Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4715379 | Journal of Volcanology and Geothermal Research | 2006 | 19 Pages |
Weak long-lived ground deformation associated with the volcanic unrest of Mt. Iwate, northeastern Japan in 1998–2002 is detected from borehole strainmeter, tiltmeter and global positioning system (GPS) data. This deformation is relatively weak (0.01 μ/day strain, 0.01 μrad/day tilt, and 0.01 cm/day displacement), but persists for a period of 5 years. To detect such weak long-term deformation, a new decomposition method based on Bayesian theory is developed and applied to the recorded data. This method is effective for extracting secular drift from the strain and tilt data, and seasonal changes from the GPS data. From a selection of possible models, a composite source model consisting of a dike series and a Mogi source is identified as the best fit to the observed data. The deformation is divided into 8 phases based on the characteristic variations in tilt and strain. Based on this model, the deformation sequence is considered to have initiated with the intrusion of a dike at intermediated depth below the summit region between February and April 1998, with short periods of shallow dikes intrusion toward the west in March and April. Shallow dike intrusion resumed in September, reaching a final volume of 5.9 × 106 m3. The Mogi source is inferred to have been located 3 km beneath the western flank of the volcano, slowly inflating to a maximum volume 1.2 × 107 m3 by the end of 2000, then deflating by 49% to 2002. This ground deformation was accompanied by seismic migration, from the summit region in the vicinity of the intruding dike to the west following the dike extension. This dike-correlating seismic activity ceased simultaneously with deflation of the Mogi source, suggesting that the Mogi source was a hydrothermal reservoir that was breached by the intruding dike. The Bayesian decomposition method developed in this study makes it possible to clarify the detailed characteristics of dike intrusions and the long-lived activity of the Mogi source utilizing the high sensitivity of strain and tiltmeters coupled with the stability of GPS.