Gamma-Ray Spectrometry in Radioactive Prospecting: Application Tool as Detecting Fault Trace

Abstract

In order to clarify the possibility of fault detection by gamma-ray spectrometry (GRS), radioactivity prospecting including soil radon gas survey, etc., was carried out on already-known faults of four areas included active faults. As a result, it was shown that it is possible to detect fault traces with the following three indices: (1) the continuity of anomaly points with an increasing rate Bi/Tl above the threshold defined for each study area. (2) The continuity of peak points of non-diffusive radon. (3) The continuity of conversion points where the average value changes in the stepped fluctuation pattern of nuclide and nuclide ratio. The index of (1) was near all four faults within 0–30 m of the known fault location. All fault-related anomalies of the index (1) were formed by local maxima in Bi and local minima or decreasing trends of Tl. Therefore, it is difficult to detect faults using only total gamma ray measuring equipment such as survey meters, which has been done so far. In order to continue to develop analysis technology for GRS, in addition to accurate GRS measurements, it is also necessary to develop analysis technology using Artificial Intelligence (AI) technology, etc.

1. Introduction

Development of the radioactive prospecting in the early stage was closely related to the uranium deposit development. After World War II, the nuclear development competition (nuclear arms race) intensified. A lot of uranium (U) deposits were discovered by the radioactive prospecting, etc. In the United States of America, although only a little uranium vein was known until 1948, 525 U deposits were discovered in 1953. Although there was only one U mine in Canada in 1952, Canada came to account for about 30% of the world U production in 1959 [1]. The former Soviet Union established the government-run U development enterprise in the Czech Republic, Germany, and Poland. The center of uranium development was Erzgebirge and the circumference near the border between the Czech Republic and Germany. U deposits in 164 places were discovered from 1946 to 1990 [2]. The radioactive prospecting played an important role in these uranium developments. The radioactive prospecting is divided into soil radon (Rn) gas survey that measures α-ray from soil radon gas and gamma-ray spectrometry (GRS) that measures gamma-ray.

The terminology of GRS differs as follows depending on the different methods of acquiring gamma-ray spectra: Air-borne is measured by mounting a measuring device on an aircraft or helicopter, Car-borne is measured by mounting a device on a car, Man-borne (portable measurement) is measured by a person on the ground with a portable detector, etc.

Gamma-ray spectrometry could not discriminate radionuclides from the 1940s to the 1950s, but since the mid-1960s, the demand for monitoring the effects of nuclear tests and advances in computer technology have made field measurement of potassium (K), U, and thorium (Th) possible. The International Atomic Energy Agency (IAEA) began studies in the 1970s to standardize this method. It led to the publication of the first standard technical guide in 1991 [3]. In 2003, the IAEA published a second guideline that added the gamma-ray spectrometry theory to consider the use of GRS for environmental surveys other than for geological survey purposes [4]. This completes the technical system. GRS has already become a mature measurement method with measuring instruments and standardized analytical methodologies [5]. One of the remaining issues is the development of utilization technology for using the U, Th, and K concentration data measured by GRS.

Suran [2] examined which radiological exploration method contributed to the discovery of the 164 U deposits discovered in the Czech Republic between 1946 and 1990. He found that the most effective method was soil Rn gas survey (44%), while air-borne survey (3%) and car-borne survey (9%) were less effective. Therefore, the results of Suran [2] show that it is difficult for air-borne and car-borne surveys to directly identify uranium mines as uranium anomalies.

The reason why GRS, especially air-borne, has regained attention in the field of exploration is that GRS has found not only U and Th deposits but also many metal deposits in the altered zone defined as the high K and low Th/K ratio (e.g., [6]). This is the first turning point in the development of technology for using GRS. This turning point occurred due to the understanding of radionuclide behavior in hydrothermal and alteration/weathering processes. Since this turning point, GRS has been accepted as a geological survey technique for mapping wide-area radionuclide concentration zones, rather than as a technique for exploring the location of anomalous radiation spots.

Discontinuous planes of all sizes, such as faults and joints, are present in rock masses. In particular, research about faults is of special interest to researchers in scientific and engineering disciplines. Because a fracture zone has extensive shearing, the strength and modulus of elasticity of a fracture zone are lower and the hydraulic conductivity is higher than that of the surrounding area. Consequently, assessment of the fracture zone’s properties is considered important to the selection of a site, stability analysis, and design of construction works.

Not only for researchers related to earthquake hazards, but also for the general public, distinguishing between “inactive” and “active” faults is one of the most important issues because of the seismic hazard commonly associated with fault activity. In general, scientists define active faults as follows: (1) evidence of historical and/or instrumental seismic activity and/or (2) stratigraphic/morphologic displacements within a time period which can span from the Holocene (10,000 years) to the entire Quaternary. Moreover, it is important to remember that fault slip behavior is often complicated by the mixture of two endmember styles: stick slip (seismic fast slip) and stable sliding (continuous slow creep). These styles can occur in different sections of the same fault and occur in the same section at different times. While the first tends to rejuvenate and keep the fracture network open near the fault, the second allows for fault self-shielding to occur due to the precipitation of fluid constituents flowing through the fracture [7].

Fault trace survey is a major theme of radioactive prospecting. Ambron [8] was the first to find the relationship between faults covered with sediments (buried faults) and radioactivity through α-ray surveys using an ionization chamber. He suggested that the peak positions of α-ray survey can be applied to fault trace survey. Lane and Bennett [9] reported that radon concentrations in groundwater are indicative of known faults. However, it was difficult to investigate a wide area of faults of interest with the α-ray analysis technology at that time. Therefore, Ochiai [10] developed a vehicle-mounted scintillation counter that can continuously measure γ-rays as a Radon index on the ground based on the assumption of the radiation equilibrium of uranium series nuclides. He estimated locations on the faults for bedrock groundwater extraction wells from sequences of γ-ray anomalies.

After his research, in Japan γ-ray prospecting with a survey meter was frequently used for detecting fault traces. This is in contrast to the general use of γ-ray radioactivity prospecting research in the world for the purpose of mineral exploration. On the other hand, some studies in Japan denied the effectiveness of this method for fault survey from some cases where γ-rays did not increase on the fault (e.g., [11, 12]). Imaizumi et al. [12] suggested that the contribution of Bi from soil pore Rn gas to the γ-ray count was about 1.5% based on model calculations, and that it is difficult to estimate the soil Rn gas concentration from γ-rays.

In the field of oil geophysical prospecting, γ-ray (GR) logging was introduced in the late 1930s as the first non-electrical logging method [13]. GR logging is useful in distinguishing between clean sand and shale formations. Then in the 1950s, spectral γ-ray (SGR) logging that could discriminate between nuclides was developed. An initial successful application was the detection of radioactive fractures in Austin Chalk wells, Texas, USA [14]. The Austin formation was originally interpreted as a shale zone. SGR logging was able to distinguish between shale (moderate potassium (K) and thorium (Th) content but low uranium (U)) and U-bearing fracture zones (low K and Th but high U). Based on this information, recompletion of wells increased production sevenfold in some cases [14].

Kimura [15] introduced γ-ray spectrometry for measuring three nuclides into fault surveys and proposed using the Bi/Tl or Bi/K ratio (hereafter, the ratio of three nuclides) as a fault index. He developed a car-borne survey system that detects faults from the increasing rate that is the ratio of the measurement point to the moving average around the measurement point. Imaizumi [16] verified the effectiveness of Kimura’s increasing rate on various faults by combining Kimura’s system [15] with other surveys such as soil radon gas surveys. The results confirmed the effectiveness of the increase in rate [12, 16, 17, 18].

Since 2000, thanks to the development of low-cost, high-precision Si semiconductor detectors for alpha-ray detection and advances in geostatistical analysis technology, significant increases in radon carrier gas (e.g., CO₂) and radon gas were found along active faults (e.g., [7]). In other words, the effectiveness of the alpha-ray exploration method as a tool for fault trace survey has been established. Moreover, numerous radon gas data have advanced our understanding of the structural properties of faults (echelon faults, open cracks, fault clay-sealed cracks, etc.). According to these studies, radon anomalies along fault lines can be continuous or intermittent [7].

On the other hand, regarding γ-ray fault surveys, the relationship between γ-rays and faults remains ambiguous. For example, Szabó et al. [19] reported that faults can be detected using geostatistical spatial structure analysis even with sparse gamma-ray dose rate data (average 3 points/100 km²), while Jolie et al. [20] suggested that gamma-rays are affected by man-made structures other than faults, so this survey method is not necessarily effective. Yoshimura and Matsumoto [21] showed various examples of the fact that distribution patterns of anomalies depend on the form of the fracture zone (fault clay or open crack, etc.). Some cases even showed that the dose rate decreased over faults. Therefore, at present, even the assumptions of the γ-ray fault surveys are suspicious.

The reason why the radiation dose increases on the fault is not only the radon gas rising along the fault crack but also the concentration of radon parent nuclide elements U and Ra in the fault clay. Therefore, it is impossible to uniquely determine the distribution pattern of γ-ray nuclides, which is an indicator of faults, unless the physicochemical conditions of the faults are elucidated. One way to overcome these problems is to classify and catalog behavioral patterns of gamma-ray-emitting radionuclides (⁴⁰K, ²¹⁴Bi, and ²⁰⁸Tl) and alpha-rays (soil radon gas) along survey lines crossing faults of various geological settings and activities (active and inactive faults). We can extract fluctuating patterns of γ-ray nuclides as common fault indices based on the data. However, few studies have clarified the fluctuation behavior of γ-rays on faults caused by U leaching and adsorption/enrichment on clay.

