Our study uses a dense temporal stack of 78 Landsat 8 images for surface water extraction using automatic Earth Observation (EO) image pre-processing, coupled with analyses over time for flood detection. The analysis is conducted with our IQ (ImageQuerying) system developed in-house, which allows ad-hoc executing of spatio-temporal queries against semantically enriched EO images. To facilitate high performance analyses, the data are stored as a spatio-temporal data cube in an array database. The analyses are automatically-translated database queries, which increase reproducibility, readability and comprehensibility for a human operator and can be conducted within just a few minutes. The specific analysis for this contribution is based on flood-extent mapping over different user-definable time spans. The results indicate areas that have been flooded at least once in the selected time span and are therefore prone to being flooded in future events. Additional spatial queries (e.g., for the indication of cloud cover) support the quality assessment of the flood analyses. We compared our result with a flood mask derived from a SAR (synthetic aperture radar) image of a single event in Somalia (Hiran province). Larger flooded areas overlap in both analyses, despite the non-synchronous acquisition times of the images. The results can be used as input for improved risk assessment and management of floods.