Closure to 'Parameter Estimation of Extended Nonlinear Muskingum Models with the Weed Optimization Algorithm' by Farzan Hamedi, Omid Bozorg-Haddad, Maryam Pazoki, Hamid-Reza Asgari, Mehran Parsa, and Hugo A. Loaiciga

摘要

The main argument of the discussion is that the parameterized initial storage that was considered in the original paper did not consider the physical aspects of flood routing. The Muskingum method is commonly used for river routing. It involves several parameters, including the initial river-reach storage, S_0. These parameters represent the physical properties, and their values must be estimated for the purpose of flood routing. Hydrologic routing methods, including the Muskingum, estimate the model parameters from the input and output flow data in the river reaches. The original paper estimated the Muskingum parameters by posing them as unknown variables. An initial condition is needed to estimate the parameter S_0 (the initial river-reach storage), and for this purpose, the initial condition I_0 = O_0 is applied in the flood routing, in which I_0 denotes the initial reach inflow and O_0 denotes the initial calculated outflow (e.g., Chow 1959; Das 2007; Chu and Chang 2009; Orouji et al. 2012; Easa 2013; Hamedi et al. 2014; Bozorg-Haddad et al. 2015a). The authors intended to improve the initial condition to produce a well-calibrated Muskingum model in the original paper. Therefore, S_0 was estimated by the optimization algorithm in the original paper. The results in the original paper proved that the optimized nonlinear Muskingum flood-routing models NL1 (Chow 1959), NL2 (Gill 1978), NL3 (Easa 2013), and NL4 (Bozorg-Haddad et al. 2015a) performed better than the nonopti-mized versions. The optimization of S_0 in the original paper was constrained to obtain physically meaningful values (Mehrabian and Lucas 2006; Asgari et al. 2016).