摘要
针对采用普通最小二乘法对水文模型进行参数估计时,模型残差需满足一定的内在统计假定的要求,分别选用Levene检验、Kolmogorov-Smirnov检验和残差自相关系数图等方法对水量平衡模型WASMOD的模型残差的同方差性、正态性以及相互独立性等统计假定进行了检验。结果表明:当对原始数据未经变换而直接采用普通最小二乘法进行参数估计时,得到的模型残差满足相互独立假定,但并不满足同方差性和正态性假定;对原始流量数据进行开根号变换,可以很好地解决模型残差的异方差性和非正态性问题。在模型残差统计假定得以满足的条件下,在月时间尺度上WASMOD模型可以为莺落峡流域的流量模拟与预报提供良好的工具。
When the parameters for hydrological models are estimated by using the ordinary least square method, the model residuals should satisfy certain statistical assumptions. The Levene test, Kolmogorov-Smirnov test and autocorrelation coefficient are employed to check the statistical assumptions of model residuals of the conceptual lumped Water And Snow balance MODeling system （WASMOD） such as homoscedasticity, normality and independence features. As the original discharge values without any transformation are calculated by using the ordinary least square method, the yielded model residuals are proved to be independent, which greatly deviate from the features of homoscedasticity and normality. However, a square root transformation for the discharge values is capable of solving the problems of heteroscedasticity and non-normality in model residuals. The proposed WASMOD model may be satisfactory tool for the simulation and predication of discharge in Yingluoxia watershed on monthly temporal scale when the statistical assumptions of the model residuals are satisfied.
出处
《水利水电科技进展》
CSCD
北大核心
2013年第1期13-17,共5页
Advances in Science and Technology of Water Resources
基金
国家自然科学基金(41101038)
中央高校基本科研业务费专项(2011YXL038)
关键词
WASMOD
参数估计
普通最小二乘法
模型残差
水量平衡模型
莺落峡
WASMOD
parameter estimation
ordinary least square method
model residual
water balance model
Yingluoxia watershed
