Author(s): Anabela Oliveira; Goncalo De Jesus; Joao Rogeiro; Joao Nuno Fernandes; Rui Rodrigues

Linked Author(s): Anabela Oliveira, João Rogeiro, João Nuno Fernandes, Rui Rodrigues

Keywords: Real time data; Flood forecast; Hydraulic modelling; Machine learning-based simulations; High performance computing

Abstract: Flood forecasting in small watersheds is a complex problem, requiring the integration of both real-time data and models to produce inundation maps and timely warnings to avoid human lives and material losses. The Water Information Forecast Framework (WIFF, Fortunato et al., 2017) addresses this problem by integrating seamlessly computational models that simulate all relevant processes at the adequate time and spatial scales, based either on numerical representations of these processes or on data-based models. This framework can be applied from the rivers to the sea, being already validated through the successful implementations to urban, river, estuarine and coastal flood forecast (Gomes et al., 2017). Currently, it takes advantage of the SCHISM and SWMM models, implemented in high performance computing environments at several European Open Science Cloud providers. WIFF offers multiple choices of atmospheric forcings available from NOAA, METEOFrance and MeteoGalicia, with outputs from models GFS, Arpege and WRF at several time and space scales. Herein, we present the latest developments in WIFF that address floods in small watersheds, where retention time is short and routing of flood waves is on the order of tens of minutes. The availability of predictions is therefore required at very short time scales, making a hybrid process necessary to convey accurate alerts in due time. In this process, WIFF executes two procedures in parallel. First, a large-scale approach, based on conventional numerical models, runs continuously everyday, to detect significant rain events. If a predicted rain event crosses a warning threshold, a second approach is triggered, involving a small-scale data-based model to predict flooding for the following hours, based on real time monitoring networks data and on the use of high performance computing for machine learning-based simulations. For the first step, we are updating the framework to integrate both hydrological and hydraulic models of the HEC model family (Brunner, 2021). This methodology will be validated in the Ribeira das Vinhas basin, an area prone to torrential floods that inundate the urban area of the city of Cascais, located at the Tagus estuary mouth. Brunner, P.E. (2021) HEC-RAS 6.0, Hydraulic Reference Manual, May, Hydrologic Engineering Center, US Army Corps of Engineers Fortunato; A.B., A. Oliveira; J. Rogeiro; R. T. Costa; J. L. Gomes; Kai Li ; G. Jesus; P. Freire; A. Rilo; A. Mendes; M. Rodrigues; A. Azevedo. (2017). Operational forecast framework applied to extreme sea levels at regional and local scales. Journal of Operational Oceanography, Volume 10/1, DOI: 10.1080/1755876X.2016.1255471 Gomes J.L., Jesus G., Rogeiro J., Oliveira A., da Costa R.T., Fortunato A.B. (2017) An Innovative Web Platform for Flood Risk Management. Advances in Intelligent Systems and Computing, vol 461. Springer, https://doi.org/10.1007/978-3-319-44354-6_13

DOI: https://doi.org/10.3850/IAHR-39WC252171192022737

Year: 2022