基于PCA-Fisher最优分割法的汛期分期影响研究

摘要

Current flood season division methods are often subjective,it is difficult to make flood season staging easily and reasonably.Taking Shencun Reservoir basin as an example,the average value change-point method is firstly used to find out the correlation coefficient between monthly rainfall and runoff for quantitatively determining the flood season and non-flood season,and then,the weight of each index is determined by using Fisher optimal dissection method combined with principal component analysis (PCA) method.The staging rules of flood season are studied based on comprehensive index,flood index and rainstorm index,and the staging results are analyzed according to regional climate characteristics.The results show that PCA-Fisher optimal dissection method is feasible.The staging results in flood season are closely related to staging indexes,so a variety of indicators should be comprehensively considered in practical application.Because there is a lack of consideration of climate change and extreme weather in the determination of flood season and non-flood season,it is need to make an in-depth study on the impact of climate change on flood season.%针对传统汛期分析方法主观性强、研究论域单一等问题,以申村水库流域为例,首先采用均值变点分析法寻找月降雨-径流的相关系数列变点,定量确定汛期研究域;再应用Fisher最优分割法,结合PCA法确定各项指标权重研究了基于综合指标、洪水指标以及暴雨指标的流域汛期分期规律,并依据区域气候成因对分期结果进行分析.结果表明,基于PCA-Fisher最优分割法可行;汛期分期结果与分期指标有密切关系,在实际应用,应综合考虑多种指标;人为判定汛期研究域缺乏对气候变化和极端天气等因素的考虑,需对气候变化等因素对汛期的影响做深入研究.