Identification Problem of Open-Channel Friction Parameters

Abstract

Values of calibration parameters embedded in open-channel flow equations normally are ascertained using observation data that often contains Gaussian noise. These values subsequently suffer from induced errors even if they are determined by using optimization methods. If these parameters are to be considered a reliable representation of their true values, investigations are needed to establish their response to a range of factors, such as data errors, through statistical methods. A systematic study was carried out by a set of synthetic database for effective control of test conditions and by using typical flood events. The investigations showed that the identified parameters could be affected by such factors as data error, objective function, or gauge sites. However, through a better understanding of the behavior of the induced errors the mean of the identified parameters was found to lie within a prescribed confidence interval that contained the true value, even in the presence of high noise levels. Selection of objective function was found to be prone to undue biases affecting the identified parameters, which could be avoided through a careful consideration of the problem.