对洪水形态进行准确的识别,是分析不同洪水形态形成原因的基础,也为防洪工程提供有效的信息。本文从一般性指标、洪水时域、频域角度选取13项与洪水形态相关的指标,依据主成分分析法进行降维处理,从中提取出6项指标,采用验证为具有很好分类效果的投影寻踪模型,来验证所提取的6项指标的有效性。结果表明:1)利用投影寻踪模型,结合6项洪水形态指标,进行洪水形态识别的有效性;2)武江流域的双峰型洪水占39.62%,且均为主峰偏后型,这对武江流域的防洪工程显得尤为不利。本研究不仅为洪水形态指标的提取、洪水形态的识别提供参考,也为武江流域的防洪提供了理论指导。
Morphology of flood is an important feature to classification and recognition of flood and flood control. Six quantitative parameters can be extracted for description of flood morphology in time and frequency domains. This paper uses a projection pursuit model to confirm the validity of classification and recognition methods of these parameters. The results indicate that the method of combining this model with the six parameters is very effective in correct recognition of various flood shapes. In a case study of Wu River (Guangdong) , it is revealed that two-peak floods account for 39.62% and all such floods have a prominent second peak, showing an adverse case to flood control. The results are useful for extracting flood shape parameters and provide a theoretical basis for flood control of this river.