Data from the Kauai ATOC Acoustic Source

Last updated November 23, 2005.

Data from ATOC are almost but not quite freely available for use. There are a number of peculiar issues at the moment that prevent posting of these data here for ready download. Those interested in using these data should contact Peter Worcester ( or Brian Dushaw ( There should be no problems obtaining the data; at the moment we merely want to keep track of who is using the data, and maintain a dialog with users on how they are using and interpreting the data.

This past year (2005) there have been a number of problems with data collection associated with the greater security restrictions. These restrictions have made remote monitoring of the health of the collection system impossible, which has been the main issue (i.e., when the remote computers crash for whatever reason, we are unaware.)

Comparison of Temperature with the ECCO model and TOPEX/POSEIDON Data

Time series of range- and depth-averaged temperature derived from the acoustic travel times (red, with error bars) compared to TOPEX/POSEIDON (blue) and ECCO model (green; JPL incarnation) data (ECCO model link) averaged along the acoustic paths. The sea-surface height variations are assumed to be caused only by thermal expansion in upper 100 m; the coefficient used to convert between temperature and sea-surface height is strongly temperature dependent. In tropical regions, the ray sampling is subsurface, so the annual cycle of temperature is sometimes not observed. Therefore, a better comparison to the ECCO model is made by calculating the equivalent ray travel times in the ECCO model, shown below.

Comparison with the ECCO Ocean State Estimates

The analysis of path integral data is made much simpler by modern ocean state estimation methods, using travel times as integral constraints on the model variability. If the data estimated by the model do not match the observations, then the ocean model state is adjusted to bring the model into better agreement. As implemented by the ECCO Consortium (Estimating the Circulation and Climate of the Ocean) and others (Stammer and Chassignet 2000, Stammer, et al. 2001), state estimation also serves to best combine disparate data types, which can then be evaluated for their contributions to reducing the uncertainty of the solution. As a first step towards incorporating travel times into the ECCO model cost function, ECCO model output was converted into travel times for several source-receiver pairs. Different ray paths have different sensitivities to the surface and to the deep ocean, and the estimation can exploit this to obtain vertical information from a set of rays.

The ocean state estimate used here is based on an integration of the MIT General Circulation Model in a global configuration that spans 75 S to 75 N, with latitudinal grid-spacing ranging from 1/3 degree at the equator to 1 degree at the poles and longitudinal grid spacing of 1 degree. The model assimilates a variety of satellite and in-situ data and data products, including TOPEX/POSEIDON, WOCE hydrography, XBT sections, etc. A description of this state estimate and the complete fields are available at

Travel time anomalies for the ATOC sources at Pioneer Seamount and Kauai transmitting to receivers k, o, and f (red and blue lines, see the ATOC figure for receiver locations), compared with the travel times calculated from the ECCO state estimates described above (green lines). Each line represents a different ray path, which samples the ocean in a different way. A deeper turning depth means the ray is sensitive to both the near surface and the deep ocean. Deep-turning, surface-reflecting rays are less affected by the annual temperature cycle than rays with shallow turning depths. The time series for sources or receivers near Hawaii have larger variability because of the mesoscale variability near the Hawaiian Ridge.