Jamal Samani - Associate Professor of Water Structures Eng., Tarbiat Modarres University, Tehran, Iran
Mahsa Vaez Tehrani - Ph.D. Student of Water Structures Eng., Tarbiat Modarres University, Tehran, Iran.
Morteza Mojtahedi - Moderator of Water office, ABFA Co., Tehran, Iran

Due to different factors affecting flow discharge process and their randomness, uncertainties arise which cause failure and may force designers to consider safety factors in their designs. Therefore, uncertainty analysis may introduce a better understanding of the effect of different input parameters and their separate contributions to the whole output error. In this study data collected from Roodan River in Iran and Arun & Chayang Rivers in Nepal are used. Manning n-values for each study reach were estimated using Jarrett's Equation, Limerinos's Equation and Cowan Method and each set of n-values was then varied by and . In this research, Latin Hypercube Sampling (LHS) method was also investigated. To estimate the effect of various factors on the uncertainty of flow discharge, sensitivity analysis was conducted with correlation method and with and change in n-values. The results show that the roughness coefficient introduces the highest contribution to the whole uncertainty of flow discharge. In LHS, the uncertainties of flow discharge were 0.31 in Roodan River, 0.24 in Arun River and 0.25 in Chayang River. In most cases, varying the n-value by shows no change in the estimated discharge. Varying the n-values for the study reaches by generally introduces discharge variation by an average of 20-30%. In general, the discharge variation associated with varying n-values is proportional to channel gradient and to roughness, and inversely proportional to width/depth ratio. Therefore, with the obtained results, uncertainty in n may be translated into uncertainties in discharge. Both methods, in sensitivity analysis introduce almost similar results.

flow discharge, roughness coefficient, uncertainty, LHS, sensitivity analysis