Flooding is the most frequent natural hazard on Earth and affects an increasing number of people. Major events are responsible for huge loss of life and substantial destruction of infrastructure. Detailed information about the location, time, or extent of present and historic floods help in improving emergency response or planning of prevention actions. For this purpose, the new Global Flood Monitoring (GFM, https://gfm.portal.geoville.com) service provides satellite-based flood mapping information derived from Sentinel-1 Synthetic Aperture Radar (SAR) data in near-real time (NRT) on a global scale to the user community (Salamon et al, 2021). This service is part of the Copernicus Emergency Management Service (CEMS), and is available in its beta-version through the Global Flood Awareness System (GloFas, https://www.globalfloods.eu/). In order to improve the overall reliability of the flood mapping, three independent Sentinel-1-based algorithms are combined within one ensemble product. As basis for all activities within the GFM service, a global Sentinel-1 datacube has been created (Wagner et al, 2021). In the initial phase, more than 1.6 million Sentinel-1 scenes from 2015 – 2020 were preprocessed using the new 30m Copernicus DEM for terrain correction. The observations were resampled to a spatial gridding of 20m and are provided in a tiled and stacked image structure based on the Equi7Grid (https://github.com/TUW-GEO/Equi7Grid).This setup allows for an efficient extraction of spatiotemporal subsets. The Sentinel-1 datacube is updated in NRT to enable continuous flood monitoring. One of the algorithms going into the ensemble product is the algorithm developed by the Technische Universität Wien (TU Wien, https://www.geo.tuwien.ac.at/). The algorithm performs a pixel-wise decision between flooded and non-flooded conditions. The historic Sentinel-1 measurements of the datacube and derived temporal parameters allow to statistically describe the backscatter signature of both states. The water backscatter differs significantly from non-flooded land due to the specular reflection of the impinging radiation and the side-looking geometry of the SAR system. Contrary to water surfaces, the backscatter signals over non-flooded land are much more heterogeneous and most show strong seasonal variations. This seasonality is caused by variable factors within the signal like soil moisture or vegetation conditions. To parametrise the backscatter under non-flooded conditions, and by considering the backscatter’s seasonality, a harmonic regression model was found to be best suited (in particular for NRT operations). The model’s parameters were computed for each pixel of the Sentinel-1 datacube by a least-square estimation which made use of measurements from 2019-2020. Based on the resulting global parameter database and the underlying model, one is able to estimate the non-flooded backscatter for every day of the year. Using Bayes interference, the incoming Sentinel-1 scene is pixel-wise compared to the modelled backscatter signature of flooded and non-flooded conditions, and the more probable condition is then chosen. When working with SAR data, water-look-alikes like deserts, radar shadows, or tarmacs could be confused easily with inundated areas. Additionally, one is limited to areas where the Sentinel-1 signal is able to reach the ground undisturbed in order to distinguish between flooded and non-flooded situations. Consequently, areas which are densely vegetated or built-up areas need to be excluded as well as areas which permanently feature low backscatter. Therefore, we utilise exclusion layers that are derived from temporal parameters of the Sentinel-1 datacube. By masking the flood mapping results with the exclusion layers, potential uncertainties are avoided and the algorithm’s robustness is increased. In this contribution, we present the TU Wien Sentinel-1 flood mapping algorithm, which exploits the historic measurements of a dedicated Sentinel-1 2015-2020 datacube, and which is already integrated within the GFM ensemble approach. We evaluate the globally operated algorithm in representative sites of a set of world regions, highlighting its strengths and caveats. Additionally, we focus on the suitability of the Sentinel-1 signal history to exclude areas that show low sensitivity for flood mapping or could potentially be classified wrongly as flooded. Salamon et al. (2021) The New, Systematic Global Flood Monitoring Product of the Copernicus Emergency Management Service. In 2021 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), IEEE, pp. 1053-1056. Wagner, Wolfgang, et al. (2021) A Sentinel-1 Backscatter Datacube for Global Land Monitoring Applications. Remote Sensing 13.22 , 4622.