摘要：

Abstract:

【Purpose/significance】With the deepening of urbanization process and the increasing uncertainty of climate change, urban

flood disasters occur frequently. At present, there are many characteristics of urban flood disaster emergencies, such as complicated

flood disaster information, confusion of decision-making demand information and weak pertinence of response measures. The reason is

that decision makers at different levels cannot get valuable decision-making information. Therefore, it is particularly urgent to provide

different levels of decision makers with scenario information of appropriate granularity.【Method/process】Aiming at the characteristics

of rapid evolution and uncontrollability of the situation of urban flood emergencies, this paper decomposes the situation elements of

emergencies, introduces support vector machine and similarity algorithm, and constructs a dynamic response model based on the granu⁃

larity principle to realize scenario refinement from the high level decision-making to the grass-roots level.【Result/conclusion】The

model can quickly focus on the key nodes of event evolution, provide decision support for different levels of users, and has been effec⁃

tively verified by the flood disaster events in Changzhou City.【Innovation/limitation】By fine-grained scenario units of emergencies,

the decision-making needs of users at different levels are met and the emergency decision-making is scenarioized. However, the map⁃

ping relationship between scenario levels and the transformation algorithm still need to be further clarified.