Three-state Network Research Articles

The height dependence of TID and gravity wave parameters

The height dependence of amplitude and both horizontal and vertical phase trace speeds, for individual spectral components of F region TID's, are obtained for the first time by means of a three‐station network of rapid‐run ionosondes. The corresponding causative gravity wave parameters are derived using a previously developed inversion technique, which is extended by incorporating the effects of dissipation and ion diffusion. Reasonable agreement is achieved between the height variation of the real part of the vertical component of the gravity wave phase propagation vector, kzr, obtained from the inversion process, and solutions of a dissipative dispersion relation for kzr at different heights. Height dependent values of the imaginary part of the vertical component of the gravity wave phase propagation vector, kzi, which are required for the inversion, are computed from the dissipative dispersion relation. Values of kzi obtained from analytic formulae which assume an isothermal atmosphere and small dissipation yield erroneous results for the waves considered, suggesting that care should be exercised when interpreting previous theoretical results that have employed these assumptions.

The optimal estimation of earthquake parameters

The precision with which various earthquake parameters can be determined using an existing or a proposed network of seismographic stations has recently been of considerable interest. Analytical solutions for the optimal resolution are presented for the epicenter, hypocenter, focal mechanism, seismic moment, and local magnitude estimation problems. The algebraic method is to decompose orthogonally the equations of condition for each problem using the singular-value theorem. The orthonormal equations and the data covariance then yield the uncertainties in the estimations of the unknown earthquake parameters. Possible biases due to significant linear correlations among the observed data are included in the analysis. As an example, the ability of data from the Berkeley three-station network of broadband seismographs to determine the above earthquake parameters is discussed. The parameters of the M L 5.9 Coyote Lake earthquake of August 6, 1979 are determined in detail. In order to use all of the resolution that a network has for determining earthquake parameters, it is necessary to calibrate the network for regional earthquakes which have well-determined parameters. If the network is not calibrated, the use of relatively simple velocity models will lead to significant systematic errors in the estimation of some of the parameters.

Variations in the relationship of seismic velocities before two earthquakes in Imbria (Italy)

The variation of thevP/vS relationship between the velocities of longitudinal and transverse seismic waves in an area of Central Italy has been examined between January 1979 and March 1980. SP seimograms, recorded by a three-station network, have been used duringM≤4.2 earthquakes. Such events took place along an active zone which included the epicentre areas of the two strong shallow earthquakes which occurred in Umbria on 19-9-1979 (M=5.4) and on 28-2-1980 (M=4.8), respectively. This study has allowed us to mark clear anomalies before such events took place. It has been possible also to mark certain differences in the anomalous variation of thevP/vS relationship for different values of the focal depth with respect to the events which have been studied.

An evaluation of a signal-summing technique for improving the signal-to-noise ratios for seismic events

Formulas are derived for evaluating as a function of frequency the improvement in the root mean square signal-to-noise ratio for seismic events obtainable by the technique of summing the trace amplitudes recorded at a number of closely spaced seismometer stations. A modification of this technique whereby the signals are corrected for phase shift before summing is also studied. The theory is applied to noise data from a three-station network for the ranges of periods 6.7 to 20 sec and 0.25 to 1 sec. The signals in the short-period range are taken to be P and S body waves. Rayleigh waves are used in the long-period range. Improvement comparable to and occasionally in excess of that expected for uncorrelated noise is indicated for the modified technique in the short-period range, whereas little improvement is obtained in the long-period range for the given station spacing and noise conditions.