The water erosion modelling in small catchments is developed on the distributed runoff model, by their coupling at the raster scale and 10 min time intervals.

In distributed runoff modelling, the output hydrograph is estimated by integrating space and time variables of the water balance components over the catchment area. The basic idea is that only taking into account in the computing process of a larger number of variability factors, the estimation accuracy may be increased, due to the use of more proper averaging scales of the involved variables. The basin digital terrain model stands for the numerical integration domain of water depth values, whereas physically-based models of the water balance components may be easily tailored to the local terrestrial peculiarities by means of a constant parameter set has been verified on several events. The time dependence is given by two infiltration parameters characterising the soil moisture dynamics as short and long term behaviours, being estimated in terms of the previous rainfalls' occurrence.

The rain erosivity at the impact with the land surface is estimated by means of a linear function of 10 min rainfall intensities, where the scale parameter has a time evolution exponential law due to a soil moisture effect. The sediment source is a distributed variable according to the area variability of rainfall input and interception losses, and a potential amount for the sediment transport.

The sediment transport rate is a two-parameter parabolic function in terms of local water discharges. Under the circumstances, the usual high non-linearity of the process at the catchment scale is reduced at similar shapes like the runoff one, being dependent on a combined scale parameter of the above processes, to be updated at the analysed event by means of a time constant in the exponential function and the initial state according to the soil short term behaviour. The fixed parameters may be easily estimated from simulated successive events.

In framework of the local sediment balance, the process non-linearity may lead to bed erosion in case of a higher output than input, and, otherwise to sediment deposition. The sediment dynamics is also available in framework of area estimates all over the computing period at each 10 min time step either in case of washed materials from the potential amount, or as cumulative values in the transport process.

Being highly connected to the catchment land cover and topography, reasonable predictions may be performed at the event scale for hydrologic effects being generated by different land cover changes, as well as long term estimates due to an acceptable accuracy the time evolution of the soil system may be described.

The numerical examples in the paper refer to simulations of successive events, predicted effects on runoff and suspended load hydrographs at scenarios for watershed afforestation and cumulative effects on a long period to be compared with two topographic maps at a 25 years' interval.