Ferramenta de auxílio à emissão de alerta de inundações urbanas: metodologia associando as características das precipitações com a probabilidade de ocorrência de alerta

Abstract

Urban floods occur when a natural or artificial channel can no longer accommodate increased discharges caused by storm events and overflows to its banks affecting areas occupied by the population. Although floods are natural events, these disasters in an urban environment are directly related to the current social and economic development system Thus, its management requires an interdisciplinary and intersectoral approach to the formulate policies for prevention, mitigation and response to future disasters. Early warning alert systems are important management tools for consolidated urban areas, allowing a potential reduction of the damage caused by floods. However, the effectiveness of this system is influenced by several, including the available time to send the warning prior to the event, reliability of the alert and expertise of the involved personnel to perform the correct course of action. Given that the reliability of the early warning alert system is extremely important for its effectiveness, this research presents different approaches whose objectives are to guarantee this reliability through the improvement and understanding of the numerous steps involved in this process. The main objective of this work is to present a methodology that allows the correlation between the rainfall characteristics and the probability of alert occurrence, serving as a tool to support decision-making by the competent authorities given a rain forecast. To achieve this goal, synthetic rainfall datasets, generated by a stochastic generator were simulated in the SWMM model, enabling the development of a correlation between the duration and total precipitation volume with probability of occurrence of the alerts. Comparison of these relations with the councils alert records, evidence the tendency of the simulated results to underestimate the probability of the alert. This result shows the difficulty to represent the peak levels accurately in the SWMM calibration and validation process. Although the results obtained in this research are not satisfactory from the operational point of view, they are extremely promising from the methodological perspective. Therefore, the analysis presented here can be applied for other regions and hydrological and hydraulics models