<p>Real-time or near real-time prediction of coseismic landslide distribution is crucial for disaster response. The spatial distribution of ground motions, such as peak ground acceleration or seismic intensity, has been used as a trigger to make such predictions. However, prior research has reported that these ground motions are not often correlated with the actual landslide density. The sparse distribution of strong-motion seismographs makes this problem challenging to comprehend, leading to uncertainty in the spatial distribution of ground motions. In this study, we estimated the areal distribution of peak ground velocity (PGV) at a 1-Hz sampling rate (PGV_1Hz) using accelerometer and GNSS observation data for the 2016 Kumamoto Earthquake and investigated their relationship with coseismic landslide distribution. Consequently, the distribution map of PGV_1Hz estimated using accelerometer and GNSS observation data is better correlated with the landslide distribution than that estimated solely from accelerometer observation data. In the Kumamoto Earthquake, GNSS close to the epicenter fault or on the extension of the rupture direction of the epicenter fault made a substantial contribution. These findings suggest that the use of PGV_1Hz calculated using accelerometer and GNSS data yields more accurate estimates of the areal distribution of landslides.</p> Graphical Abstract <p></p>