Temporal variations of the coda decay rate on Kamchatka: Are they real and precursory?

Abstract

Temporal variations of the decay rate of the coda of local earthquakes are studied for four Kamchatka stations. To suppress any bias caused by the coda Q versus lapse time dependence, I use the deviation of an individual coda decay function from the empirical reference decay function, and the analysis is done over a fixed lapse time window. For each individual coda record, a decay rate parameter “α” is determined by the following procedure. First, the reference log coda decay function is subtracted from an individual one giving a residual function; second, the slope of this residual versus lapse time is determined over the fixed lapse time window giving an α estimate. The data used are records of 150–250 Kamchatka earthquakes of 1967–1990 (ML=3.5–5) obtained by four stations employing three‐component 1‐s seismographs. The processed data show apparent temporal variation of α with high statistical significance. Several possibilities are then investigated of mimicking the genuine temporal variation of α by systematic variation of other parameters, first of all epicenter wandering and nodal plane rotation. This analysis does not reveal any strong bias and thus suggests that the observed variation is genuine. Then I show the significance of five apparently precursory anomalies identified retroactively. Also, I describe briefly the real‐time prediction experiment conducted on Kamchatka for 1982–1990 using coda decay, aimed at the intermediate‐term forecasting of large Benioff zone earthquakes. This experiment resulted in the successful forecast of the August 17, 1983, Mw=7.0 event with an accuracy of 2 months in time and 100–200 km in location; magnitude was overestimated by 0.5. Later, a false alarm was also issued. Thus the experiment confirms the reality of the coda decay rate precursor but also shows that it needs further improvement to become reliable. Physically, the variations of the coda decay rate are associated with time‐dependent local variations of scatterer density in the lithosphere.