Mapping climatic temperature changes in the ocean with acoustic tomography: navigational requirements

In eddy-resolving hydrodynamic models, first-mode baroclinic Rossby waves linked to El Nino/Southern Oscillation are the dominant features which change basin-wide temperatures below the seasonal thermocline in the northeast Pacific at periods less than a decade. Simulations are carried out in which Rossby waves are mapped using acoustic tomography. Based on the model which propagated these waves, a Kalman filter is used to map temperature signals for a year. The modeled data are taken from a dense network of acoustic tomography sections. At 300-m depth, where the temperature perturbations associated with Rossby waves are about /spl plusmn/1/spl deg/C, 80% to 90% of the model variance is accounted for with tomographic estimates. The corresponding standard deviations of the estimates are less than 0.1/spl deg/C at 400-km resolution. About 80% of the model variance is accounted for with tomography when the navigational errors of the sources and receivers are as poor as one kilometer. Consequently, it may be unnecessary to accurately navigate actual tomographic instruments to map climate change. Modeling results are insensitive to: 1) a reduction in data due to a significant number of instruments which fail; 2) whether the instruments are mobile or fixed; 3) the detailed trajectories of mobile receivers; 4) the shape of the a priori spectrum of ocean fluctuations; 5) the corrections to the acoustic travel-time biases; and 6) the errors in the sound-speed algorithm. In basin-scale arrays, the modeled variance of acoustic travel time depends on the horizontal wavenumber of temperature as k/sup -5.5/. Because sound has little sensitivity to small wavelengths, modeled Rossby waves can be mapped in a day from a few sources and of order ten receivers. The results only depend on the model having large scales in space and time.