نوع مقاله : پژوهشی
نویسندگان
1 دانشجوی دکتری، دانشگاه تربیت مدرس
2 هیات علمی/دانشگاه تربیت مدرس
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
In order to analyze the uncertainty of the geometric characteristics of the river on the hydrodynamic model, the uncertainty of the river cross-sections was investigated for a hypothetical example and a real river. Accordingly, taking into account 5%, 10% and 20% errors in the harvested points and ±6%, ±3% and 0 errors in the entire section, 15 scenarios were defined for generating new cross-sections. In order to investigate the effect of the selected distribution in generating random points on the output results, the random points of each section in each of the proposed scenarios were generated once by normal distribution and once again by uniform distribution. Five statistical indicators were used in river analysis mode and section by section analysis to analyze the two characteristics of the flow output, i.e. velocity and cross-sectional area. The results indicate that with the increase of the error in picking the points of each cross-section, the thickness of the 95% confidence interval, the coefficient of variation, the dispersion index and the result of dividing the actual value of each characteristic by the deviation from the criterion of that characteristic in different repetitions for both river and cross-sectional conditions increases. This is despite the fact that increasing the error of the entire section does not change much in the output of the results. By comparing the results of the two mentioned distributions, the uncertainty indices in the scenarios implemented with a uniform distribution show more dispersion.The result of dividing the thickness of the 95% confidence interval by the standard deviation of the data in each scenario in both distributions is also around a constant axis with little fluctuation in change. The results of the investigation of two types of systematic and random errors showed that the change of systematic error does not impose uncertainty on the output of the model. Also, in the case of normal error distribution, with the increase of the percentage of error, the statistical indicators change in such a way that the statistical indicators do not undergo unacceptable fluctuations up to the 20% error that was investigated in this study
کلیدواژهها [English]
)