We developed a prototype system of the very short-range forecast of precipitation in Vietnam by merging kinematic extrapolations of composite hourly rainfall analysis and NWP, verified its performance for the case of a heavy rainfall event in July 2021 over central Vietnam. First, we produced hourly composite rainfall analysis over Vietnam with a grid distance of 5 km using AWS, radar, and satellite data. Next, we computed lag correlations between two hourly rainfall intensities at specific templates of 50 × 50 grids, and obtained lag indexes that maximize the lag correlation at 11 × 10 points. The moving vectors of precipitation areas at all grids are obtained by Cressman interpolation of the lag indexes, and a quality check using NWP horizontal winds at 700 hPa level was applied. Kinematic extrapolation of rainfall analysis was conducted using the above moving vectors and was merged with hourly rainfall prediction by a regional NWP model at NCHMF of VNMHA (WRF3kmIFS-DA) by weighted averaging. The magnitude of weight for the NWP in the merger was linearly increased from 0 to 1 for FT = 2 to 6 (from 03 UTC to 07 UTC, 12 July 2021). Verifications showed that the merged rainfalls outperformed both NWP and kinematically extrapolated precipitations for the time range of FT = 3 to 5.