This report studies the application of some methods for high-resolution bearing estimation on real data from an active sonar array. Conventional beamforming is compared with MUSIC, MVDR and DML on data generated by one moving target and two stationary noise sources. The results of a data reducing beamspace-transform has been estimated indicating almost as good resolution as with non-reduced data. Simulated data has been generated to estimate the resolution-critical separation for two targets. The DML technique is superior in this simulated data application. The high-resolution methods give better result in tracking one single target than conventional beamforming. Range resolution by conventional matched filtering gave a very clear identification of the target as long, as it is not close to either of the noise sources. Analysis for prediction of multipath propagation via surface interaction has been performed showing that due to the shallow depths of the sonar and the target the time differences are too small for separating, the two incoming signals in the time domain. The report is a Master of Science Thesis at the Royal Institute of Technology.