In this paper, Section 2 summarizes the behavior of γ-ray nuclides during weathering and faulting processes, and describes the results of RGS and soil gas radon surveys of exposed faults where the soil has been stripped away. Based on the results, indices for fault detection, such as the Bi/Tl ratio, are proposed. Section 3 presents surveys of four known faults using the proposed index. Section 4 presents the results of regional fracture system surveys, geological surveys, and weathering surveys of the Atera fault area and the southern Abukuma area by a car-borne survey. The original data for Sections 3 and 4 are taken from Imaizumi [16] and Imaizumi et al. [12, 17, 18]. However, these data have been reviewed using Geographic Information System (GIS) techniques. Finally, the study results are summarized and discussed in Section 5, and future studies are recommended based on the indices established by this study.

2. Environmental γ-ray radionuclides

Gamma-ray spectrometry (GRS) is already a mature technology, and its principles and analysis methods have been compiled by the IAEA [3, 4]. Here, only the basics of using GRS as a fault detection method are described. See the IAEA for other basics.

2.1 Natural γ-ray radionuclides

Radioactivity is a phenomenon that changes into a stable nucleus when atomic nucleus with unstable balance of proton and neutron emits elementary particles of α particles (radiation), β particles, and γ photons (ray) over time. Elements with this feature are called radionuclides. γ-rays are electromagnetic waves of surplus energy emitted when an unstable excited nucleus changes to a new stable nucleus. When γ-rays pass through matter, they interact with the electrons and nuclei of matter atoms through phenomena such as photoelectron effects, Compton scattering, and electron pair generation. Under the practical conditions of GRS, Compton scattering is the dominant process. Gamma-ray radiation from soil is predominantly attenuated by water, soil, and organic matter [22].

The most of natural radionuclides that were produced during the formation of the Earth and concentrated in the crust turned into stable nuclides after 4.5 billion years. Currently, there are 17 nuclides with a half-life of more than 700 million years. Among the existing nuclides, the only nuclides that have high radioactivity and are subject to γ-ray spectrometry are ⁴⁰K, ²³⁸U, and ²³²Th. In all, 0.012% of potassium (K) is ⁴⁰K, 83.3% of which is β-ray-emitting nuclide, and 10.7% is decayed to ⁴⁰Ar by EC (electron capture), and 1.46 MeV γ-ray is emitted in the process. ²³⁸U and ²³²Th decay to produce other nuclides (daughter nuclides), and the daughter nuclides undergo further radioactive decay to form a radioactive decay chain (Figure 1). Only ²¹⁴Bi and ²⁰⁸Tl are the nuclides that emit γ-rays that are the target of γ-ray spectrometry (Because the counting rate of γ-rays of ²¹⁴Pb and ²²⁸Ac (actinium 228) is low, they are not the target nuclides of γ-ray spectrometry described below).

Figure 1.

Radioactive decay series of ²³⁸U and ²³²Th. Figures from Wikipedia decay chain uranium series and thorium series are edited and added. *https://en.wikipedia.org/wiki/Decay_chain, **https://commons.wikimedia.org/wiki/File:Decay_Chain_of_Thorium.svg

Since the half-life of each series of parent nuclides is significantly longer than that of daughter nuclides, the radioactivity of each nuclide in the series is closed state in secular equilibrium, which is the same as that of the parent nuclide in a closed system. In secular equilibrium, the radioactivity of all nuclides in the series is equal, so the nuclide concentration at one stage of the decay chain can be estimated from all daughter nuclide concentrations. Since ⁴⁰K exists in a fixed ratio to the non-radioactive K isotope, K (%) can be analyzed directly from ⁴⁰K γ-rays of GRS. The U concentration of GRS is estimated from ²¹⁴Biγ rays of the daughter nuclide of 238 U. Therefore, the uranium concentration by GRS is an indirectly estimated concentration. Similarly, Th concentration is estimated indirectly from ²⁰⁸Tl γ rays of the daughter nuclide of ²³²Th [4]. When one or more decay products in the decay chain are completely or partially removed or added to the system, the decay chain becomes non-equilibrium state. ⁴⁰K has nothing to do with the non-equilibrium problem. In the Th series, non-equilibrium state rarely occurs due to the low mobility of daughter nuclides. However, in the U series, non-equilibrium states inevitably occur due to selective leaching of decay products (e.g., ²²⁶Ra), diffusion of ²²²Rn gas from soil, and dissolution of ²²⁶Ra in groundwater [6].

2.2 Geochemical behavior of radionuclides

2.2.1 Behavior of K, U, and Th in the igneous process

The concentrations of K, U, and Th in igneous rocks vary systematically from ultramafic to felsic. These elements have large ionic radii and are called incompatible elements because they do not fit easily into the crystal lattice of minerals. Therefore, they stay in the magma until the final stage of crystallization differentiation. In igneous rocks, the higher the SiO₂ concentration, the higher the radionuclide concentration in the rock [23]. In acidic rocks such as granite and rhyolite (SiO₂: more than 66%), U is found in the main minerals like feldspar and mica, as well as in accessory minerals (zircon, monazite, etc.) or as a tetravalent element in the interstitial space of the crystals. Th also occurs not only as a tetravalent element in a separate mineral, but also by replacing U, Zr, rare earth elements, etc., in the mineral [23].

2.2.2 Behavior of U in intrusive rock emplacement process

Intrusive rock is formed when plutonic rocks such as granite and diorite intrude and solidify into the crust as batholiths and dykes. It is well known that rocks that touch granite intrusive rock undergo contact metamorphism and contact metasomatism because of the intrusions. Analysis of the batholith’s drill core shows that the closer to the surface layer, the stronger the radioactivity. In other words, radioactive materials are more concentrated near the roof pendant (Figure 2(a)).

Figure 2.

Radioactivity distribution in granitic intrusive rocks (a) and radioactivity distribution near the interface of contact metamorphism (b) (edited from figures in Hatsuda [24]). P: Roof pendant, H: High activity part, and L: Low activity part. The density of dots represents the density of radioactive elements.

Uranium remains in magma until the final stage of magma crystallization differentiation. As it consolidates into crystals at the terminal stage, a relatively distinct radioactive anomaly forms at the upper rock boundary of the granite body. Under the same conditions during contact metamorphism, the curves of radioactivity intensity distribution across the contact area of the intrusive rocks show similar patterns. These patterns in Japan are divided into four types, as shown in Figure 2(b) [24]. Type I is a pattern that appears during typical contact metamorphism. It is thought that this type of intrusive rock boundary can be detected as an anomaly in radiological surveys. An example of this is shown in the car-borne survey of the Atera fault area in Section 4. The pattern of radioactive anomaly becomes weaker from type II to type IV. Hatsuda [24] interpreted that type IV indicates the radioactivity distribution in the deep part of the intrusion where granitization was remarkable. An example of this is shown in car-borne surveys of the Abukuma area of Section 4.

2.2.3 Behavior of K, U, and Th in the sedimentary process

Ordinary sedimentary rocks always contain U and Th, although there are differences depending on the region and material. U is correlated with C in sedimentary rocks and is relatively well correlated with K in Neogene mudstones. Humus and other organic matter adsorb U. U is also adsorbed by iron hydroxide and clay minerals. U is more abundant in rocks like black shales that are deposited in strongly reducing sedimentary environments with a lot of organic matter and mainly clay minerals. U is also concentrated in phosphorite. Therefore, the distribution of U concentration in sedimentary rocks is determined by the paleogeography of sedimentary basins [25].

2.2.4 Behavior of K, U, and Th in the weathering process

When rocks underground are exposed to the temperature and pressure above the ground, they weather and decompose, changing their composition. Rock components dissolve and recrystallize to form clay minerals, turning the rock into a parent material for soil. Moreover, when organisms and organic matter interact with the parent material, soil is formed. The concentration of radionuclides in soil mainly depends on the geology of the bedrock, the formation of clay minerals by weathering, and the geochemical behavior of radionuclides [6]. Generally, the composition of radionuclide concentrations in the parent material is inherited by the soil, but some are changed through the pedogenesis process. General trends in soil concentrations of K and U, Th series nuclides show a positive correlation as well as rock trends.

Potassium (K) is rich in felsic rocks because it is found in main minerals like feldspar and mica. It is rare in basic and ultrabasic rocks. It is not found in carbonate rocks at all [22]. Feldspar and mica are present in the sand fraction in the early stages of soil formation. But they recrystallize into clay minerals and disappear because of weathering. The K in the feldspar crystal lattice is usually washed away by rainwater when the feldspar changes into kaolinite with almost no K. When mica changes into illite at the first weathering stage, some K stays in the illite crystal lattice. At more weathering, illite changes into low K-concentration clay minerals (vermiculite and smectite). At the final weathering stage, low K-concentration clay minerals change into kaolinite [26].

Uranium and Th have similar chemical behavior in magma. In unweathered igneous rocks, both elements are tetravalent. But in the process of weathering, they show a completely different chemical behavior. This is because of the difference in sensitivity to oxidation and reduction changes. In the oxidized atmosphere, U becomes hexavalent uranyl ion (UO₂²⁺) and forms many kinds of complex ions. Since these complex ions have relatively high solubility, hexavalent uranium easily moves with groundwater. But uranium in a water solution is adsorbed to an adsorbent such as clay minerals, organic matter, trivalent iron hydroxides, and zeolites. Montmorillonite has more uranium exchange capacity than other clay minerals. When hexavalent uranium is reduced for some reason, it becomes tetravalent and precipitates as poorly soluble uranium dioxide (UO₂), and recrystallizes as a secondary mineral. So, U is sensitive to oxidation and reduction changes and is mobile [26]. On the other hand, Th keeps its tetravalent state, so its solubility and mobility are low [26]. Even after K and U are removed by weathering, Th remains as a residue and tends to keep the source rock composition. Note that in the analysis of GRS data, this Th characteristic, specifically the Th nuclide ratio, can be used to analyze the migration and enrichment of U.

