基于相似性理论的洪水超前预报预警研究.

Predicting the future evolution of the flood according to the similar inlor^nation in historical rainstor^n and flood is an effective way to realize advance flood warning and prediction. This paper developed a multi-index similarity measure method for rainstor^n based on “ quantity” and “type” similarity indexes, the similarity index of rainstor^n distribution and the earth mover's distance (EMD) ； searched the similar rainstor^n process and its corresponding typical flood process in historical records and then scaled the typical flood process according to the ratio of rainfall amounts to achieve the advance waning and forecast of flood. The ideal rainstorm series generated from an actual rainstorm process were applied to validate the proposed approach, proving its rationality； and this paper used the method in Xinmiao Station of Kuye River, located in the middle Yellow- River. The results show that the proposed flood forecasting method based on rainfall similarity analysis is effective and feasible. [ABSTRACT FROM AUTHOR]