HYDROLOGICAL PROCESSES Hydrol. Process. 24, 798–799 (2010) Published online 17 February 2010 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/hyp.7511 Comment Comment on ‘Sivapragasam C, Maheswaran R, Venkatesh V. 2008. Genetic programming approach for flood routing in natural channels. Hydrological Processes 22: 623–628’ A. H. Alavi, 1 A. H. Gandomi 2 * and M. Gandomi 3 1 College of Civil Engineering, Iran University of Science & Technology, Tehran, Iran 2 The Highest Prestige Scientific and Professional National Foundation, National Elites Foundation, Tehran, Iran 3 School of Civil Engineering, College of Engineering, University of Tehran, Tehran, Iran Received 29 January 2009; Accepted 22 September 2009 The discussers wish to thank the authors for examining the potential of the application of genetic programming (GP) for constructing a routing model for a channel reach along the Walla Walla River, USA. The discussers would like to present the following important viewpoints, which the authors and potential researchers need to consider. The discussion shall focus on main points that are not considered in the study. Descriptions given on page 625 and Figure 2 of the paper studied by Sivapragasam et al. (2008) clearly indicate that the method utilized for constructing a routing model for a channel reach is a tree-based genetic programming (TGP) approach. TGP was introduced by Koza (1992) as an extension of the genetic algorithms, in which programs are represented as tree structures and expressed in the functional programming language, LISP (Koza, 1992). Besides the traditional tree-based representations, there are linear and graph representations (Banzhaf et al., 1998; Poli et al., 2007). According to the last paragraph of descriptions of GP on page 625, the software package Discipulus  , which is developed by Conrads et al. (1998), was applied to the flood routing in natural channels problem. Discipulus  is a machine-code-based, linear genetic programming (LGP) software (Deschaine and Francone, 2002; Fran- cone and Deschaine, 2004; Francone et al., 2005; Lang- don and Banzhaf, 2005). LGP (Brameier and Banzhaf, 2007) is a subset of GP that has emerged, recently. Comparing LGP to the traditional Koza’s tree-based GP, there are some main differences. LGPs have graph- based functional structures and evolve in an imperative programming language C/CCC (Brameier et al., 1998) * Correspondence to: A. H. Gandomi, The Highest Prestige Scientific and Professional National Foundation, National Elites Foundation, Tehran, Iran. E-mail: a.h.gandomi@gmail.com and machine code (Nordin, 1994) rather than in expres- sions of a functional programming language like LISP (see Figure 1). Unlike tree-based GP, structurally non- effective codes coexist with effective codes in LGPs (Brameier and Banzhaf, 2007). Because of the imperative program structure in LGP, the non-effective instructions can be identified efficiently. This allows the correspond- ing effective instructions to be extracted from a pro- gram during runtime. Because, only effective programs are executed, evaluation can be accelerated significantly. Automatic Induction of Machine code by Genetic Pro- gramming (AIMGP) is a particular form of LGP. In AIMGP evolved programs are stored as linear strings of native binary machine code, which are directly executed by the processor. The absence of an interpreter and com- plex memory handling results in a significant speedup in AIMGP execution compared to TGP. This machine-code- based, LGP approach searches for the computer program and the constants at the same time (Nordin, 1994). Utilization of Discipulus  software confirms the essen- tial fact that the approach employed by Sivapragasam et al. (2008) is machine-code-based, LGP not TGP. Fur- ther information on basic LGP operators and their differ- ent types can be found in (Brameier and Banzhaf, 2007; Gandomi et al., 2009a). A popular modularization concept in LGP is the evolution of program teams (Brameier and Banzhaf, 2001). A team solution is formed by an uneven number of programs, of which every program has one vote. Because a team solution, in general, performs better than a single solution (Francone and Deschaine, 2004; Brameier and Banzhaf, 2007; Gandomi et al., 2009b), investigating the prediction qualities of team solutions can be an interesting topic for future work. Considering the above arguments, this study suffers from straightforward application of LGP without insight Copyright  2010 John Wiley & Sons, Ltd.