The spatial step size for the second-order accurate Muskingum-Cunge (M
-C) method is determined by the spatial weighting factor. Both the spa
tial weighting factor and the time step must be selected judiciously t
o obtain accurate solutions. In this study, accuracy criteria for the
linearised diffusion routing problem are discussed. Starting from the
truncation error analysis of the general finite-difference scheme used
to solve the kinematic wave equation (of which the M-C method is a sp
ecial case), conditions necessary to obtain second-, third- and fourth
-order accurate solutions to the linearised diffusion routing equation
are derived. For given diffusion coefficient and celerity, the spatia
l step of the fourth-order scheme is fixed, whereas third- and second-
order solutions are available for independently selected spatial steps
. In order to achieve optimal solutions to the second-order accurate s
cheme, the truncation error criteria are combined with a condition der
ived from the concept of column holdup. This combination is shown to p
roduce results as good as those from the third-and fourth-order accura
te schemes. The simplest explicit method is shown to give satisfactory
results for flood data from the River Wye (UK).