A Discussion on the measurement and interpretation of changes of strain in the Earth - Methods for prediction and evaluation of tidal tilt data from borehole and observatory sites near active faults

Tilt has been measured continuously at tidal sensitivity in borehole and observatory sites in the San Francisco Bay region, California. These sites, within a few kilometres of a m ajor fault and the Pacific Ocean, are part of an extensive network measuring strain and microseismicity. Static response of the Presidio site to ocean loading at the M 2 frequency best fits a finite-element model with lower shear modulus in the San Andreas fault zone than in adjacent material at the same depth. The Presidio tilt data exhibit a secular trend less than 3 (rrad/year superim posed on local earthquake and meteorological effects. On two occasions earthquakes (M„ > 4.3) occurred within 55 km of a station and were preceded by anom alous tilt accumulating to 1 ^rad over several days with an accelerated rate of tilt a few hours before the events. The root-mean-square (r.m.s.) difference for two stations of 25 km apart for 700 h before and after one of these events was 5 x 10-8 and 2 x 10 ^ respectively . A similar r.m.s. difference was observed before and after a larger ( M b> 5) but more distant (180 km) earthquake from the same two stations. This latter event did not, however, exhibit the extreme linear slope (10~9 rad/h) of the two earlier earthquakes. Although such anomalies can be correlated with meteorological activity over short periods of time, they do not correlate for periods approaching one month. A transfer function derived during a period w hen there were no local earthquakes can be used for calculating tilt response to surface loading from telemetered meteorological and tilt data. These results can then be input for a prediction beyond the data, and the error in prediction monitored as a final output for instrument performance and potential earthquake hazard.