Estimates of crustal transmission losses using MLM array processing

Abstract

Seismic refraction experiments have been used extensively in the past thirty five years in investigations of the structure of the oceanic crust. The longer range of the refraction or wide angle reflection technique, on the order of tens of kilometers, permits a deeper and wider area of examination, although with less resolution, than the spatially limited seismic reflection experiment. Observations of arrivals from the Mohorovicic discontinuity, at an average depth of seven kilometers below the sea floor, are routinely made. The major focus in interpreting refraction data has been the analysis of travel time/range data and the "inversion" of this data for the purpose of determining a velocity versus depth profile of the crust. The most frequent application of this procedure is the geophysicist's use of velocities for postulating geologic structures and rock types below the sediment (Christensen & Salisbury, 1975). Another area using refraction data, less widely seen, falls into the ocean acoustician's domain. In studying the behaviour of sound in the ocean, the sea floor is often modelled as a boundary with a half space below, and with some form of reflection characteristic and/or loss mechanism. If acoustic energy, upon encountering the bottom, was either reflected or transmitted directly, this would be appropriate, and the determination of reflection and transmission coefficients for the sea-sediment interface would probably be sufficient. However, sound energy does penetrate beneath the sea floor and is both reflected and refracted back to the water. In an active acoustical experiment, especially at longer ranges, a significant amount of the received energy may come from waves that have interacted with the earth's crust and have been reinjected into the water. Since these arrivals can be detected in the ocean, their study is of concern for the acoustician. The role of bottom interaction, especially at low frequencies, is now an area of intense research activity in modelling acoustic propagation. In particular, in the language of the sonar engineer, the TL, or transmission loss, of this energy is of major importance for i) predicting the character of the sound field at a receiver in future experiments, ii) for comparing crustal loss with the better known TL of paths remaining primarily in the water layer, and iii) expanding the role of arrival amplitudes in inversion theory. Just as there may be a number of possib1e paths in the sea between a source and receiver, each with a different loss characteristic, trajectories in the crust are variegated and exhibit different TL behaviors. It is important to be able to differentiate the energy partitioned among the different paths, and to determine which paths are most important. Resolving the locus of a particular acoustic path is intimately tied to the problem of determining the velocity structure of a medium. To the limits of the geometrical optics approximation of acoustic behaviour, sometimes sorely pressed at low frequencies, a completely detailed knowledge of sound speed variations, both laterally and with depth, plus known source characteristics and attenuation losses in the medium, enables one in principle to predict signals observed at a receiver. For an ocean acoustician, the requirement of environmental knowledge of the sound speed profiles, both in water and crust, needed to predict the amplitude and timing of data, is clearly very burdensome. In the past twenty five years, however, models of the oceanic crust have been formulated which are statistically consistent over much of the oceans. These models divide the crust into three or more horizontal layers with certain average thicknesses and velocities (Raitt, 1963). At least within the confines of these models, if a typical transmission loss were known for each of these layers, an acoustician can make predictions of the expected strength and timing of crustal arrivals at other stations. Most of this environmental information has been obtained from refraction and/or wide angle reflection data, usually via travel time analysis. Little has been done in developing models accounting for amplitude dependence.