Influence of rainfall spatial variability for the hydrological modelling of small catchments: a simulation approach

The present work aims to quantify the effects of neglecting rainfall spatial variability for runoff modelling at the outlet of catchments ranging from ten to several hundred km(2) (of urban or peri-urban type). We have proceeded by simulation in order both to overcome modelling and measurement errors, in addition to controlling rainfall variability and the characteristics and hydrological behavior of catchments. For this purpose, it has been used a simulation chain including a stream network model, a rainfall simulator and a distributed hydrological model (with four production functions and a distributed transfer based on the Hayami model). It has also been decided not to conduct an exhaustive investigation but instead a study of contrasting situations. Results showed the difficulty to obtain general conclusions from studying just a few events and thus contribute to explain the divergent conclusions drawn in the literature. Moreover, the proposed methodology revealed a number of trends. For the different studied rainfall configurations, the effects of neglecting rainfall spatial variability appear to be significant in situations where rainfall contrasts are created and maintained over the catchments. Such situations were identified for three scenarios: 1) the dimensionless "rainfall field velocity to catchment flow celerity" ratio is close to 1; 2) the rainfall field direction is perpendicular to the main catchment flow direction; and for 3) a Horton or SCS production function. At last, in order to verify that simulation results are close to reality, all results will have to be confronted to those obtained on real catchments with some real rainfall measurements.