• Deception Island;
  • long-period events;
  • seismic arrays;
  • volcano seismology.

The zero-lag cross-correlation technique, used for array analysis in the hypothesis of plane waves, has been modified to allow the wave front to be circular. Synthetic tests have been performed to check the capability of the method, which returns the input test data when the source–array distances are not greater than two or three times the array aperture. For this distance range the method furnishes an estimate of the apparent velocity and the epicentral coordinates of the source. For more distant sources the method becomes equivalent to that based on the planar-wave approximation, which gives an estimate of the backazimuth to the source and the apparent velocity. The method has been applied to seismic data recorded at the active volcano located at Deception Island, Antarctica. 35 volcanic long-period events occurring in a small swarm were selected. Results show that the epicentres are close to the array (between 0.4 and 2 km) and aligned in a SW direction, in agreement with one of the main directions of the fracture system of Deception volcano.