Muography. Группа авторов

Figure 1.7 Density variations of the central southern region of Honshu, Japan. The densities plotted in this map were derived by excluding the data from the low‐density regions located underneath the seismic fault segments. (a) The map of Japan along with an indication of four tectonic plates. The dashed lines indicate the interface of these tectonic plates. (b) The enlarged map showing the geographical configuration of Izu, Miura and Boso peninsulas. (c) The muographically surveyed area in Miura and Boso Peninsula. (d) The muographically surveyed area in Izu Peninsula. (e) Geological maps of the Miura peninsula, indicating the distribution of the accretionary complex formed (1) between 40 and 20 million years ago; (2) marine and non‐marine sediments formed between 15 and 7 million years ago; (3) those formed between 7 and 1.7 million years ago; (4) those formed between 1.7 and 0.7 million years ago; and (5) those formed between 0.7 and 0.15 million years ago. Black dots show the locations of observation points. (f) Muographic image of the subterranean density distribution of Miura peninsula. The colors in this map indicate the terrain density. The region used for plotting this figure is shown by the inset in (e). Two blue dashed lines respectively indicate Kinugasa and Kitatake active fault segments.

      The density variations between the submarine volcano‐originated peninsula and accretionary prisms‐originated peninsulas could also be found in the results of the gravimetric measurements, which compared Izu peninsula (2.2–2.3 g/cm³) with Kii peninsula (2.5–2.7 g/cm³) (Nawa et al., 1997). The gravimetric density of Izu peninsula is in agreement with the muographic density before excluding fault segments. While gravimetric measurements provide geologically meaningful subterranean bulk density information, the derived density values are strongly affected by local topography and geology. In particular, sharp offsets generated by faults in Bouguer anomaly result in erroneous and unreasonably high or low density (Nawa et al., 1997).

      As a consequence, Tanaka et al. (2015) derived bulk densities for different rock types: (i) accretionary complex formed between 40 and 20 million years ago (2.79 ± 0.05 g/cm³); (ii) marine and non‐marine sediments formed between 15 and 1.7 million years ago (2.57 ± 0.03 g/cm³); (iii) marine and non‐marine sediments formed between 1.7 and 0.7 million years ago (2.43 ± 0.05 g/cm³); and (iv) basaltic and andesitic rock formed between 7 and 1.7 million years ago (2.51 ± 0.05 g/cm³). These muographic values were consistent with the gravimetric values of 2.63 ± 0.09 g/cm³, 2.54 ± 0.15 g/cm³, and 2.58 ± 0.2 g/cm³, respectively, for Mesozoic, Tertiary, and Quaternary sedimentary rocks, and 2.53 ± 0.15 g/cm³ for Tertiary volcanic rock (Nawa et al., 1997).

      1.3.5 Underground Water

      Risks in landslips (smaller‐scale landslides, mudslides, debris slides, and flows), landslides (larger‐scale mass movements), and flood in tunnels (for example, the flood accident known as “The Great Spring” inside the Severn Tunnel in 1879) are all related to the existence and mobility of subterranean water. Therefore, monitoring the subsurface water content is a key factor to predict underground‐originated disasters. The subsurface water content has been monitored with water gauges and soil moisture sensors inserted into vertically and horizontally drilled boreholes (Yuliza et al., 2015), however, from only a few data points, it is generally difficult to extrapolate the actual movements of underground water that are affected by the complicated surrounding geology. Deployments of an array of multiple sensors in multiple boreholes are not practical, thus our current quantitative understanding is far from the level of constructing a satisfactory model for prediction. As a consequence, the current risk management is rather palliative, for example, concrete shotcrete to stabilize mechanically unstable slopes or drainage tunnels to lower the level of the underground water table (Tanaka & Sannomiya, 2012).

Muography can measure (i) variations of bulk water contents over a much larger area than possible with a soil moisture sensor or a water gauge; and (ii) the total mass of water, inflow and outgoing, within a given volume. Examples are shown in Fig. 1.8. The extensive precipitation (> 100 mm/day) and muon flux have a negative correlation in the fracture zone associated both with the seismic fault (Fig. 1.8a) (Tanaka et al., 2011) and the landslide fault (Fig. 1.8c) (Azuma et al., 2014). The difference between these faults is either vertical (seismic fault) or horizontal (landslide fault). A delay in timing of the muon flux reduction between higher elevation (700 mrad) and lower elevation (500 mrad) corresponded to the motion of rainwater from a higher place (220 m a.s.l.) to a lower place (185 m a.s.l.), however, no such a delay was observed in the landslide fault. This simultaneous muon flux reduction was consistent with the water gauge results that showed a quick ascent (~15 m) of an underground water table from the steady elevation (480 mrad; dashed line in Fig. 1.8d) within a short time (a few hours) at a local water well (W‐10). The soil underneath an elevation of 480 mrad was already saturated with water. Therefore, there was no rainfall‐muon‐flux correlation at a lower elevation (480 mrad) as shown in Fig. 1.8c.

The precipitation of 100 mm increases the mass of the soil only by 100 kg/m². The resultant fractional density variations in the target volume will be less than 0.5%; hence the muon flux reduction of less than 1%. Therefore, muography is not so sensitive to the precipitation itself. In the previous cases, muography measured more localized larger‐scale mass movements due to permeation of rainwater and inflation of underground water. Three tons/m² of water moved through the seismic faults from surface to underground (at a depth of 35 m) within a few days, and 3.6 × 10⁴ tons of water inflow were measured in the underground water pathway. Rainwater gradually permeates to the underground direction, and inflation of underground water gradually propagates to the downstream direction. However, this propagation direction was not so simple. For example, the underground water level that was located closer to the detector did not change at W-9 (~50 m from the detector) (Fig. 1.8c). In general, the increased water flux couldn't be processed uniformly within the complicated underground structure, and the water level locally ascended like drifts of snow or leaves. If there is no water inflow or outgoing, the absolute amount of water situated inside the underground porous media can be derived only when a degeneracy of porosity and fraction of water occupation can be solved by, for example, joint inversion with seismic measurements (Matsushima, 2019).