Significance test for seismicity rate changes before the 1987 Chiba-toho-oki earthquake (M 6.7) Japan

A precursory seismic quiescence lasting 1.5 ± 0.5 years was observed prior to the 1987 M 6.7 Chiba-toho-oki earthquake, Central Japan. This event was the largest mainshock to occur in the region in 60 years. A quantitative analysis of the seismicity rates, using two independent catalogs provided by the NIED and JMA networks, shows that the precursory seismic quiescence is centered in the shallower part of the rupture zone of the subsequent mainshock, at a depth of 20-40 km. At the hypocenter of the 1987 Chiba-toho-oki mainshock, a 50% increase in the seismicity rate was detected in the NIED data, coinciding in time with the onset of quiescence (1986.4 ± 0.5). The simultaneous appearance of both quiescence in the shallow part of the rupture zone observed in two catalogs, and a rate increase in the immediate hypocenter region, suggest that these phenomena are causally linked to the subsequent mainshock. However, a quantitative analysis of both catalogs reveals that the precursory quiescence and rate increase are not unique, since rate changes of this duration and significance often occur in the data. A rate change of this significance rating could probably not be detected as a precursor in a real time approach. For the aid of real time monitoring of seismicity rate changes, we introduce the method to calculate the 95-percentile of confidence level for the significant rate changes.

Quantitative mapping of precursory seismic quiescence before the 1989, M 7.1off-Sanriku earthquake, Japan

The first main shock of the off-Sanriku earthquake sequence (02/11/89, M 7.1; 18/07/92, M 6.9; 28/12/94, M 7.5) was preceded by a precursory seismic quiescence lasting 2.5 ± 1 year and up to this main shock. The detailed properties of this quiescence were mapped as a function of time and space by a gridding technique using the ZMAP computer code, and the statistical significance was estimated by generating a synthetic catalog based on the microearthquake catalog of Tohoku University, which was the data set used. The statistically most significant expression of this precursory quiescence has a probability of 0.1% to have occurred at random and was located in the eastern part of the 1989 aftershock area, at a point to which the 1994 aftershocks extended also. If we define the dimensions of the quiescence anomaly by a vertical cylinder with the depth of the entire seismogenic layer, centered at the point of most significant quiescence and showing a rate decrease of 75%, then we find its radius is 25 ± 9 km. If we allow other shapes, such as the simplified aftershock volume of 1989, or other simple geometric figures, to define the rate decrease we find dimensions of 80 by 80 km. The characteristics of the quiescence anomaly do not depend strongly on the choice of free parameters within the following ranges: 100 ? number of events ? 400, 2.0 ? Mmin £ 3.0, 1 ? time window ? 3 years. With our method, a thorough analysis of the period before the 1994 main shock is not possible because of the interference of the extended aftershock sequence of 1989. Nevertheless, we identified a quiescence of nearly zero earthquakes located near the center of the 1994 aftershock area that lasted for one year up to that main shock. However, this quiescence period ranked only 46th in significance, behind other quiescences of equal duration and similar dimensions distributed in time and space through the data set. Because of the ubiquitous existence of periods of near zero activity during short periods like one year, we find that quiescences shorter than about 1.5 years cannot be defined with high statistical significance in most earthquake catalogs. In the last two years of the data (1995.3-1997.3) we see no extensive quiescence of high significance off the east coast of Honshu between 36.5° and 42°N in the currently available data.