Abstract Robust estimators are parameters insensitive to the presence of outliers. However, they presume the shape of the variables’ probability density function. This study exemplifies the sensitivity of turbulent quantities to the use of classic and robust estimators and the presence of outliers in turbulent flow depth time series. A wide range of turbulence quantities was analysed based upon a stepped spillway case study, using flow depths sampled with Acoustic Displacement Meters as the flow variable of interest. The studied parameters include: the expected free surface level, the expected fluctuation intensity, the depth skewness, the autocorrelation timescales, the vertical velocity fluctuation intensity, the perturbations celerity and the one-dimensional free surface turbulence spectrum. Three levels of filtering were utilised prior to applying classic and robust estimators, showing that comparable robustness can be obtained either using classic estimators together with an intermediate filtering technique or using robust estimators instead, without any filtering technique.

中文翻译：

---

自由表面湍流特征的稳健估计器：阶梯式溢洪道应用

摘要 稳健估计器是对异常值的存在不敏感的参数。然而，他们假定变量的概率密度函数的形状。本研究举例说明了湍流量对使用经典和稳健估计器的敏感性以及湍流深度时间序列中异常值的存在。基于阶梯式溢洪道案例研究，使用声学位移计采样的流动深度作为感兴趣的流动变量，分析了广泛的湍流量。研究的参数包括：预期自由表面水平、预期波动强度、深度偏度、自相关时间尺度、垂直速度波动强度、扰动速度和一维自由表面湍流谱。