Threshold and real-time initiation mechanism of urban flood emergency response under combined disaster scenarios

Scientific and reasonable emergency response initiation mechanisms can provide important support for decision making regarding the emergency management of urban floods. However, there is a lack of a unified paradigm on how to calculate the threshold for emergency response initiation and reasonably initiate emergency response. Therefore, this study proposes a loss-driven urban flood emergency response initiation framework from the perspective of combined disasters. A discrimination mechanism of the emergency response initiation level was established based on the optimal threshold and loss function. And the rainfall event that occurred in Zhengzhou, China, on July 20, 2021, was taken as an example to realize real-time emergency response discrimination and initiation driven by forecast data. Results showed that the initiation time of the Level I emergency response using the proposed method was 9.5 h earlier than the time of the government release, thereby significantly increasing the preparation time for flood management personnel. In addition, the results of the optimal threshold selection indicated that the Natural Breakpoint method was the optimal method for loss threshold partitioning, with the comprehensive evaluation index (CEI) being 3.56–9.53 % higher than those of the K-means, Equal Interval, and Quantile method. These results constitute a reference for urban emergency management and related research.