2.2.5 K/Th and U/Th ratios as the indicator of the sedimentary environment

Based on the difference in how U and Th react to oxidation and reduction changes, the Th/U ratio (hereafter, the “ratio” is omitted) is used as an indicator of the paleo-redox environment during deposition, and Th/K is used as an indicator of clay mineral types.

2.2.5.1 Th/U

Adams and Weaver [27] concluded that Th/U ratios are often strongly related to the sedimentary environment based on laboratory analysis of many samples. So far, studies have shown a strict relationship between the Th/U and the sedimentary environment. This relationship shows that high Th/U is typical of the continental environment and low Th/U of marine settings. The boundary value of this ratio was determined as follows: Th/U > 7 for continental deposits, Th/U < 7 for marine deposits, and Th/U < 2 for high chance of reducing conditions. Th/U > 7 for high chance of oxidation conditions [28].

2.2.5.2 K/Th

Schlumberger [29] proposed a reference thorium-potassium cross-plot that distinguishes five typical minerals that are important for the oil industry: chlorite, glauconite, illite, kaolinite, and smectite by the Th/K ratio, based on many data analysis results (Figure 3). In this figure, plagioclase (Th/K = 0.3–0.6) has the lowest Th/K ratio. The Th/K ratio increases in the order of mica, illite, mixed layer clay, and montmorillonite. Kaolinite as the final weathered mineral has Th/K = 12–25.

Figure 3.

Thorium-potassium cross-plot for mineral identification. The Th-K cross-plot is modified from Schlumberger [29] by Nuţu-Dragomir et al. [30]. The data plotted are for each rock body in the Tanagura fracture zone in the southern Abukuma region. See text in section 4–2 for details.

Many researchers have used them to identify types of clay minerals. Nuţu-Dragomir et al. [30] used an improved Schlumberger Th-K cross-plot for fault survey in shale and marl regions. Their Th-K cross-plot (Figure 3) showed that the clay minerals in the fractured zone were montmorillonite and the clay minerals in the outcrops of non-fractured zone were illite-smectite mixed layer minerals. Stahr et al. [26] applied the cross-plot for the discrimination of soil types of Acrisols (cation exchange capacity (CEC) <24 cmol. kgL⁻¹) and Alisols (CEC ≥ 24 cmol. kgL⁻¹) that require laboratory measurements in northern Thailand. Th/K = 16 by GRS in the field was able to distinguish between Acrisols and Alisols.

Using cross-plot diagrams to determine the types of clay minerals has some drawbacks, because the interpretation results cannot be shown as a function of depth. Showing the Th/K as a function of depth allows us to track the changes in the types of clay minerals in the stratigraphic column. In this case, the relationship between K/Th and clay minerals is classified by the K/Th value in Table 1.

Th(ppm)/K (%)	Minerals
<0.6	Feldspar
0.6–1.5	Glauconite
1.5–2.0	Mica
2.0–3.5	Illite
>3.5	Mixed-layer clays
> = 10	Kaolinite and chlorite

Table 1.

The relationship between K/Th and clay minerals [28].

2.2.6 Behavior of K, U, and Th around the fault fracture zones

2.2.6.1 Structure of the fault fracture zones

A fault is shown as a geometrical plane on geological maps. But an actual fault has the following forms: The fault gouge (fault clay) and the fault breccia (both together called the intrafault materials) that are formed by the crushing and friction of fault movement are layered between two fault planes that run almost parallel [31].

When a large-scale shear fracture occurs, the breccia and clay mix because the rock mass is broken in a relatively wide width. This state, made of clay (fault gouge) and relatively coarse-grain fault breccia, is called a fault fracture zone. In the fault fracture zone, the rock mass, which usually has many cracks, lies beyond a certain constant width of breccia and clay. The cracks gradually decrease and turn into normal rock at a certain distance from the center of the fracture zone.

A clay mineral in fault gouge is usually formed by chemical weathering and alteration due to reaction with groundwater after crushing [31]. The texture of the fault fracture zones seen in outcrops shows a banding structure, with a zone mainly made of fault gouge (clay zone) and a zone of fault breccia (brecciated zone) running parallel to the fault. Generally, the clay zone is located on both sides of the fracture zone (Figure 4) [31]. Therefore, U tends to be enriched in clays at both ends of the fracture zone.

Figure 4.

Schematic figure of fault fracture zones (modified from Tanaka [31]).

2.2.6.2 Geochemistry of the fault

The formation of the clay minerals in the fault fracture zone is strongly influenced by the type of host rock (original materials) and the surrounding temperature and pressure, as well as the generation of clay minerals by general weathering. The composition of the clay minerals in the Atotugawa fault fracture zone in Japan showed that feldspar and quartz, etc., which are main minerals in the host rock (granite), are hardly found, but montmorillonite has been newly formed [32]. The amount of montmorillonite, compared to the whole mineral composition, is high in the argillization zone, and low in the brecciated zone.

The non-equilibrium data of uranium series show that U very slowly leaches from the primary mineral [33]. On the other hand, the crack in the fault is often characterized by various filling and alteration materials. The most common materials are hematite and goethite. It is often observed that uranium is concentrated around the secondary Fe⁺³ mineral. Smellie et al. [34] showed a Fe-U-Ra vertical profile in the crack of faults in three different granite regions, where total Fe, Fe³⁺, and U generally increased toward the crack, although the Th concentration was almost constant.

2.2.6.3 U/K and Th/U curves as indicators for crack exploration

The principle of detecting fracture systems in carbonate formations by spectral γ-ray (SGR) logging is based on the fact that under the reducing conditions in which carbonate rocks form, hydrothermal circulation can cause uranium salts to precipitate within the fractures [28]. However, in complex crack systems, low concentrations of radioactive elements can obscure the crack locations in the uranium curve of the SGR curve. Therefore, plotting the ratio curves, such as U/K, Th/U, and Th/K, helps to better identify the crack locations. Of these ratio curves, matching the U/K and Th/U curves is highly recommended. These curves are inversely proportional, with changes in uranium content plotting the curves in opposite directions [28].

2.3 Indices for fault trace detection for radioactivity prospecting

Buried faults are faults that are covered by newer sediment or soil and cannot be seen directly. The planned inspection gallery site of a dam with a fault is a valuable example where the buried fault can be seen because the surface sediment has been removed. The planned site for F Dam was one of these examples.

Figure 5 shows a geological section of the valley across the M River at the F dam site. The width of the M river is about 20 m. The slope on both sides of the valley is about 30°. About 140 m above the slope there is a plateau. The geology of the plateau consists of Neogene basalts in the upper part and Mesozoic granodiorite in the lower part. The dashed line in the figure shows the terrain before the dam construction. At the time of the survey, river bed gravel, river terrace deposits, and weathered parts of the bedrock had been taken away to expose fresh bedrock, so F1 faults could be seen (Figure 6).

Figure 5.

Geological section of the valley across the M River at the F dam site.

The fracture occurrence of bedrock is explained by Japanese rock mass classification system. Rock mass classification in Japan classifies rock mass into four ranks from A to D. C ranks are subdivided into C_H (high), C_M (middle), and C_L (low). D ranks are subdivided into D_H and D_L. A and B indicate hard rock with few cracks, and C_H ∼ C_L indicate medium hard rock with cracks. Crack density increases from C_H to C_L. D_H is very soft, has no cohesion between joints, and falls apart with a slight hammer blow. Clay is on the peeled surface. D_L is a clayey state (e.g., [35]).

Figure 6 shows the bedrock classification map and the distribution of springs at the planned inspection gallery construction site. The classification map shows the locations of high-angle f1-f8 faults. Except for the f1 fault, there were small faults with a fracture zone width of 0.3 to 0.5 m. The f1 fault was represented by D_L rock with a width of 0.5 to 2 m. D_H rock with a width of 2 to 5 m was located on the west side. Five meters away from the f1 fault, C_H-C_M class hard rock with few cracks was distributed. Many springs were distributed in the riverbed. The amount of spring water from each spring flowed at the rate of several liters per minute (liters/min). The spring distribution was mainly located in the cracks of C_H rocks on the east side of the fault and the boundary cracks between C_H and C_M rocks. There was almost no distribution of springs near the f1 fault.

Figure 6.

The bedrock classification map and the distribution of spring water at the planned inspection gallery construction site of the F dam site.

A 50-m-long survey line centered on the f1 fault for radioactivity prospecting was set up. GRS and radon gas concentration were measured at 5 m intervals along the survey line. The GRS sampling time was 300 s. Methods for GRS and radon gas survey are described in Section 3.3.

Figure 7 (a) shows the fluctuation patterns of Bi, Tl, Bi/Tl, and Rn gas concentrations and (b) shows the distribution of fractures (extracted from Figure 6). The white areas in the figure show the C_M and C_H rocks with few cracks. Across the main fault zone from f1 to f6, where the density of fractures is the highest, Bi and Bi/Tl drop quickly to minimum values (indicated by a green triangle). On the other hand, Tl increased slightly. Since the pattern of Tl shows the original pattern, the Bi and Bi/Tl patterns may reflect the leaching of U from the host rock, granodiorite.

Figure 7.

Fluctuations of Bi, Tl, Bi/Tl, and Rn gas concentrations (a), the distribution of fractures (b), and the relationship between Bi and rock class classification for the entire excavated surface in the F dam site. In the figures of fluctuation curves below, the values have been processed in order to plot multiple curves in one figure. For example, Bi/Tl*1000 is Bi/Tl multiplied by 1000. See text for red and green triangles.

