fuzzy C-Means unsupervised classification approach to the flood mapping conditions by considering the InSAR coherence information. In this context, two flooding events in Tunisia (2005) and Spain (2019) were examined using the C-band imagery of Envisat and Sentinel-1 data.

Furthermore, we were interested to map the floodwater in a complex environment that contains dense urban and vegetated areas. We took advantage of the bistatic TerraSAR-X/Tandem-X InSAR data and the hydraulic modeling of LISFLOOD-FP to study the flooding event that happened in Quebec (2011). As a-priori data of the flooded region were available, we considered a flood mapping approach based on the Random Forest (RF) supervised classification. The achieved mapping results using SAR/InSAR data were validated by the means of the flood map derived from the generated LISFLOOD-FP hydraulic model.

Regarding the monitoring of post-flood soil deformations, we considered the ground surface movement that happened after a flooding event in Tunisia on October, 18, 2018 and we used the Parallel Small Baseline Subsets (P-SBAS) differential InSAR approach with a set of multitemporal Sentinel-1 data covering the period between 2016 and 2019. A qualitative analysis was given by the means of the soil characteristics of Lithofacies and erodability maps. The generated velocity map of the soil displacement revealed that the peak of soil movement time series happened after the reported flooding event.

Finally, we presented an insight into the deep learning capability to process the automatic classification of InSAR interferograms in the context of post-flood soil deformations.

Keywords :

flood mapping, post-flood landslide, SAR, InSAR coherence, InSAR interefrogram, LISFLOOD-FP, FCM, RF, P-SBAS, deep learning.