摘要
When using long range sound travel-time measurements to monitor the ocean temperature changes, the tidal effects must be corrected from the data. On the basis of a linear model of the tidal signal, and using measurements taken within a specified period and at 4 hour intervals, a pseudorinverse method is used to predict the traveltime change due to the aggregate effect of barotropic tides along the sound path. The sampling period should be sufficiently long to give an acceptable prediction accuracy. In order to estimate all the major tidal constituents and separate closely spaced frequency components, a sampling period of 18 months is recommended. The linear model should include as many constituents as possible to minimize the predictioll error. This is feasible because in modeling the tide, the only parameter needed for each constituent is the frequency; and the freqencies of the astronomical components are known to a high precision, while the nonastronoIIilcal components are trivial in this application. Quantisation errors are reduced by means of multipath averaging.
When using long range sound travel-time measurements to monitor the ocean temperature changes, the tidal effects must be corrected from the data. On the basis of a linear model of the tidal signal, and using measurements taken within a specified period and at 4 hour intervals, a pseudorinverse method is used to predict the traveltime change due to the aggregate effect of barotropic tides along the sound path. The sampling period should be sufficiently long to give an acceptable prediction accuracy. In order to estimate all the major tidal constituents and separate closely spaced frequency components, a sampling period of 18 months is recommended. The linear model should include as many constituents as possible to minimize the predictioll error. This is feasible because in modeling the tide, the only parameter needed for each constituent is the frequency; and the freqencies of the astronomical components are known to a high precision, while the nonastronoIIilcal components are trivial in this application. Quantisation errors are reduced by means of multipath averaging.