On both sides of the main fault zone, relatively large peaks in the Bi/Tl ratio were seen. The largest peak is about 10 m from the f1 fault on the east side (shown by the vertex position of orange triangle). In this part, Tl makes a local minimum. Since there is a proportional relationship between Bi and Tl in the original granodiorite, the original Bi should also have made a local minimum. But it is thought that Bi was locally maximized due to the enrichment caused by precipitation of U (indicated by a red triangle). Bi/Tl highlights this phenomenon more than the fluctuation of Bi (the orange triangle is bigger than the red triangle). The Rn gas concentration showed a big peak of about 3 kBq/m³ near the boundary between the D_H and C_L rocks, which is several meters away from the f1 fault (D_L rock). Note that this radon gas peak near the f1 fault does not affect the Bi fluctuation pattern.

Figure 7(c) shows the relationship between Bi and rock class classification for the whole excavated surface. As the bedrock condition gets better (D- > C_H), Bi tends to increase. This phenomenon shows that the leaching of U progresses with the weathering, and the lower the rock class, the lower the U content.

From this figure, the following points can be made about the detection of faults in granitic regions by radioactivity prospecting: (1) The fault fracture zone was the U leaching zone. (2) The locations of the local minimum of Bi and Bi/Tl are indicators of the central location of the fracture zone. (3) The local maximum locations of Bi and Bi/Tl, here the peak on the left side of the f1 fault was about 10 m away from the center of the fracture zone, which is the entrance of the fracture zone. (4) Although the Rn gas peak location is several meters away from the center of the fracture zone, Rn gas is the most sensitive fault indicator. (5) If an error of about 10 m is allowed, the local maxima of Bi/Tl also serve as fault indicators. (6) The leaching zone of U is wider than the enriched zone of U, so the local maximum point of Bi/Tl is better than the local minimum point of Bi/Tl as an index for exploration.

Based on the above observations, in the exploration of known faults in the next section, the local maximum peak of Bi/Tl (anomaly point) and peaks of radon concentration are chosen as indices for fault trace detection.

2.4 Definition of anomaly point for fault detection index

Traditionally, anomaly points have been defined as outliers with a threshold of mean ± 2 × standard deviation of a single population assuming a normal distribution in a parametric test, and anomaly points outside this threshold [36]. This method is intended to identify the extreme values of the statistical normal distribution, and geochemical problems often do not warrant statistical assumptions (e.g., [37]). In the analysis of radiological exploration, especially in the analysis of car-borne survey data, a survey line is set across the distribution of multiple background populations (stratum), so it is not possible to define outliers by statistical methods.

Kimura [15] proposed the following equations for increasing rate, specifically assuming the evaluation of anomalies by the car-borne survey (Figure 8). Increasing rate: F_{(f, p)} is defined as the ratio of the movement average of forward and backward to the measured value given by:

Figure 8.

Schematic diagram explaining the calculation method of radioactive survey anomalies by Kimura [15].

$$F_f=\frac{V_i}{\left(V_{\left(i-n-1\right)}+V_{\left(i-n\right)}+\cdots \cdots +V_{\left(i-1\right)}\right)/n}-1\times 100$$E1

$$F_p=\frac{V_i}{\left(V_{\left(i+n-1\right)}+V_{\left(i+n\right)}+\cdots \cdots +V_{\left(i+1\right)}\right)/n}-1\times 100$$E2

where V_i indicates the measurement value at i point. V_(i-n-1) and V_{(i + n + 1)} indicate the measurement values at the n-front and n-post of i point.

A radioactive anomaly cannot be detected by values of either Ff or Fb alone. But when both are used, the position of the stratum boundary line with different nuclide compositions can be identified as the point of rapid increase of either Ff or Fb. The position of the fault line can be clearly identified as the point of rapid increase of both (Figure 8). The increasing rate: R is defined as the smaller of Ff and Fp. Anomalous points are set at points above the threshold based on the coincidence of increasing rate points and fault lines at each field. For example, on the Atera fault, anomalies above the threshold of R = 10% tended to match with the fault location.

Figure 7(a) shows the R-value of Bi, Bi/Tl, and Rn gas concentrations and the rate of increase (F-value) from the average value of the survey line. The F-value for the small Bi peak is 20%, while the R-value decreased to 7.8%. This is due to the distorted peak shape (Ff = 42 and Fb = 7.8). But for Bi/Tl, which produces a clear peak, the F-value was 37%, but the R-value grew to 49%. Similarly, at the Rn gas concentration that produces a distinct peak, the F-value increased by 619%, but the R-value grew to 2872%. This large rate of increase shows that Rn gas is a sensitive fault index. In this way, when the shape of the peak is clear, using the R-value can increase the rate of increase from the simple average value.

3. Fluctuation patterns of nuclide, nuclide ratio, and radon gas concentrations across known faults

3.1 Method

3.1.1 Man-borne survey

The portable pulse height analyzer made by Clearpulse Co., Ltd. and scintillation detector (sodium iodide (NaI) crystal-like cylinder of 13 cm in diameter and height with a photomultiplier tube) were used for the man-borne survey. A detector with energy resolution of 7% or less was used. The NaI detector was placed on the surface of the earth for the measurements. Each measurement time was for a duration of 5 min. The error ($\sqrt{N}$) in the measurement value (N), in this case, was 6% or less ($\sqrt{N}/N$) concerning Bi, whose discharge energy is smaller than those of the other nuclides.

3.1.2 Car-borne survey

The pulse height analyzer (made by Clearpulse Co., Ltd.) with 1024 channels designed for car-borne survey and 18 NaI detectors (the same size as the man-borne detector with resolution 7% or less) were used for the car-borne survey. A stabilizer mechanism that tracks the peak of gamma-rays from a 137Cs source (37 kB) installed near the detector was constantly operating during measurements to keep the amplification rate constant. The detectors of car-borne were placed on one side of the luggage compartment and one side of the detectors were shielded to reduce γ-ray counts from the road pavement under the car. These measures gave direction to the measurement efficiency of the detector [15]. The stripping ratio method [3] was applied to correct the Compton scattering of the γ-ray spectra.

The measurement method was as follows:

1. Running car measurement [15]: The speed of movement was kept at 4 km/h. Sampling time was taken at 30 s intervals. The maximum measurement error ($\sqrt{N}/N$) of Bi was 5% or less. The measurement value indicates an integrated value of radioactivity distribution for about 30 m in length. The location of the measurement point includes an error of about 30 m back, either side, due to various conditions of the measurement.

2. Stopping car measurement: The car was stopped at intervals of 10 m for measurement. The measurement was done three times, every 30 s. The average value of the three measured values was used for analyzing the fault.

3.1.3 Soil radon and CO₂ gas surveys

The soil radon gas survey using the Picorad detection (1) and soil CO₂ gas survey using the detection tube (2) were used in conjunction with the GRS survey.

1. Picorad detection (Packard Co., Ltd.) that is a granular-activated carbon canister method was used for soil radon gas survey. A Picorad detection device consists of a 20-ml plastic vial with cap, including activated carbon. A hole was made with a diameter of 20 cm and 30–70 cm in depth. The Picorad vial, without a cap, was placed inside the bottom of the hole. The container was covered up with soil. The Picorad was left undisturbed for 24 or more hours to adsorb radon surrounding it. A liquid scintillator was injected into the detector after carrying it to the laboratory. The α- and β-rays of the liquid, into which the radon was transferred from the activated carbon, were measured by a liquid scintillation counter. The correction of adsorption and collapse under exposure, and the attenuation of time after collecting and measuring, was executed with the special software offered from Packard Co., Ltd. The limit of detection is 14.8 Bq/m³.

2. A detection tube made from glass of Kitagawa method (KOMYO RIKAGAKU KOGYO K.K.) was used for the soil CO₂ gas survey. A handheld vacuum gas sampler was used to vent a constant amount of sample gas to the detector tube. The measurement range was 0.1–5.2%.

3.2 Overview of topography and geology for the known fault test sites

3.2.1 Selection of known fault test sites in Japan

Since Japan has a complex geological structure, it is possible to investigate various geological faults. Four test sites with different geological settings shown in Figure 9 were selected. In this section, first, the relationship between the four test sites and the geological tectonic regions of Japan is described. After that, an overview of the topography and geology of each test site is presented.

Figure 9.

Location map of four known fault test sites on the geological tectonic region map of Japan. The geological tectonic region map is edited from the original map of the National Institute of advanced industrial science and technology (AIST) (https://www.gsj.jp/geology/geology-japan/geology-japan/index.html), and the active fault lines are edited from the original map of the active fault society [38]. Note that not all geological tectonic regions are shown.

Basically, the subduction in the trench and the related volcanic activity form the backbone of the Japanese island arc. Sedimentary rocks and metamorphic rocks in Japan, especially, are characterized by the continuous zoning of rocks that were formed over a certain period of time. These boundaries are often connected by faults, some of which are called tectonic lines. The geological structure of present-day Japan was formed during the new orogeny due to the development of the curved archipelago after the Miocene. Figure 9 is a map compiled by the Geological Survey of Japan showing geological regions in Japan that have been proposed so far. This figure shows the differences in rocks underlying the Japan island arc, mainly based on large-scale faults, accretionary prism formation age, and metamorphic age.

Japan’s largest island (Honshu) is roughly divided into two parts, northeastern Japan and southwestern Japan, by a rift zone called the Fossa Magna that runs north-south through the center (as indicated by <- a ->) (The tectonic line on its west side is the Itoigawa-Shizuoka tectonic line (b line), and the tectonic line on its east side is not clear. Here, it is indicated by a dotted line). The Fossa Magna has thick deposits of submarine volcanic products and sediments from the mid-Neogene period. They are covered by a north-south chain of volcanoes, including the Quaternary Mt. Fuji.

