The Site-Specific Early Warning System (SS-EWS) follows an impact-based approach which has been advocated by the WMO to properly communicate information to the end-users regarding what the weather will do, rather than what the weather will be. This study focuses on the design and real-time evaluation of the Site-Specific Early Warning System Framework in Terrassa City where Urban Flooding has been prominently observed. The four key components of the SS-EWS have been established through a complete analysis of the components of risk (haszard, vulnerability, exposure) and the utilization of all the available products to trigger impact-based flood warning levels. These flood warning levels have been established in three new points, namely: Av. Joaquim de Sagrera, Av. Del Valles, and Underpass (Av. Del Valles), and brings the total number of vulnerable locations to six. The six vulnerable points have been evaluated in real-time, with the objective of studying the performance of all the flood warning levels and attempting to improve the Real-Time Activation component of the SS-EWS by using all the pixels within the catchment area of the vulnerable location, rather than only the surrounding pixels. The real-time evaluation also gave us an opportunity to explore how temporal changes to vulnerability, in the form of traffic data, can be integrated into the SS-EWS. Through the real-time evaluation, we have concluded that utilizing all the pixels within the catchment area resulted to a higher probability of false alarms, due to the inherent uncertainty of estimates by the nowcasting products (proportionate to the number of pixels used), and by the possibility that rainfall far from the location can still be diverted to other areas of the city by the drainage network. This has led to impacts that were not observed in real-life, even though flood warning levels have been issued. These inherent uncertainties were found to be highly affected by the type of rainfall, which demands for a larger sample of evaluated events for a 3 more representative analysis with regards to the effects of the rainfall type to the products, and eventually, the warning levels. Our evaluation also showed that we cannot conclusively say if there is a correlation between traffic data and rainfall because data was not recorded during rainfall events. Finally, the evaluation also showed a possible recalibration of the thresholds for Av. del Valles and its underpass since impacts of lower warning levels were observed, than what was issued. A continuous real-time evaluation is needed to conclusively recalibrate the thresholds to properly match the warning levels with the observed impacts.
gmail.com
