Quels apports hydrologiques pour les modèles hydrauliques? Vers ...

∫ tO(t) = Ψ(t)+0with: Ψ(t) = L[I(t − τ) − Ψ(t)] K(τ)dτ (2)∫ t0q(t)dτ290This solution was applied by several authors (Baptista and Michel, 1989; Corradini,1991; Moramarco et al., 2005) to model propagation in natural rivertel-00392240, version 1 - 5 Jun 2009295300channels. It has two main limitations. First, the analytical solution of Equation1 is valid only for constant values of C and D, which reduces its precisioncompared to a fully dynamic model. Second, it does not handle downstreamboundary conditions and hence neglects all backwater effects.Compared to other simplified routing schemes, it has the following advantages.First, it accepts point and uniformly distributed lateral inflows, which is ofprimary importance in this study. Second, it can calculate flows along theentire reach allowing interior simulations to be evaluated (see Section 4.5).Finally, the convolution product makes model execution very fast. This greatlyfacilitates calibration of the coupled models that may require a significantnumber of iterations.305Although the linearised diffusive wave is based on the Saint-Venant equations,it is clear that this solution is not equivalent to a fully dynamic model. So itshould be necessary to check that similar results are obtained with refinedhydraulic models. This point will not be investigated in this paper. The workof Butts et al. (2004) provides an encouraging example on the usefulness ofsimplified routing schemes: they compared two routing procedures in a semidistributedmodel and found that a Muskingum-Cunge routing model wasequivalent to a full dynamic model. This suggests that the benefits of com-15