I have few near-field ground motion records having a range of 100-150cm/sec PGVs. can we find out which ground motion was recorded forward and backward? Also, does ground motions recorded backward have high velocities of the above range?
The answer that Shahid has given you is correct. It is used in seismology a lot to determine movement direction with respect to your recording station. It is explained in detail by Cronin (2008) in his paper: "Cronin, V.S., Millard, M.A., Seidman, L.E. and Bayliss, B.G. 2008. The seismo- Linearment Analysis Method (SLAM). A reconnaissance tool to help find seismogenic faults: Environmental and Engineering Geoscince, v. 14, no. 3, 199-219 p.
In essence what you are looking for is depicted in the attached figure. You can also look at the frequency content. The forward content may be a little bit higher than the backward content.
The detection of forward and backward directivity on accelerograms, when you have near fault data from many stations, is possible by comparing waveforms from different stations. Stations recording forward directiviity generally show higher amplitudes and shorter duration (single pulse with highest amplitude in the frequency band from 0.5 and 2 Hz). Backward directivity rupture lengthens the signal and reduces waveform amplitudes. Important details on how to rigorously separate forward directivity can be found in this document http://www.seismo.ethz.ch/research/groups/comp/Publications/Mena2010_GEORISK