Cenozoic rocks are widely distributed in northeastern Japan. Many of the Pre-Tertiary rocks are isolated massifs. Granites are distributed in the Abukuma belt (2), and sedimentary rocks and granites are distributed in the Ashio belt (3). Both belts are separated by the Tanagura fracture zone (the d line). On the other hand, in southwestern Japan, Paleozoic and Mesozoic rocks are widely distributed, while Cenozoic rocks are narrowly distributed. Southwestern Japan is further divided by the Median Tectonic Line (the c line) into an inner belt (on the Sea of Japan side) and an outer belt (on the Pacific Ocean side). The following conspicuous zonal structures parallel to the Median Tectonic Line are distributed in the outer belt of southwestern Japan: the Sanbagawa belt (a high-pressure metamorphic belt) (6), the Chichibu belt (a Permian-Triassic accretionary prism: hereafter written as AP) (7), and the Shimanto belt (a Cretaceous-Neogene AP) (8). The Shimanto belt is divided into three sub-belts by two tectonic lines that include the Honguu fault, which are not shown in the figure. The inner belt includes the Ryoke belt (5) (a high-temperature metamorphic belt) and the Mino belt (mainly Jurassic AP) (4). The Mino Belt has several active faults, including the Atera Fault. The eastern part of the Sorachi-Ezo belt (10), which runs north-south through central Hokkaido, is a geological division different from that of the main island of Japan. To the east of the Sorachi-Ezo Belt, the Tokoro Belt (12) (a Cretaceous-Paleogene AP) and the Nemuro Belt (13) (Cretaceous-Cenozoic sedimentary rocks of the Kuril Arc) are distributed. The boundary between the Tokoro belt and the Nemuro belt is the Abashiri Tectonic Line (the e line).

3.2.2 Yamaguchi village area-Atera active faults

The study area is Yamaguchi Village, which is now west part of Nakatsugawa City, Gifu Prefecture, which is located at the southeastern end of the Atera main active fault (Figure 9). The Atera fault can be traced for about 67 km. Yamaguchi village is located about 340 km west of Tokyo. The Kiso River flows east-west through the northern edge of the study area and north-south through its west side (Figure 10). The Atera fault splits the study area into northeastern and southwestern parts. The Atera fault is the boundary between mountainous area (Mt. Takadoki, with an altitude of 1038 m) and a plateau area (Mt. Bonten, 696 m) in the Mino belt. This geography is caused by a large-scale left slip of the fault so that the NE side of the fault is uplifted relative to the SW side.

Figure 10.

Topographic map (a) and geological map (b) of the Yamaguchi village area (edited from Yamada [39]) of the study area. Topographic map (a) is displayed with a red relief image map overlaid on the geospatial information Authority of Japan (GSI) tiles. The red relief image map [40] is a topographic map that emphasizes ridges (white) and valleys (red) created by system for automated geoscientific analyses GIS (SAGA-GIS) (https://saga-gis.sourceforge.io/en/index.html). The black frame indicates the man-borne survey area. The blue curves indicate survey route of the car-borne survey. The geological map legends are as follows; 1: Alluvium or talus deposit, 2: Terrace deposit or old alluvial fan deposit, 3: Seto group, 4: Naegi-Agematu granite, 5: Inagawa granite, 6: Granodiorite, 7: Non-classified Nouhi rhyolites, 8: Fujimi-dai welded tuff, 9: Fault, and 10: Strongly radioactive zone by contact metamorphism.

The geology of the study area is mainly underlaid by late Cretaceous to Paleogene Nohi rhyolite and granite. The granite caused contact metamorphism during intrusion. The late Miocene to Pliocene Seto Group covers them and is distributed in the Kiso River tributaries. Quaternary sediments are only distributed on a small scale along the Kiso River (Figure 10(b) [39]).

The Atera fault is a fracture zone that is composed of parallel faults or echelon faults in a Northwest-Southeast (NW) direction. Faults that are diagonal to the Atera fault are the Magome-Pass fault and its parallel faults. These faults pass from Agatuma Pass to Magome Pass and are cut by the Atera fault [39].

3.2.3 Memanbetsu area-Lake Abashiri east coast active faults

The study area is located in Memanbetsu Town, which is now part of Ōzora Town, in eastern Hokkaido, where active faults run north-south in the eastern coast of Lake Abashiri. The distance from Sapporo to Ōzora Town is 257 km. According to the Active Fault Society [38], the Lake Abashiri east coast active fault group comprises three faults that run parallel to each other in a north-south direction on the east coast of Lake Abashiri (Figure 11). The Notori active fault, located on the eastern side of the group, was chosen as the target fault. The survey area has a hilly topography at the northern foot of Mt. Mokoto (with an altitude of 1000 m), which is part of the outer rim of Lake Kussharo, Japan’s largest caldera lake. The altitude of the surrounding area is less than 200 m. The hills are eroded by the Memanbetsu River, which flows north-south and empties into Lake Abashiri (Figure 11(a)).

Figure 11.

Topographic map (a) and geological map (b) (edited from 1:200,000 geological map from the geological survey of Japan) of the Memanbetsu area. The dark blue lines show the locations of faults are according to ref. [38]. The black frame indicates the man-borne survey area. The red curve indicates the survey lines.

The fault group is part of the Abashiri Tectonic Line, which marks the boundary between the Nemuro and Tokoro belts (Figure 9). The Nikoro Group, consisting mainly of basaltic volcaniclastic sedimentary rocks and other formations, belongs to the Cretaceous-Paleogene accretionary complex and is distributed in the Tokoro Belt. On the other hand, the Nemuro Group, composed of Cretaceous-Paleogene sandstone, mudstone, and conglomerate, is distributed in the Nemuro Belt.

These strata are overlain by the Abashiri Formation (Miocene) and the Kusharo Pumice Flow Deposit (Quaternary) (Figure 11(b)), among other formations. The Abashiri Formation consists of basaltic andesite and dacite lavas, pyroclastic rocks, conglomerates, sandstones, mudstones, and coeval intrusive rocks. The Kusharo pumice flow deposit consists of dacite pumice and volcanic ash. The study area is covered by the Kusharo pumice flow deposits with an average thickness of more than 10 m. Therefore, the background radionuclide concentrations in GRS are entirely derived from the radionuclides in the Kusharo pumice flow deposits.

3.2.4 Kii-Tanabe area-Honguu active fault-

The survey area is located in the southeastern part of Tanabe City, approximately 110 km southeast of Osaka City (Figure 9). The survey area belongs to the Shimanto belt. The Honguu fault is an active fault that runs in an east-northeast direction [38]. The topography on the north side of the fault is an east-west mountain range connecting Mt. Takao (606 m) and Mt. Maki (796 m). The topography on the south side of the fault is a mountainous area oriented north-northeast, including Mt. Kinugasa (235 m) (Figure 12(a)). The Right Aizu River and Left Aizu River, which flow north-south and intersect the Honguu Fault, originate in the Hatenashi Mountains (the highest point, 1262 m) in the north, join near Kii-Tanabe, and flow into the Pacific Ocean.

Figure 12.

Topographic (a) and geological (b) maps of the Kii-Tanabe area. The fault of the sky-blue line and geological map (b) are according to ref. [41]. The square frame indicates the survey area. The black curve shows the car-borne survey stop measurement route.

The Shimanto belt, which runs from almost east to west, is an accretionary prism formed from the Cretaceous to the middle Tertiary (Figure 9). It mainly consists of alternating sandstone and mudstone layers, accompanied by greenstones and chert. The boundary between the Chichibu Belt in the north and the Shimanto Belt in the south is the Butuzou tectonic line. The Shimanto belt is divided into three sub-belts by two tectonic lines. The Gobo-Hagi tectonic line separates the Hidakagawa belt (Cretaceous) in the north from the Otonashigawa and Muro belts (Paleogene) in the south. Furthermore, the Otonashigawa belt and the Muro belt are separated by the Honguu fault (Figure 12(b)). The development of the Quaternary system is extremely poor, and only terrace sedimentary layers and alluvium are found in a very narrow area along the river [41].

The Honguu fault divides the study area into northern and southern parts (Figure 12(b)). The northern part belongs to the Otonashigawa belt. The southern part belongs to the Muro belt. The fault runs east-west, dips 20–50 degrees to the north, and is accompanied by a fault clay zone several tens of centimeters wide. The Otonashigawa Group of Eocene is distributed in the Otonashigawa belt, and the Muro Group of Oligocene-lower Miocene is distributed in the Muro belt. The Tanabe Group of middle Miocene covers the Muro Group with unconformity (Figure 12(b); [41]). The Otonashigawa Group consists of alternating sandstone and mudstone layers. The Muro Group is composed of sandstone, mudstone, conglomerate, and alternations of sandstone and mudstone. The Tanabe Group consists of an alternation of sandstone and mudstone.

3.2.5 Southern Abukuma area-Tanagura fracture zone and Abukuma belt

The study area is located in the Kuji River lowland area and the southern Abukuma Mountains, covering Hanawa Town and Yamatsuri Town, Fukushima Prefecture. Hanawa Town is about 200 km north-northeast of Tokyo (Figure 9). The topography of the study area consists of the Kuji River Lowland and the Southern Abukuma Mountains. The Kuji River lowland in the western part of the survey area belongs to the Ashio (Yamizo) belt, and the eastern part belongs to the Abukuma belt (Figure 9). Both belts are separated by the Tanagura fracture zone, which is 2–5 km wide (Figure 13).

Figure 13.

