Post-War Urban Damage MappingUsing Optical and Radar Data: the case of Douma, Syria

Sammanfattning

Urban infrastructure is vital in maintaining societal stability. Its destruction duringarmed conflicts creates critical challenges for post-war reconstruction and planning.This work assesses the effectiveness of Sentinel-1 radar data in detecting structuralchanges in destroyed urban environments, comparing the results with Sentinel-2optical data, focusing on Douma within Rif Dimashq (Rural Damascus) as the studyarea. The value of this work lies in the need for reliable, free, and rapid methods indifficult-to-access areas after the war, where traditional assessments, such as fieldsurveys, lack temporal accuracy. This work is based on the Delta coherence (∆C)analysis approach using Sentinel-1 data processed in SNAP, using visually verifiedcalibration reference samples from Google Earth. In parallel, Sentinel-2 data wereanalyzed using the NDBI index in ArcGIS Pro to examine spectral changes andcompare performance. The results’ accuracy was evaluated using UNOSAT data as areference dataset. The results indicate that the delta coherence model, compared tothe model based on absolute coherence, achieved a statistically significantimprovement in detecting structural damage with an overall accuracy of 61.5% andan F-score of 0.73. On the other hand, Sentinel-2 data achieved an overall accuracyof 61.1%, but with a clear overlap between damaged and undamaged areas and ahigher number of false positives, since they respond to general surface changes ratherthan structural changes. Analyses show that the most reliable source of data formonitoring structural damage is Sentinel-1, while Sentinel-2 data contribute to theexploration and understanding of general change patterns.Consequently, the combination of these sources provides a more integrated andcomprehensive approach to assessing urban areas affected by conflicts and disasters.