*Takuto Minami1, Mitsuru Utsugi2
(1.Division of Frontier Planetology, Department of Planetology, Graduate School of Science, Kobe University, 2.Aso Volcanological Laboratory, Graduate School of Science, Kyoto University)
Keywords:inversion, MT, volcano, Aso, ACTIVE, controlled source
The resistivity structure beneath a volcanic edifice is an important proxy for understanding the hydrothermal and magmatic systems. The Audio-frequency Magnetotelluric (AMT) method is a useful tool to investigate shallow resistivity structures up to depths of several kilometers. Kanda et al. (2019) inferred three-dimensional (3-D) resistivity structure by dense AMT survey data, which implies that the resistivity of Aso volcano is accounted for mainly by the background hydrothermal system balancing the effective rainfall and the heat input at depth (Minami et al. 2021, SGEPSS). On the other hand, controlled-source electromagnetic (CSEM) methods are also important for the monitoring purpose, because of their stable signal-to-noise ratio. A CSEM volcano monitoring method called ACTIVE (Utada et al. 2007) is sensitive to temporal changes in resistivity structures. Minami et al. (2018) applied a 3-D finite element inversion to the ACTIVE data set obtained in Aso volcano and inferred the temporal variation in the 3-D resistivity structure associated with the magmatic eruption of Aso volcano starting in November 2014. The resistivity structures from AMT and ACTIVE data, however, have a significant discrepancy in the resistivity values. The resistivity structure inferred by AMT data is relatively conductive compared to those inferred by ACTIVE data sets in Aso volcano.
To obtain consistent view of resistivity structure of Aso volcano, we are currently developing a joint inversion code that can deal with AMT and ACTIVE data sets simultaneously, based on the finite element inversion code for ACTIVE developed by Minami et al. (2018). For effective inversions, we prepare a common coefficient matrix at each frequency for both AMT and ACTIVE forward linear equations. This was achieved by forcing the Dirichlet-type boundary at all the boundaries of the calculation domain in both AMT and ACTIVE forward problems. We have already confirmed that both AMT and ACTIVE forward solutions obtained from the common tetrahedral mesh are accurate enough with respect to the analytical solutions. In the preliminary inversion of our developed code, we found that the derivatives of boundary conditions in MT problems with respect to the model parameters is necessary to include in the inversions for long periods around ~1 second. In the presentation, we report the details of our developed joint inversion code and plan to show application of our code to the AMT and ACTIVE data sets obtained in Aso volcano.