基于GAMLSS模型的大渡河流域极值降水非一致性分析

以GAMLSS模型为基础,分析大渡河流域降水频率的非一致性特征,并将时间和多项气候因素作为解释变量进行参数拟合。结果表明,大渡河流域年最大日降水序列均呈现不显著变化的趋势,与一致性模型相比,大渡河流域降水序列的时间非一致性的特征不明显,而气候指标可以更好地反映大渡河流域降水非一致性的特征,表现为所有气象站点GAIC值与一致性模型的值相比明显减少。GAMLSS模型可以模拟水文序列的方差、偏态系数等其他统计参数非线性变化趋势,能够反映水文序列的非一致性特征。

GAMLSS model-based analysis on nonstationarity of extreme precipitation in Daduhe River Basin

Based on GAMLSS model，the nonstationary characteristics of the precipitation frequency within Daduhe River Basin is analyzed herein，and then the parameter fitting is carried out by taking the time and several climate factors as the explanatory variables. The result shows that all the series of the annual maximum daily pricipitation present the trends of non-significant change，and compared with those from the stationary modelling，the nonstationary characteristics of the time serieses of the precipitation within Daduhe River Basin are not obvious，while the climate indexes can better reflect the nonstationary characteristics of the precipitation therein，which is presented as that if compared with the values from the stationary modelling，the GAIC values from all the meteorological stations concerned are obviously reduced. Generally，GAMLSS model can simulate the non-linear changing trends of the other statistic parameters of the hydrological series，such as variances，deviation coefficients，etc. and then refelct the nonstationary characteristics of rhydrological series.