Topography (a) and geological map (b) of the Southern Abukuma area. The geological map is modified from ref. [42] and Koshiya et al. [43]. Red lines on the topographic map (a) indicate faults from the geological map (b). The black square frame indicates the investigation area of the Tanagura EM fault by man-borne survey. The dark blue curves indicate the investigation route of the car-borne survey. Abbreviations for geological map (b) are as follows: HG: Higashidate, Yujimata, HD: Hidoro, KY: Koya-yachi, MR: Morinoshita, TN: Tonohata, OW: Oiwake, HS: Hashiba, SN: Sankomuroyama. Geological legends for geological map (b) are as follows: 1: Quaternary system, 2: Tertiary system, 3: Higashidate rock body (Mylonitic granodiorite), 4: Onukari rock body (Mylonitic adamellite), 5: Mylonitic and cataclastic sedimentary rocks of the Ashio belt, 6: Cataclastic amphibolite, 7: Mylonitic and cataclastic metamorphic rocks of the Abukuma belt, 8: Fine-grained granular granodiorite, 9: Mylonitic granodiorite, 10: Large sphene-bearing granodiorite, 11: Porphyritic granodiorite, 12: Foliated tonalite and granodiorite, 13: Fine-grained quartz diorite and tonalite, and 14: Metamorphic rocks of the Abukuma belt.

The Abukuma mountain range is gently undulating, and the elevation gradually decreases from east to west, except for the highlands of the watershed with an altitude of about 890–870 m, which is about 10 km east of the study area. Near the Tanagura east marginal fault of the Tanagura fracture zone, the elevation suddenly drops from 550 m to 300 m (Figure 13(a)).

The Ashio belt is underlain by sedimentary rocks of the Paleozoic and granitic rocks. These are unconformably overlain by Tertiary strata. In the southwestern part of the study area, granitic rocks and Tertiary alternating layers of sandstone and mudstone are distributed. The Takenuki metamorphic rocks and the Hanawa plutonic rock mass of the Late Cretaceous are distributed in the Abukuma belt of the survey area. Takenuki metamorphic rocks are high-grade metamorphic rocks of high-temperature type. Hanawa rock mass consists of quartz diorite, tonalite, granodiorite, mylonite granodiorite, and fine-grained diorite [44]. The Hanawa rock mass intruded into metamorphic rocks and sedimentary rocks of unknown age and caused contact metamorphism to them.

The Tanagura fracture zone is a fault fracture zone bounded by two NNW-trending faults: the Tanagura West Margin Fault and the Tanagura East Marginal Fault (hereafter, Tanagura EM Fault). The following fractured rocks are distributed between these faults: mylonite and fault gouge, which originated from sedimentary rocks, granites, metamorphic rocks, etc. The sedimentary rocks are thought to have originated from the Ashio belt, while the metamorphic rocks are thought to belong to the Takenuki metamorphic rocks of the Abukuma belt [42]. Granitoids are distributed as several small rock bodies. Higashidate and Ohonukari rock bodies are distributed in the study area. The Higashidate rock body is greenish granodiorite. The Ohonukari rock body is gray-white biotite adamellite [42].

Since the Tanagura EM marginal fault can be clearly identified topographically, and the underground geology is known by electrical prospecting [42], it was selected as a target fault for GRS. The geological structure in the vicinity of the survey area has been clarified by many researchers including in Ref. [42] and Koshiya et al. [43]. Here, only the results of photolineament surveys are described. Photolineament interpretation can be seen in various orientations such as NNW, NW, EW, NE, and NNE directions, but the NNW or NW direction is the most prominent (See Figures 26 and 27 in the Section 4.2). The NNW-oriented lineaments corresponding to the Tanagura EM fault were clearly recognized inside the Tanagura fracture zone area. However, the number of lineaments corresponding to boundary faults of small rock bodies inside the Tanagura fracture zone is smaller than that of the Abukuma zone. The reason for this observation is thought to be that the erosion of the topography in the Tanagura fracture zone made it difficult to interpret the lineament.

3.3 Investigation results

3.3.1 The Atera fault

Man-borne survey, soil radon gas survey, and dipole/dipole array electric prospecting were conducted along the A survey line (extension 660 m), as shown in Figure 14(a) [17]. The measurement interval was 20 m. The number of measurement points was 33. According to Yamada [39], survey line A crosses Atera faults A and B. The A fault passes near point 17, and the B fault passes near point 22.

Figure 14.

Location map of survey points and outcrop sketches (a) and fluctuation patterns of K, Bi, Tl, Bi/Tl, and radon concentrations (b) along a survey line across the Atera faults of the Yamaguchi village area. The location of the Atera fault is based on Yamada [39]. The red % number indicates the increasing rate R, and the black number indicates the survey point number. See text for green and red triangles.

Fresh granite (C_H bedrock) was exposed at survey points 1–2, 7–8, and 30–33 (Figure 14(a)). At points 20–26, weathered and sandy C_L bedrock was exposed. Survey point 18 exposed a D bedrock of pale green fault gouge. On the east side of the Atera B fault, springs of about 1 to 2 liters per minute are distributed in several places (Figure 14(a)). Since no springs are distributed on the west side of the fault, it is thought that the fault gouge functions as a water barrier. Based on the above observations, the area between Atera faults A and B is the most fractured.

Figure 14(b) shows the fluctuation patterns of K, Bi, Tl, Bi/Tl, and radon concentrations along the A line. The Atera faults are indicated by arrows in the figure. The largest Bi/Tl peak is at point 18, 20 m east of fault A. The increased rate R of Bi/Tl at point 18 is 48%. This anomaly is composed of the local maxima of Bi (red triangles) and the local minima of Tl (K) (green triangles). Note that there is a fault gouge near point 18.

Since Tl is the least mobile of the three nuclides, the fluctuation pattern of Tl indicates element concentration variation in the original rocks (Section 2.2.4). There should be proportional relationships between Bi, Tl, and K in the original granite. Bi variations deviating from Tl variations were therefore caused by secondary uranium migration and enrichment. In other words, the Bi/Tl ratio allows us to distinguish Bi anomalies into primary magma crystallization anomalies and secondary weathering anomalies occurring near faults. For example, at points 32 and 3, three nuclides, Bi, Tl, and K, are increasing at the same time, so the Bi/Tl variation does not form a peak. Fresh granite distribution is predicted at these locations. In fact, outcrop observations confirmed the distribution of fresh C_H rocks.

Radon concentrations ranged from 0.2 to 19.2 kBq/m³ with an average value of 3.8 kBq/m³. The variation pattern of radon concentration along the A line can be divided into two zones. One is the zone where the radon concentration near faults A and B forms several peaks (points 13–31, average 5.4 kBq/m³). The other is a zone with a low radon concentration and a flat pattern (points 1–12 and 32–33, average 1.2 kBq/m³) (Figure 14(b)).

The highest radon concentration in the former zone is at point 14 (19.2 kBq/m³), 60 m west of fault A. The third highest peak is at point 22 (10.7 kBq/m³) near the location of fault B. These peaks are thought to be due to long-range uplifted non-diffusive radon through open fissures associated with faults A and B. However, we need a measured value of a carrier gas such as CO₂ to determine whether it is non-diffusive radon, but there are no carrier gas data. Since the fault gouge is distributed under point 18, the small peak at station 18 is considered to be due to radium near the surface. Point 31 has the second highest peak at 12.2 kBq/m³. However, since it is a primary anomaly in which three nuclides are high at point 32 at the same time, there is a possibility that this radon is also diffusely transported radon due to radium near the surface.

Granite is distributed on the west side of fault A (points 1–16), and granodiorite porphyry is distributed on the east side of the fault (points 17–33) [39]. Comparing two sides of K (7263:6790), Bi (550: 511), and Tl (1264:1114) shows that the granite distribution area is higher. However, the difference is not large enough to define a fault.

Figure 15(a) shows the variation patterns of Bi/K, Tl/Bi, radon concentrations, and increasing rate of Bi/Tl. Figure 15(b) is a simplified pseudo-resistivity profile [12]. The measured resistivity values ranged from 45 to 11,264 Ωm. Since the resistivity of general granite is generally 3 × 10² to 1 × 10⁴ Ωm, various grades of rocks from fault gouge to fresh granite should be distributed on the resistivity profile. The resistivity values are classified into three categories for ease of viewing the resistivity profile in Figure 15(b). Although it is difficult to accurately identify the lithology from the resistivity value, the relationship between the resistivity range and the lithology is as follows, considering the general relationship between the resistivity value and the lithology and field observations: 0–120 Ω·m is from fault gouge to sandy granite; 120–200 Ω·m are sandy to brecciated granites; 200–300 Ωm is granite with cracks; and 300–500 Ωm is sheared granite. Above 500 Ωm is fresh granite [12].

Figure 15.

Fluctuation patterns of Bi/K, Tl/K, radon concentrations and Bi/Tl increasing ratio (a) and the simplified resistivity profile (b) across the Atera faults of the Yamaguchi village area. The numbers on the increasing ratio of Bi/Tl curve indicate the increasing ratio at each point. The resistivity profile has been edited and modified from Imaizumi et al. [12]. The red double-headed arrows in graph (a) indicate the locations where uranium was deposited in the cracks.

Since the Atera fault in the study area is not a boundary fault between two stratums with significantly different resistivity values, it is difficult to draw a fault line in the resistivity profile. The subsurface structure of the Atera fault is inferred from the fault location of geological map and resistivity distribution pattern under the ground surface. Underground traces of fault A and fault B are considered to be shown in the resistivity profile as lines or bands between rock masses of about 300 Ωm or more that have not been broken by fault. Probable fault lines of fault A and fault B are shown in Figure 15(b). According to this rule, unknown faults a and b are also estimated.

