Observation Methods and data
Five 200-Hz transceiver moorings, each equipped with a sound source and a 5-element vertical hydrophone array [Nakamura et al., 1995] were deployed at 1100-m depth over a 1000-km-scale domain centered at 33°N, 149°E; the five-mooring array produced 10 horizontal connecting paths. Each mooring is equipped with a surface buoy for Global Positioning System (GPS) timekeeping and real time data telemetry, so that the acoustic travel time and inverse analyses can be done in parallel with sound transmission and reception. All of the transceivers (T1-T5) were placed TOPEX/POSEIDON (T/P) satellite altimeter ground tracks. Paths 12-T5 and T3-T4 were put along the descending T/P ground tracks p238 and pl36 (Figure. 1), respectively, to simplify the comparison of the acoustic tomographic data and the T/P altimeter data. Sound was transmitted every 3 hours in August and every 6 hours in July and September; transmissions were sequential, taking 2 hours to complete. After beamforming at the vertical arrival angles of-10°,0°, and +10°, the received acoustic data were correlated with a replica of the transmitted pseudo-random M-sequence to detect travel times for the largest 150 peaks. Travel times for the peak data were recorded on hard disk and sent to a land-station in real-time from the surface buoys using the INMARSAT-C satellite communications system. The mooring motion was tracked using local long baseline acoustic navigation nets. Transceiver position uncertainties were estimated to be approximately 2 meters, corresponding to a few milliseconds uncertainty of travel time.
The measured arrival patterns with +10｡?rrival angle from T5 to T2 (1000.4 km range) contain approximately ten early, discrete arrivals followed by an extended cluster of arrivals (Figure 2); these early ray arrivals are very stable and persistent and can be followed from day to day. Travel times for all the arrivals had prominent changes: minimum on July 26 (marked A), maximum on August 10 (marked B) and minimum on August 31 (marked C), implying a sequence of water warming and cooling process. Arrival patterns predicted from the historical NODC (National Oceanographic Data Center) hydrographic data [NODC, 1991] are in excellent agreement with the measured arrival patterns (Figure 2). The earliest 5 arrivals correspond to surface reflected rays with lower turning points deeper than 4300 m. The clustered arrival corresponds to rays that propagate near the sound channel axis and have the upper turning points deeper than 500 m and lower turning points shallower than 1800 m.
XBT, CTD, XCTD, and shipboard acoustic Doppler current profiler (ADCP) measurements were obtained on several acoustic transmission sections during the deployment and recovery cruises. These profile data are used to compare with the acoustic tomography results.
