The Fengshan fault, originally identified by the geomorphic analysis, penetrates from the southern area of Tainan city in the north to the downtown of Kaohsiung city in the south in SW Taiwan. Based on previous GPS analysis, this fault is a major active creeping fault with approximately 15 mm/yr left-lateral strike slip rate.  However no historical disastrous earthquakes occurred on this fault and no geological evidence to prove the existence of the Fengshan fault. In order to realize the location and kinematics of the Fengshan fault, GPS observations during 2007-2015 from 44 continuous and campaign-mode stations and another 68 GPS satellite control point data between 1997 and 2010 are used to estimate the surface horizontal velocity field in this study relative to the station KMNM in Chinese continental margin. In addition, 17 Synthetic Aperture Radar (SAR) images from ALOS PALSAR from 2007-2011 are also adopted in this study to evaluate the Line-of Sight (LOS) velocity field in SW Taiwan for increasing the spatial resolution of surface deformation. Because the southern tip of Fengshan fault is passed through the subsidence region of the Pingtung Plain and this fault is a strike-slip fault, leveling vertical velocities are used to remove the contribution of land subsidence from the LOS velocities. According to the analysis of above all data, the northern segment of the Fengshan fault is creeping with the creeping rate of approximately 13 mm/yr and the southern segment of the fault is probably locked with the rate of approximately 17 mm/yr. In addition, two mud volcanos are discovered at two end points of the northern creeping segment of the Fengshan fault, which may imply the reason of surface creeping on the Fengshan fault. The thick mud may decrease the friction on the fault plane and promote the generation of the creeping fault in S Taiwan. If any constructions are passed through the creeping fault, a continuous damage will be occurred due to the movement of the fault. Therefore it is important to build up a real-time monitoring network to identify the kinematics of the creeping fault.