The Bi/K and Tl/K curves using SGR logging crack detection methods (Section 2.2.5.3) and resistivity profiles showing subsurface structures are compared in Figure 15(a) and (b). The opening positions in both curves (indicated by the red arrows in the figure) show the locations where uranium salts have precipitated in the cracks. At point 18, which has the highest increasing ratio, the opening between the Bi/K and Tl/K curves is the largest (the longest red arrow). Other opening locations (22 and 31% increase) may be related to inferred faults a and b. These faults are “closed fissures” because point 18 is the location of the fault gouge. On the other hand, the radon gas concentration peaks indicating open cracks are located between these closed crack faults.

3.3.2 The Abashiri Lake east coast fault

Man-borne survey, soil gas radon, and CO₂ surveys were carried out along two survey lines, the first line (length 940 m) and second line (820 m), which are set up across the Notori fault (Figure 11). The measurement interval was 20 m. Here, fluctuation patterns of K, Bi, Tl, Bi/Tl, radon, and CO₂ along the first survey line are shown (Figure 16). For the estimation of fault line (Figure 17), the data of the second survey line are also referred. In Figure 16, arrows indicate the location of the Notori active fault according to Ref. [38]. The Notori Fault is thought to pass through survey point 25 on the first survey line (Figure 16(a)).

Figure 16.

Fluctuations of K, Bi, Tl, Bi/Tl, and radon concentrations (a) and fluctuations of CO₂ and radon concentrations along the first survey line across the Notori fault of the Memanbetsu area.

Figure 17.

Relationship between presumed faults and the active faults of the Memanbetsu area.

The nearest Bi/Tl anomaly to survey point 25 is point 26, 20 m east of the fault, with an increasing rate R of 15%. This anomaly is composed of a small local maximum of Bi (red triangle) and a decreasing trend of Tl. Other anomalies are points 2 (R = 19%) and 14 (R = 21%), which are more than 200 m west of the active fault. These anomalies also consist of small local maxima in Bi (red triangles) and local minima in Tl (green triangles). These two anomalies can be faults because they have the same patterns.

Radon concentrations in soil range from 0.1 to 4.8 kBq/ m³, with an average value of 1.4 kBq/m³. Several peaks are formed near station 25 where the active fault passes (Figure 16(a)). These radon peaks and CO₂ concentration peaks tend to coincide (Figure 16(b)). Therefore, these peaks of radon are non-diffusive radon that has been transported through cracks from deep faults by the carrier gas of CO₂.

The first survey line (as well as the second survey line) has another index indicating the fault. This index is a stepped change in nuclide or nuclide ratio that indicates different material layer boundaries (stratum boundaries). The fluctuation patterns of Bi, K, and Bi/Tl show a sharp decrease at points 29–30 (indicated by green dotted line). Averages of K, Bi, and Bi/Tl at points 0–29 and at points 30–47 are as follows: 3563:2482 on K, 315:235 on Bi, and 0.56:0.46 on Bi/Tl (Figure 16(a)). Therefore, the fault shown by the stepped change index passes through point 30. Although the survey site is covered with the Kussharo pumice layer with an average thickness of 10 m, the stepped change in Bi/Tl may indicate the boundary fault of the different Kussharo pumice layers, which have different nuclide concentrations. Since the Kussharo pumice layer sandwiches a thin sand layer, the different nuclide concentrations may be due to the different depths of the sand layer.

Figure 17 shows the relationship between the inferred faults and active faults along the first and second survey lines. The two inferred faults are the Bi/Tl anomaly points linking fault (black line) and the Bi/Tl stepped fluctuation of turning points linking fault (brown line). Both the faults estimated by the two methods are shifted 20–100 m to the west from the active fault. The non-diffusive radon peak linking faults are distributed as several crack zones 100–250 m apart to the east and west of the fault.

3.3.3 The Honguu fault

A two-stage survey, reconnaissance and detailed survey, was conducted to locate the Honguu fault. In order to extract seven routes for detailed survey, the running car measurement of car-borne was carried out along 19 routes around the Honguu fault and a total survey distance of 64.8 km. For the detailed survey, the stopping car measurement of car-borne was performed (Figure 18). The total distance surveyed was 4930 m. This paper describes the results of the C5 line (Figure 19) that localized the Honguu fault using GRS and soil radon gas surveys.

Figure 18.

Detailed car-borne survey route and distribution of Bi/Tl anomalies of the Kii-Tanabe area.

Figure 19.

Fluctuations of K, Bi, Tl, Bi/Tl, and radon concentrations in soil (a) and fluctuations of radon and CO₂ concentrations in soil along the C5 survey line across the Honguu faults of the Kii-Tanabe area. See the text for an explanation of hatching patterns and green dotted line.

Figure 19(a) shows fluctuation patterns of K, Bi, Tl, Bi/Tl, and soil radon concentrations. According to the ref. [41] and topographical interpretation of aerial photographs, the Honguu fault is considered to pass through survey point 16. Point 16 has a Bi/Tl anomaly (increasing rate R = 11%). There are two other anomalous points, 100 m north of the fault (point 11, R = 11%) and 180 m north (point 7, R = 15%). All anomalous points are in the Otonashi Belt and not in the Muro Belt. The cause of this observation is unknown. These anomalies are all formed from Bi local minima (red triangles) and Tl local minima (blue triangles), which should indicate fault patterns.

Radon gas concentration ranges from 0.05 to 11.1 kBq/ m³ with an average of 1.7 kBq/ m³. There are several peaks around the Honguu fault. The non-diffusive radon peaks that overlap with the soil CO₂ peak (Figure 19(b)) have the highest concentration peak at point 15 (5.9 kBq/ m³), 20 m north of the fault. Other peaks include point 10 (2.2 kBq/ m³), 120 m north of the fault and point 5 (4.4 kBq/m³), 220 m north of the fault. These peaks are not related to the Honguu fault, but are related to the derived faults estimated from the Bi/Tl anomaly.

The Honguu fault is also the boundary fault between the Otonashigawa group and the Muro group, which have different sediment ages and nuclide concentrations. In fact, the mean values of K, Tl, and Bi/Tl at survey points 0–16 and 17–31 differed: 5229:5927 on K, 1295:1395 on Tl, and 0.33:0.30 on Bi/Tl, respectively. The stepped change in Bi is indicated by the green dot line.

Figure 18 shows the distribution of anomalies with a Bi/Tl increase rate of 5% or more. Due to the small threshold for fault detection on all survey lines, the anomaly points are distributed not only near the Honguu fault but also at locations far from the fault. Strict threshold settings are required to eliminate these extra (noise) anomalies. This issue is considered in Section 5.

3.3.4 The Tanagura east marginal fault

The Tohoku Agricultural Administration Office (hereafter, TAAO) evaluated bedrock groundwater development technologies, such as fault analysis, electrical surveys, and groundwater geochemical surveys, in rocky areas where groundwater development is difficult [42]. The southern Abukuma area, where various rock bodies are distributed, was selected as the test site. The author was in charge of the fault crack survey using radioactive prospecting. This section presents reviewed results of man-borne survey and soil radon gas surveys along the survey line across the Abukuma EM fault. The results of a regional running car-borne survey are described in the next section. TAAO [42] conducted an electrical survey of 900m survey line across the Tanagura EM Fault (Figure 20). The resistivity profile shows that the fault gouge and mylonite of granodiorite origin (resistivity 400 Ωm or less) are distributed on the west side of the fault (Tanagura fracture zone) and the mylonitized granodiorite (700 Ωm or more) is distributed on the east side of the fault (Abukuma zone). TAAO [42] concluded that the fault underlies under the survey line 530 m. Gamma ray spectrometry and radon gas survey were carried out along the electrical survey line of 450–600 m. The measurement interval was 10 m.

Figure 20.

Electric prospecting survey line and radioactivity prospecting survey point (a) and apparent resistivity profiles by the Wenner electrode array across the Tanagura east marginal fault (b) of the Southern Abukuma area. The resistivity profile from the TAAO [42] has been revised and added.

The fluctuation patterns of K, Bi, Tl, Bi/Tl, and radon concentrations along the survey line are shown in Figure 21. These patterns can correspond to the fluctuation patterns of the F dam site (Figure 7) where U leaching occurred in the fault fracture zone. Bi begins to decrease from both sides of the fracture zone (survey lines 500 and 590 m) and forms a minimum in the center part of the fracture zone. Comparing the variation patterns of K, Bi, and Tl, only Bi decreased (gray triangles). The minimum occurred at 550 m. This location is the center of the fracture zone indicated by GRS. The local maximum value of Bi (red triangle) caused by U enrichment is at 500 m at the entrance of the fracture zone, where Tl is a local minimum (blue triangle). Remember that the typical fracture zone shows fault clay layers form at both ends of the fracture zone (Section 2.2.5.1). U enrichment might have caused U to get adsorbed in these clay layers. As a result, Bi/Tl formed the only anomalous point (orange triangle) with an increasing rate of 11% along the survey line.

Figure 21.

Fluctuation patterns of K, Bi, Tl, Bi/Tl, and radon concentrations in soil along the electrical prospecting lines 450 to 600 m across the Tanagura east marginal fault of the southern Abukuma area. The position of the electric survey line is the number attached to the curve of Bi and radon. Red numbers are the increasing rate of Bi/Tl and radon concentration at the peak point, respectively. The Tanagura EM fault passes through the 530 m electric prospecting line.

Radon concentrations range from 0.4 to 2.5 kBq/ m³ with an average value of 0.9 kBq/ m³. The fluctuation pattern of radon concentration is also the same as that of F dam. There is a single peak of 2.5 kBq/m³ at line 520 m, 10 m east of the fault location. It is not possible to determine whether this peak is non-diffusive radon because no CO₂ measurements were made in this study, but the only peak is thought to be radon moving upward through an open crack associated with the fault.

