Newly developed 3D velocity structure model of the Osaka sedimentary basin

Three dimensional subsurface structure model of the Osaka sedimentary basin is revised with additional survey data conducted under Comprehensive Research on the Uemachi Fault Zone (FY2010-2012) by MEXT. We improved the three-dimensional basin velocity model by adding new observations and applying newly developed methodology to describe the three-dimensional model under the Comprehensive Research Project on the Uemachi Fault Zone by MEXT. 3D velocity structure models have been developed for the Osaka sedimentary basin from earlier time than in other areas thanks to relatively dense data of underground structure surveys. Former 3D models are classified into two types. One, we call them J-type here, includes Kagawa et al.(1993), Miyakoshi et al.(1997), Miyakoshi et al.(1999), Kagawa et al.(2002), Iwata et al.(2008) and Iwaki and Iwata (2011). Another one, H-type, includes AIST model (Horikawa et al., 2003) and Osaka Prefecture model (Os-aka Prefecture, 2004). These two types adopt quite different description of their 3D structure. J-type models divide the sediments into three layers with constant Vp, Vs and densities and adopt spline-function to model the layer boundaries, which make it easy to derive medium properties at arbitrary point. H-type models are given in fixed 3D grids to express complex heterogeneity and steep material-boundaries like overhang faults. Medium properties are given by empirical formulas depending on the depth and the depositional age which were constructed based on geophysical prospection data. In this study, we aimed to model the layers and medium property structure as faithful as possible to survey data (like H-type models) and to describe the layer boundaries by interpolation functions so that we can get the model in arbitrary mesh (like J-type models). To realize this, we construct our 3D velocity structure model with the following way. In order to get information to improve the velocity structure, we have conducted the microtremor array observation for obtaining phase velocities at 6 sites in southern part of the Osaka basin, single-station microtremor observation for obtaining H/V spectra at 100 strong motion stations, continuous microtremor observation at 15 stations and seismic interferometry, and the reflection survey along 2 lines. We also collected strong motion records from seismic intensity observation network by Osaka prefectural government and other strong motion networks (CEORKA, K-NET, KiK-net, etc.), and used them to estimate PS-P travel time by the receiver function analysis and to compare with the synthetic waveforms of moderate size events. We found that the velocity structure model needed improvement especially in the southeast of the basin, southern part of the Osaka Bay area, and northern edge of the basin. We made necessary modification to empirical formula for medium properties and depths of layer boundaries. The purpose of this study is to develop an easy way to estimate deep ground structure. We combine gravity survey and long-term microtremor survey. Seismic interferometry is applied to analyze long-period microtremor, where the influence of deep soil structure appears. Our method is applied to the case in Hsinchu, Taiwan and structure model is modified. Natural frequency is one of the important characteristics which are determined with physical properties and geometry in subsurface structure including mountains. In particular Monitoring of natural frequency for active volcanic mountain can facilitate our understanding of its dynamic change, such as the intrusion of magma for prediction of the eruption. In this study, we verify if we can estimate the frequency characteristics of Mt Fuji, the highest mountain in Japan, with microtremor observation. The microtremor observation was conducted from 6 to 9 August 2012. 7 locations are prepared in the observation at the 2nd and from the 5th to the 10th stations of Mt.Fuji. We temporarily installed a three-component accelerometer and a data logger at each station. In the analysis, we made a spectral analysis of the observed records, and we found the predominant frequency around 0.2 Hz in the NS component. Amplitude distribution at this frequency is similar to fundamental mode shape of vibration. However, the vibration at the 6th station at the predominant frequency shows slight different features. We confirm from a cross-correlation function in the vicinity of the predominant frequency that delay time between the 6th and 10th stations is greater than others. The result suggests the vibration mode changes near the boundary of the 6th station of Mt.Fuji. This feature of the vibration may be related with the subsurface structural changes around there, because it is located near the boundary of Older Fuji and Younger Fuji or it is close to the volcano Hoei. We need to discuss this from long-term observation data. We also conducted eigen value analysis with FEM using a simple cone model; 