Abstract Observations of synoptic variability from CTD and current meter measurements in Wakasa Bay, Japan in summer of 1980 and 1981 are compared with the results of 1979 reported by Yamagata, Umatani, Masunaga and Matsuura (1984). It is suggested that the speed and direction of propagation can basically be explained in terms of shelf wave dynamics. In the 1980 event, a dense (colder and more saline) water advanced eastward along the north coast at about 10 km day −1. The lateral scale of the phenomenon was about 30 to 40 km, in agreement with the Rossby internal radius of deformation. The T-S and current data suggest that the 1980 cold event was dominated by phase propagation. In the 1981 event, a light (warmer and less saline) water area advanced eastward at the speed similar to the 1980 cold event, but the T-S and current data suggest that Lagrangian drift of water particles associated with strong eddy motions was not negligible.