Underwater natural gas pipelines (UNGPs) used to typically transport natural gas, which is a very important energy resource, can be punctured due to various reasons such as corrosion and shipping/fishing activities. Therefore, the detection and positioning of leakages has an important role in ensuring energy reliability and preventing environmental pollution. In this study, kth nearest neighbor (KNN) algorithm based fingerprint method is proposed to locate leakages in the UNGPs. The success of this method is analyzed using a realistic time-varying multipath underwater acoustic channel and the parameters of a real life UNGP. Numerical results show that for this channel, leakages can be positioned with a margin of low errors from kilometers away depending on the ambient noise, number of receivers, and sea wave height.