20km in diameter and 3km height. The first natural frequency of the model is about 0.2 Hz, and this is almost the same as the results with the observations. This shows that it is possible to estimate the frequency characteristics of Mt.Fuji with microtremor observation. However, the used model was a very simple model, and it is necessary to consider a model closer to the actual model for detailed investigation. Moreover, we need to verify how the natural frequency changes with changing the properties. We thank participants of the observation in this study. We are also indebted to the people in the mountain hut. We would like to sincerely thank them. In this study, we propose a method to construct a subsurface structure model using microtremor, gravity and magnetic data in the Tottori plain. Recently, different types of physical exploration data are analyzed simultaneously to improve uniqueness and accuracy of subsurface structure model (e.g. Sakai and Morikawa, 2005). In the target area, granite or sedimentary rock is found for seismic basement (Geological survey of Japan, 2003). The difference of densities is about 0.2t/m3 between the two rocks. Therefore we cannot perform gravity analysis with simple homogeneous two layers model, sediment layer 2.0t/m3 and seismic basement 2.4t/m3 (Noguchi et al., 2003). To overcome this problem, we employed magnetic data with the gravity data, and performed gravity analysis assuming several types of basement rocks with different densities. We applied the MWP (moving window Poisson analysis) method (Chandler et al., 1951) to get boundaries where densities change, and estimated depth distribution of seismic basement from gravity anomaly data. Based on the result, we estimated S-wave velocity structure model through inversion analysis of phase velocities of microtremor array observation data (Noguchi et al., 2003). As the result, we constructed a 3D subsurface structure model with three sedimentary layers and bedrock layer in the target area. We have built the structure model which can evaluate the strong ground motion characteristic of a broadband for the purpose of the advancement of strong motion evaluation. The built contents are the structure models which unify the subsurface part structure model and the deep structure model, and can reproduce seismic observation record. In this report, the contents of examination of the structure model construction in a south Kanto area(5 prefectures except Tochigi and Gunma) and a concentrated deformation zone (Niigata, Yamagata, and Akita) area are introduced. The final contents of examination adjusted the flow of structure model creation, and the valuation method of the periodic characteristic and the amplifying characteristic of the structure model. The result is improved for the period about 1 second in all the investigated areas. In this study, underground structures were estimated by array and single 3-componet microtremor observations. S-wave velocity structure models with 3 to 5 layers at the 10-sites were determined from array observation records. Predominant periods of H/V at 126 sites were obtained from 3-componet observation records. The S-wave velocities of alluvial layers were form 140 to 300m/s. The predominant period was about 1 second that H/V spectral ratio has clear single peak models near the coast line area. Therefore soft alluvial layer was distributed coast line area. Depth to bedrock (S-wave velocity is 600m/s layer) was about 90m maximum in the area. We present the test calculation of simultaneous estimation of subsurface structure with gravity and magnetic data. The simultaneous estimation is performed by the construction of sensitivity matrix and its inversion. For this purpose, we developed a new numerical code for solving the singular value decomposition based on I-SVD scheme. Usually, the sensitivity matrix is ill-conditioned when the number of the observation data and the model points is large. We need some regularization to solve this ill-conditioned inversion. Some technical discussion is also presented. condition smooths out the adjacent fluctuation. For this purpose, we developed a new numerical code for solving the singular value decomposition based on I-SVD scheme. Our numerical code is written by Fortran 95 in double precision, except for the lowermost DO loop for singular values calculation. This part is need to be written in quadruple precision. With this code, we can reproduce the model subsurface structure from the model gravity and the magnetic data set. We will also present the miscellaneous techniques for matrix regularization in the poster.The inclusion and the unification of microtremor data in this code may also be presented. We have applied seismic interferometry to ambient noise data recorded at Hi-net stations around Chukyo area, central Japan, to estimate velocity structure of the sedimentary basin (Hayashida et al., 2012). The estimated group velocities of surface wave from the stacked cross-correlation functions (CCFs) sho