With an average water depth of 10-15 m and hard sea-bottom, strong surface waves create severe problems for seismic body wave imaging in the Arabian Gulf. Common techniques for surface wave analysis are frequency, f-k or velocity filtering techniques. However, spatial aliasing of surface waves due to under-sampling presents unique challenges to these methods. In this paper we explore the difference in time-frequency variations of different seismic events to analyze and attenuate Scholte waves and other P-related surface waves. In this study, we utilize the S transform instead of Fourier transform to build a t-f-k filter which provides a convenient way to analyze surface waves on separate f-k panels at different times. The velocity of Scholte waves ranges from 450 m/s to about 2100 m/s depending upon the spatial variation of seabed shear wave velocity. The P-related surface waves usually have a velocity ranging from 1300 m/s to 3500 m/s, which are more severely aliased than other waves in the seismic record. Analysis indicates that the proposed S-t-f-k transform is effective to attenuate these surface waves in this area and improve reflection signal quality, which may be applicable in other similar ultra shallow water environment.