Abstract. As part of the Office of Naval Research Sea Ice Mechanics Initiative, a real-time monitoring and processing program for acoustic emission from ice fracture and ridgebuilding events was established. A wide-aperture, horizontal hydrophone array was used in combination with a vertical line array to record the acoustic signals, which were then passed through a focused beam former for real-time generation of ice seismicity maps. A number of rapidly deployable geophone arrays were used in active zones to measure the acoustic emissions in the near field for detailed seismic event analysis. During one such deployment, a highly regular transient arrival structure was recorded on all sensors located near a major lead, with a transient appearing every 5 s. These data have been processed using frequency-wavenumber analysis to show that the transients correspond to "edge waves" propagating forth and back along the edges of the lead, with the probable source being a "stick-slip" mechanical phenomenon toward the ends of the lead.

In this thesis the analysis of natural ice events is carried out based on direct measure ments of ice-borne seismo-acoustic waves generated by ice fracturing processes. A major reason for studying this phenomenon is that this acoustic emission is a signifi cant contributor to Arctic ocean ambient noise. Also the Arctic contains rich mineral and oil resources and in order to design mining facilities able to withstand the harsh environmental conditions, we need to have a better understanding of the processes of sea ice mechanics. The data analyzed in this thesis were collected during the Sea Ice Mechanics Initiative SIMI'94 experiment which was carried out in the spring of 1994 in the Central Arctic.One of the contributions of this thesis was the determination of the polarization characteristics of elastic waves using multicomponent geophone data. Polarization methods are well known in seismology, but they have never been used for ice event data processing. In this work one of the polarization methods so called Motion Prod uct Detector method has been successfully applied for localization of ice events and determination of polarization characteristics of elastic waves generated by fractur ing events. This application demonstrates the feasibility of the polarization method for ice event data processing because it allows one to identify areas of high stress concentration and "hot spots" in ridge building process.The identification of source mechanisms is based on the radiation patterns of the events. This identification was carried out through the analysis of the seismo-acoustic emission of natural ice events in the ice sheet. Previous work on natural ice event identification was done indirectly by analyzing the acoustic energy radiated into the water through coupling from elastic energy in the ice sheet.After identification of the events, the estimation of the parameters of fault pro cesses in Arctic ice is carried out. Stress drop, seismic moment and the type of ice fracture are determined using direct near-field measurements of seismo-acoustic sig nals generated by ice events. Estimated values of fracture parameters were in good agreement with previous work for marginal ice zone.During data processing the new phenomenon was discovered: "edge waves", which are waves propagating back and forth along a newly opened ice lead. These waves exhibit a quasi-periodic behavior suggesting some kind of stick-slip generation mecha nism somewhere along the length of the lead. The propagation characteristics of these waves were determined using seismic wavenumber estimation techniques. In the low frequency limit the dispersion can be modeled approximately by an interaction at the lead edges of the lowest order, antisymmetric modes of the infinite plate. Without them the data set around which this thesis revolves simply would not exist.

As part of the ONR Sea-Ice Mechanics Initiative (SIMI) experiments in the spring of 1994, arrays of high-resolution seismic sensors were deployed in areas of high seismicity identified by real-time processing of acoustic emission events recorded by a wide, horizontal aperture hydrophone array. One of the areas most rich in ice events was located on a small ice floe (∼100×100 m) 4 km Northeast of the main camp, where the seismic array consisting of five 3-component geophones in a 70-m aperture pentagon was deployed. One of the methods employed for analysis of the near-field data from this deployment was motion-product seismograms introduced by J. E. White [Geophysics 24, 288–298 (1964)]. In this method each of the horizontal components of ice motion was multiplied by the vertical component, with or without phase shift, and after integration, the two resulting products identified a vector pointing to the source of the seismic waves. The ability of this method to separate different polarizations of the seismic waves and to determine the direction to the source was especially useful for geophone array data, because other analysis methods for such data occasionally failed due to the overlapping of waves generated by different ice events. Using the polarization processing method the development of ice fractures in the array near field was successfully tracked in the time and spatial domains. One result of the polarization analysis was that these fractures seemed to mostly generate vertically polarized shear (SV) waves. [Research supported by ONR.]

A new, layered seismo-acoustic remote sensing concept was developed and applied during the SIMI field experiments in the Beaufort Sea in the Fall of 1993 and the Spring of 1994. A large-aperture, 32-element, horizontal hydrophone array was used to record the acoustic emission from ice events. Using real-time array processing, maps of the seismic activity of the ice cover out to a range of 2 to 3 km were continuously generated. Once an active zone was detected, clusters of five 3-axis geophones and a single hydrophone were deployed in the active zone for near-field recording of the seismo-acoustic emission. The data were transmitted back to the main camp via a wireless local area network, and recorded on tape. During the Spring experiment data were recorded continuously on the hydrophone array for 4 weeks, and several deployments of the geophone clusters were performed in the vicinity of active ice mechanical processes such as ridge building, finger rafting, and floe fracturing. In addition, the clusters were deployed on specimens used for artificial fracturing experiments. The layered remote sensing concept is described, and examples are given of the seismo-acoustic emission produced by the different types of ice events. Finally, the matched-field fracture plane analysis of the seismic data is described.

As part of the ONR sea-ice mechanics initiative (SIMI) experiments in the spring of 1994, arrays of high-resolution seismic sensors were deployed in areas of high seismicity identified by real-time processing of acoustic emission events recorded by a wide, horizontal aperture hydrophone array. A major episode detected by the surveillance array was a new major lead opening up 2 km north of the ice camp. Two seismic arrays, each consisting of five three-component geophones in an 80-m aperture pentagon, were deployed 400 m apart on the edge of the lead. One of the methods employed for analysis of these data was the 2-D horizontal wave-number estimation of the recorded time series using plane wave beamforming. This procedure determined the propagation directions of arrivals of different wave types along with their dispersion characteristics. This analysis revealed a large number of events, corresponding to ‘‘edge waves’’ propagating in opposite directions along the lead. The analysis revealed that these waves behaved differently from the edge waves of bending-gravitational nature, predicted theoretically in the literature. Also, the broadband nature of these waves is inconsistent with floe resonances. To explain the nature of these waves, a simplified elastic model has been developed for the lowest-order modes of a floating, semi-infinite plate. The characteristics of the edge waves and the comparison with the simplified model are discussed. [Research supported by ONR.]