基于GAMLSS模型的玛纳斯河设计年径流分析

玛纳斯河年径流在气候变化影响下发生了变异,传统径流频率分析方法的一致性假设遭到破坏,本文根据玛纳斯河上游出山口1956—2014年水文气象资料,采用遥相关分析法,并基于GAMLSS理论分别建立以时间、气候因子为协变量的时变矩模型,将其拟合效果与传统P-III分布进行对比分析,利用最优分布模型进行年径流量设计。结果表明：北大西洋涛动指数（NAO）作为气候影响因子与年径流序列相关系数为-0.322,遥联性最佳;以累积气温亏损值、降雨量、NAO为协变量的LOGNO分布模型为最优分布模型,有效地描述了在气候变化影响下玛纳斯河年径流动态变化特征,并在不同设计保证率下设计年径流比传统P-III分布偏大3.08%~16.10%,各月径流设计值与P-III分布相差较大。其研究结果为玛纳斯河水资源的高效利用及科学管理提供参考依据。

Analysis of design annual runoff of Manas River based on GAMLSS model

Considering the premise of consistency hypothesis is no longer satisfied for the traditional runoff frequency analysis methods, so the Generalized Additive Model of Location, Scale and Shape (GAMLSS) was employed to detect the non-stationarity of the annual runoff series (1956-2014) in Manas River and correlation analysis method was used to build time varying moment models based on time and climatic factors as covariate variables respectively, whose effects were compared with the traditional model (P-III). The results show that: NAO index could be as one of the climatic factors because of its best relativity with annual runoff serials and the correlation coefficient is -0.322; the optimal distribution model which is based on the accumulated temperature deficit, precipitation index and NAO as covariate variables effectively describes dynamic characteristics of annual runoff in Manas River under climate change. Under different design guarantee rates, annual design runoff by the LOGNO model is larger than the traditional model (P-III Model0) by 3.08%-16.10%. Particularly, the monthly runoff design value is rather different from Model0. The research results provide reference for the efficient use of resources and scientific management of Manas River.