Some problems on detection of earthquake precursors by means of continuous monitoring of crustal strains and tilts

Previous work carried out in Japan to search for earthquake precursors by means of continuous monitoring of crustal strains and tilts was reexamined. We find no evidence of reliable precursors observed simultaneously with more than two instruments or at more than two observation sites. There is a fair chance to detect earthquake precursors if strain meters and tiltmeters are distributed at many closely spaced sites in and around the source region, but observations with strain meters and tiltmeters have not been executed at distances less than twice the source dimension. We thus propose the following relation as a criterion of earthquake prediction at the present stage: M = 1.96 log r + 3.84, where M is the magnitude of an earthquake and r (kilometers) corresponds to 2 times the radius of the source dimension. Precursory strain and tilt changes may be detected within the distance of r, as revealed by data of geodetic surveys. Then we examined whether this criterion can be revised by employing reliable data obtained from laser strain meters installed in a deep tunnel at the Amagase station, Kyoto, Japan. During the period of 32 months from January 1, 1987, to August 31, 1989, there occurred three moderate earthquakes exceeding magnitude 4 within 30 km of the strain meter, but we could not detect anomalous strain changes associated with earthquakes. Strain data obtained from laser strain meters at Amagase do not correlate with surface meteorological changes, such as atmospheric and temperature changes and rainfalls, but are affected by the changes in the groundwater table around the observation tunnel. A search of tidal strain data from the laser strain meters at Amagase reveals no evidence of anomalous changes in tidal signals during the 32 months of observation. The limit of detection for changes in the tidal strain amplitudes are 5% for the M2 constituent during the 32-month averaging period. Judging from these facts, it would be difficult to detect earthquake precursors by employing strain meters and tiltmeters at distances of more than twice the earthquake source dimension. In searching for precursory changes, one must take into account the effect of groundwater flow, even when strain meters and tiltmeters are installed in a deep tunnel to avoid the surface meteorological effects.