Offshore jacket structures are regarded as a suitable type of support structure for offshore wind turbines in immediate water depths. Because of the welded tubular members used and environmental conditions, offshore jackets are often subjected to fatigue damages during their service life. Underwater sensors can provide measurements of the structural vibration signals and provide an efficient way to detect damages at early stages. In this work, simplified forms of the damages are assumed, random damages are imposed on the jacket structure, and damaged indicators are established from combination of modal shapes. Then, a response surface is constructed mapping the damage indicators and damages. Given that the efficiency of the damage identification depends on the locations of the damages and the location and number of sensor locations, a sensitivity study is performed to vary sensor location, sensor quantity, and damage severity. It is found that the effect of damage identification is better when sensor locations are closer to damage locations, and this effect is more sensitive to sensor placements when damages occur in the upper structures. Additionally, the identification effect is more sensitive to damage severity than to occurrence of multiple damages.