SOME MAPS
Figures 5 and 6 show two sample objective maps of the North Pacific Ocean using the ATOC data. These are maps of 0-1000-m depth-averaged sound speed, with values ranging from -0.5 m/s (dark blue) to +0.5 m/s (dark red). The uncertainty map is superimposed using the white contour lines. The yearday associated with each map is indicated by the bottom panel. The travel time residuals associated these objective maps are presently too large by far, but this is the first attempt at constructing these maps. Further, because most of the receiver positions are classified, fictive positions are used for the inversion, and so the fit cannot be expected to be as good as when the true positions are used. [Objective maps with the true positions will be completed soon.] The horizontal correlation length assumed for the inversion is shown by the width of the uncertainty contour lines across an acoustic path. Since the ray paths do not sample to the ocean surface in the region near Hawaii like they do in the northern or eastern pacific, the maps shown here must be interpreted carefully. To best estimate the ocean state, the acoustic data might be better employed by combining them with other data using a dynamical ocean model and data assimilation, but it is still important to make the zero-order ocean estimate for a variety of obvious reasons, e.g. a sanity check.
CONCLUSIONS
The work described here is at the level of a mapping excercise at the moment. This work will continue with the goals of making the most statistically consistent maps possible, and then comparing and combining these maps with other data such as the Hawaiian Ocean Timeseries (HOT) or TOPEX/POSEIDON altimetric data. Even with the brief timeseries that are available, one catches a glimpse of the impressive possiblities of the acoustic data. One can imagine what a complete 3-year timeseries of maps of the North Pacific would look like, and what an interesting oceanographic observation this would be. We hope that the results described here are just the beginning of a long and continuous timeseries.
