Abstract:

In order to improve efficiency of typhoon case retrieval method, spatio-temporal nature of typhoon hazards and importance differences regarding case attributes and information integrity were studied. Spatio-temporal trajectory similarities were measured using FastDTW algorithm, and a set of historical cases similar to target typhoon path was established. To highlight importance differences between case attributes, an optimized model of CCSD method was used to determine attribute weights. Subsequently, case similarity was obtained by aggregating that of attributes. Then, considering spatio-temporal trajectory similarities and case similarities based on disaster information, the most similar case was retrieved from the set of historical cases, and an emergency plan regarding target typhoon trajectory was generated. Finally, the method was applied to real examples. The results show that the proposed method retrieves the most similar historical case by measuring spatio-temporal trajectory similarities of typhoons, determining attribute weights and synthesizing resulting information to ensure that the generated emergency plan is comprehensive and accurate with respect to the target typhoon.

Key words: fast dynamic time warping(FastDTW), case retrieval, typhoon, emergency plan, spatio-temporal trajectory similarity, correlation coefficient and standard deviation (CCSD)