The present work aims to quantify the effects of neglecting rainfall spatial variability for runoff modelling at the outlet of catchments of about ten to several hundred km². In order to overcome modelling and rainfall data errors, to control the rainfall variability as well as characteristics and hydrological behavior of catchments, we have proceeded by simulation. For this, a simulation chain has been used, including a stream network model, a rainfall simulator and a distributed hydrological model (with four production functions and a distributed transfer). The choice has been made not to be exhaustive but to study very contrasted situations. Results showed that it is difficult to obtain robust general conclusions by studying only a few rainfall events which might explain that different conclusions were drawn in previous case studies. However, some tendencies were observed. For the different studied rainfall configurations, the effect of neglecting rainfall variability is particularly important if: 1) the dimensionless ratio 'velocity of rainfall field on celerity along flow path' is small, 2) the rainfall field direction is perpendicular to the catchment flow direction, 3) the catchment has an elongated shape and 4) an Horton or SCS production function. Stream network model
