Evaluation of detection level of crustal deformation observation in the time domain through power spectrum analysis

We evaluated detection levels in the time domain through the power spectra from time-series data in order to investigate the detection abilities of the volumetric strainmeter, the multi-component strainmeter and the GPS in the Tokai and Kanto regions for crustal deformation over a wide range of time constants. The power spectra of the strainmeter data was approximately inversely proportional to the square of the frequency f in all of the frequency domains, whereas the power spectra of the GPS data had white noise in a cycle shorter than approximately 100 days and approached f^-2 in the longer cycle. These results agree with those reported in previous studies. Detection levels evaluated from the power spectra were consistent with previously estimated noise levels at certain time intervals. The evaluated detection levels of the strainmeters were widely distributed at every observation point and appeared to be dependent on both the installed depth and the media around the strainmeter. The detection level decreased as the time interval T became longer. The detection level of the strainmeter was proportional to T^-0.5 in the strain rate, and the detection level of the GPS was proportional to T^-1. The linear strain, areal strain and displacement caused by the long-term SSE and the short-term SSE in the Tokai region were compared with the evaluated detection levels to examine the possibility of SSE detection. These were found to be appropriate indices of detection level, because the evaluated detection levels adequately represented the actual detection of the deformations. Furthermore, the comparison suggested that the GPS may not observe both types of SSE and that the strainmeter has potential for observing long-term SSE.