Abstract Floods are a very serious and frequent disaster occurring in many parts of the world. Mapping of inundated regions is crucial for determining the flood extent, deployment of emergency response teams, and assessment of damages and casualties. This thesis investigates flood mapping using Sentinel-1 time-series data in arid areas. The presented method aims to improve the flood classification results from Sentinel-1 Flood Service which are based on a single SAR image analysis. Water detection performed in arid regions derived from one SAR dataset is challenging because the backscatter of water is similar to sandy regions, leading to overestimations of flood extent. The main objective of the thesis is an assessment of the influence of time series on flood classification accuracy. This goal is accomplished by executing experimental tests on different statistical parameters, frequency classes and durations of timeseries data in the chosen areas of interest, which are in Somalia and Iraq. The results obtained from confusion matrices indicate enhancement in Overall Accuracy of ~5% and User’s Accuracy of more than 24%. Such an effort aims to advance the use of Sentinel-1 time-series data for arid areas and pave the way towards rapid flood mapping to support emergency management authorities. Keywords: Flood, Flood Mapping, Sentinel-1, Time Series, Arid Areas     «

Abstract Floods are a very serious and frequent disaster occurring in many parts of the world. Mapping of inundated regions is crucial for determining the flood extent, deployment of emergency response teams, and assessment of damages and casualties. This thesis investigates flood mapping using Sentinel-1 time-series data in arid areas. The presented method aims to improve the flood classification results from Sentinel-1 Flood Service which are based on a single SAR image analysis. Water detec...     »