High resolution rainfields are a prerequisite to many hydrometeorological studies. For some applications, the required resolution may be as fine as 1 km in space and 5 minutes in time. At these scales rainfall is strongly intermittent, variable in space and correlated in time due to the propagation of the rainy systems. The paper compares two interpolation approaches to generate high resolution rainfields from rain gage measurements: i) a classical interpolation technique that consists in interpolating independently the rain intensities at each time step (Eulerian Kriging) and, ii) a simple dynamical interpolation technique that incorporates the propagation of the rainy systems (Lagrangian Kriging). For this latter approach, three propagation models are tested. The different interpolation techniques are evaluated over three climatically contrasted areas in West Africa where a multi-years 5-minute rainfall dataset has been collected during the AMMA (African Monsoon Multidisciplinary Analyses) campaigns. The dynamical interpolation technique is shown to perform better than the classical approach for a majority of the rainy events. The performances of the three propagation models differ from one another depending on the evaluation criteria used. One of them provides a satisfactory time of arrival of rainfall but slightly smoothes the rain intensities. The two others reproduce well the rain intensities but the time of arrival of the rain is sometimes delayed. The choice of an appropriate propagation algorithm will thus depend on the operational objectives underlying the rainfield generation.