Atmospheric motion vectors (AMVs) derived from 5-min rapid scan (RS) imagery of the Multi-functional Transport Satellite are expected to capture small-scale distributions of airflows better than typical AMVs derived from 30-min imagery because the observation interval of RS-AMV is shorter. The impact of these high-frequency data on the numerical forecasting of a heavy rainfall near a stationary front was investigated by conducting data assimilation experiments. As a part of preparation for the assimilation, RS-AMVs were compared with the first-guess field obtained from the Japan Meteorological Agency (JMA) nonhydrostatic model (NHM). The comparison result indicated that the RS-AMVs were of good quality and could be used in the JMA’s operational NHM with 4D variational data assimilation (JNoVA). Assimilation experiments investigating a heavy rainfall event were conducted using different lengths of assimilation time slot and time intervals of spatial thinning for the assimilation of the RS-AMV data. The assimilation of RS-AMVs caused the initial wind fields to enhance the upper-level divergence and low-level convergence around the front. Consequently, the forecast of the rainfall amount was increased near the front, and the verification scores were slightly improved over the control experiment in the early forecast hours.