测流不确定度的插值方差估计法

目前，测流不确定度通过误差试验或通过经验数值来确定，但这些方式存在着工作量大或不确定估计不足等局限性。为解决此问题，对基于实测数据和统计理论的插值方差估计法在不同测流条件下进行了验证，选取白河、襄阳和沙洋3个流量站进行了实测数据的不确定度分析，同时对白河站进行了Monte Carlo试验，比较插值方差估计法得到的不确定度与真实误差的差异。结果表明，插值方差估计法能较好地反映水位变化的影响，插值方差估计法所得到的不确定度与真实测流误差的相关系数达0.64，与断面水位变化的Spearman相关系数达0.79，高、中水位情况下插值方差估计法的不确定度估计结果较为合理，低水位情况下偏高。

Interpolated variance estimator for uncertainty of discharge measurement

Uncertainty of discharge measurement is often estimated by error tests or empirical and laboratory studies. However, those methods have limitations such as heavy workload and inaccuracy. A method named interpolated variance estimator, which is based on statistical techniques and on-site observations to estimate uncertainty in discharge measurement, is introduced and testified. Measured data at three stations namely Baihe, Xiangyang and Shayang along Hanjiang are analyzed. Besides, Monte Carlo test at Baihe Station is developed to investigate the difference between real error and uncertainty calculated by interpolated variance estimator. The results show that interpolated variance estimator can indicate very well the relationship between uncertainty in discharge measurement and water level. The correlation coefficient of uncertainty and real error derived in the Monte Carlo test is 0.64, and Spearman coefficient of uncertainty and water level is 0.79. Also, uncertainty for high and middle water levels is credible but is overestimated for low water level. These results illustrate that interpolated variance estimator, as a sound statistical method, can be recommended for use in broad applications.