Impact of acoustic velocity structure to measurement of ocean bottom crustal deformation

We are developing a geodetic method of monitoring crustal deformation under the ocean using kinematic GPS and acoustic ranging. The goal of our research is to achieve sub-centimeter accuracy in measuring oceanic crustal deformation by a very short-time measurement like 10 hours. In this study, we focused on lateral variation of acoustic velocity structure in seawater and introduced an inclined acoustic velocity structure model to improve accuracy of the measurement. We have a few measurement sites along Nankai trough, Japan. In each sites, we deployed a trio of transponders on ocean floor (seafloor benchmark units) within distance comparable with the depth. An ultrasonic signal is generated from a surface vessel drifting over the benchmark unit, which is received and replied by the benchmark unit. In this system, both acoustic velocity structure and the benchmark unit positions were determined simultaneously for the each measurement using a tomographic technique. This tomographic technique was adopted on an assumption that the acoustic velocity structure is horizontally layered and changes only in time, not in space. Ikuta et al., (AGU fall meeting 2009) reported an approach to improve accuracy of benchmark positioning using a new additional assumption. The additional assumption was that the configuration of the transponders trio constituting one benchmark unit does not change. They determined the time evolution of weight center for the fixed transponder triangle between different measurements using all repetitively obtained data sets at once. This is contrasting to the previous method in which each data set for different measurement was solved independently. This assumption worked well in reducing number of unknown parameters. As a result, repeatability of benchmark positioning improved from 5 cm to 3 cm. We conducted numerical experiments synthesizing acoustic travel-time data to evaluate the robustness of this new approach. When acoustic travel-time data is synthesized using horizontally layered velocity structure, the new approach solves the position of the benchmark much more accurately than previous approach does. However, when given velocity structure has lateral variations, the accuracy gets even worse rather than improves comparing to the previous approach. For these solutions, we assumed no lateral velocity variation despite that given velocity structure has lateral variation. This means that the new approach has high solvability of lateral variation. We introduced lateral variation model into the new approach and solved the synthesized data. The accuracy of the positioning improved to be much better than that by old approach. We adopted this new model and the approach to solve the real data sets. The repeatability was worse than that with the old model. There should be some discrepancies between real velocity structure and given model. Acknowledgement: This research was promoted by a project of the Japanese Ministry of Education, Culture, Sports, Science, and Technology.