The Tanagura EM Fault is a boundary fault between two geological zones with different ages and nuclide concentrations. Therefore, there should be a difference in nuclide concentration on both sides of the fault. In fact, Tl count increases stepwise around line 530 m. Figure 21 shows this change with the green dotted line. The Tl averages of 450–520 m and 530–600 m are 490 and 1560, respectively. Tl count increases threefold west of this fault. Since Tl is the most resistant element to weathering, this stepped change is considered to indicate that fault movements caused rocks of different compositions to come into contact.

4. Regional crack distribution survey by car-borne survey

4.1 Atera fault car-borne survey

In order to investigate the regional distribution of the Atera faults, the running car measurement of car-borne was carried out on 19 routes across the Atera fault [17]. Total survey distance is 50.1 km (Figure 10). Car-borne anomalies were defined by the Bi/Tl increasing rate R above the threshold. Car-borne survey recorded the integrated value for 30 s at 4 km/h running, so the measurement interval is 30–40 m. Anomaly locations were plotted at the center of the measurement interval. The reproducibility of the position of the anomaly was confirmed by the stopping car measurements at intervals of 10 m around several anomalies.

To determine the R threshold, we plotted anomalies with R = > 20, 20–15%, and 15–10% on a geological map and examined the relationship between the anomaly distribution and known fault locations (Figure 22). The number of Bi/Tl anomalies corresponding to faults is 4 for R= > 20%, 5 for R = 20–15%, and 16 for R = 15–10%. Of the R= > 20% anomalies near Magome Pass, only one corresponds to the Atera fault. If the threshold is set to 10% or more, the correlation between faults and anomalies will improve, but noise anomalies will increase. Since multiple anomalies are required for fault estimation, the Bi/Tl threshold for the study area was set at R > =10%.

Figure 22.

Relationship between the distribution of anomalies and the locations of the geological faults and active fault of the Yamaguchi village area.

The anomaly point is calculated from the ratio of the average value (called Bi/Tl background here) of five intervals (150–200 m) before and after the measured value. It is presumed that the wide-area distribution of the background also has fault information. Figure 23 shows the relationship between the Bi/Tl background, anomaly points, faults, and topography. The background ranged from 0.46 to 0.20. No association was observed between background values and the distribution of anomalies. Four high background areas with Bi/Tl > 3.6 are distributed: Northwestern part (1) and central part (2) of the survey area, northwestern slope of Mt. Takadoki (3), and near Magome Pass (4) (Figure 23).

Figure 23.

Relationship maps between the Bi/Tl background, anomaly points, faults and topography of the Yamaguchi village area: (a) Plan view, (b) 3D view. The Qgis2threejs plugin for three-dimensional (3D) display of quantum geographic information system (QGIS) projects the fault and Bi/Tl background distribution on the 3D terrain.

Area (1) is distributed in an elongated area parallel to the Atera fault. Area (2) has two peak points of Bi/Tl = 3.9. A subarea with one peak is concordant with the Atera fault. Another subarea with another peak is concordant with a secondary fault from the Atera main fault. The area near Magome Pass is distributed on the Magome-Pass fault that intersects the Atera fault. These regions are located in terrain erosion zones caused by faults and are thought to be related to faults (Figure 23(b)). While region (3) is distributed on high slopes away from faults, the formation mechanism of this region (3) is unknown. On the other hand, the low background area with Bi/Tl > 2.6 extends from the southern slope of Mt. Takadoki to the Atera fault lines. The relationship between this low Bi/Tl area and the fault is unknown. According to the weathering migration mechanism of U, the Bi/Tl background may also be related to U leaching and enrichment. The high background areas of (1) and (2) may be related to U enrichment along the Atera fault fracture zone.

Figure 24 shows the fluctuation patterns of the nuclides and nuclide ratios along the survey line in the vicinity of Magome [17]. The arrow shows the fault location from the geological map. The anomalies of Bi/Tl (R = 32%) and Bi/K (R = 26%) were formed by the local maximum of Bi and the local minima of Tl and K. Most of anomalies by the car-borne survey related to the known faults showed the same fluctuation patterns. It is worth noting that the fault-indicating nuclides and nuclide ratio patterns observed by man-borne survey also hold true for the 30 m of car-borne survey interval.

Figure 24.

Fluctuation patterns of the nuclides and nuclide ratios on the survey line around the Magome of the Yamaguchi village area (modified from Imaizumi et al. [17]). The arrow shows the fault location from the geological map [39].

The radioactive anomalies of Type I might be formed by contact metamorphism near the boundaries of granite bodies (Figure 2). Possible zones of Type I are the boundary between granodiorite and granite, and the boundary between rhyolite and granodiorite. These zones are indicated by dotted lines in Figure 10(b). In fact, there is an anomaly of R > = 20% on the lithofacies boundary at the southern foot of Mt. Sizumo in the northern part of the study area. In the southern part of Mt. Takadoki and the western part of Mt. Bonten, there are anomalies with R > =10% on the lithological boundaries (Figure 10(b)). These anomalies are thought to be caused by Type I. Figure 25 shows the fluctuation patterns of the nuclides and nuclide ratios on the survey line across the boundary of rock facies between granodiorite porphyry and granite around Mt. Sizumo [17]. The Bi/Tl (R = 41%) anomaly is formed only by an increase in Bi. This pattern differs from that of the fault.

Figure 25.

Fluctuation patterns of the nuclides and nuclide ratios on the survey line across the boundary of rock facies between granodiorite porphyry and granite around Mt. Sizumo of the Yamaguchi village area (modified from Imaizumi et al. [17]).

Figure 22 shows many anomaly points that cannot be explained by known faults or contact metamorphic boundaries. Some of these anomalies may represent hitherto unknown faults if they have the patterns of faults. Other anomaly points may be noise.

4.2 South Abukuma car-borne survey

Groundwater is expected to flow through the fracture system. In Tohoku Agricultural Administration Office project [42], the purpose of the car-borne survey was to verify its ability to detect fracture systems. In order to achieve the purpose, 30 routes and a total distance of 121 km of car-borne survey were carried out. The anomaly points of Bi/Tl as fracture signs were defined as R > =15%. The possibility of detecting faults by car-borne survey was evaluated based on the degree of agreement between the distribution of car-borne anomalies and the photolineaments [42]. In addition, the geological classification ability of car-borne survey was evaluated from the relationship between lithofacies boundaries and anomalous points and clay mineral estimation by K/Th plots.

4.2.1 Degree of agreement between the distribution of car-borne survey anomalies and the lineaments

Figure 26(a) shows the relationship between anomalous points and lineaments. The total number of anomaly points is 79. Forty-nine percent of all anomaly points are associated with lineaments. The rate of coincidence with lineaments is low (13%) within the Tanagura fracture zone because there were few lineaments within the Tanagura fracture zone (see Section 3.2.5). Considering only the Abukuma belt, 58% of the anomalies are associated with lineaments. Anomalies within the Tanagura fracture zone tend to coincide with boundaries of rock masses delimited by faults (See next section).

Figure 26.

Relationship between photolineament [42, 43] and car-borne survey anomalies (a) and relationship between the geological boundary [42] and car-borne survey anomalies (b) in the Southern Abukuma area.

Figure 27(a) shows the relationship between photolineaments (771 lines) [43] and lineaments crossing the car-borne survey routes (203 routes). The predominant directions of the photolineaments and the car-borne survey lineaments show almost the same direction. This figure shows that the car-borne survey was uniformly surveyed over the area.

Figure 27.

Frequency diagrams of the direction of photolineaments (modified from Koshiya et al. [43]) and photolineaments intersecting the car-borne survey lines (a) and frequency diagram of lineaments related to anomaly points (b) in the Southern Abukuma area. The frequency diagram of photolineaments shows the diagram of area 1 by Koshiya et al. [43]).

The directionality of the lineaments that coincide with the anomalous points was investigated by the following statistical method. This method was to classify the direction of the lineament every 10° and examine the number of cases that coincide with anomalous points. The concordance rate of lineaments every 10° that coincided with the anomalous point was investigated using the following equation.

$$\mathit{\text{Concordance rate}}=\frac{\mathit{\text{Number of coincidences between lineaments and anomaly points}}}{\mathit{\text{Number of intersections between lineaments and carborne routes}}}\times 100$$E3

Figure 27(b) shows the directional frequency distribution of lineaments with a high matching rate. The direction of lineament with high matching rate is N60°-50°W (50%), N30°-70°E direction (35%), N10°-20°W direction (33%), and N30°-40°W direction (32%). This trend is the same as the frequency distribution map of lineaments in Figure 27(a) [42].

4.2.2 Relationship between car-borne survey anomaly points and rock bodies

Many anomaly points within the Tanagura fracture zone tend to be plotted on the fault line (Figure 27(b)). On the other hand, in the Abukuma belt, unlike the Atera fault area (Section 4.1), with some exceptions, the anomalous points do not coincide with the lithofacies boundary. They are scattered in each rock body (Figure 27(b)). Since uranium enrichment in the contact metamorphic zone does not occur through the intrusion of granite into the high-temperature host rock deep underground (Figure 2), the granites of the Abukuma Belt may have intruded at a deeper depth than those of the Atera area.

4.2.3 Evaluation of rock weathering by K-Tl cross-plot

In order to plot GRS data on the K-Th cross-plot, a calibration pad as defined by the IAEA [4] is required. Unfortunately, due to the lack of calibration pads available in Japan, it is not possible to convert γ-rays K and Tl to potassium oxide (K2O) and Th concentrations by formal methods. Therefore, γ-rays K and Tl were converted into concentrations by the following simple method. Japan’s National Institute of Advanced Industrial Science and Technology publishes geochemical maps of all of Japan on the Internet (https://gbank.gsj.jp/geochemmap/gmaps/map.htm). According to this information, the study area is within the areas of K2O = 1.609–1.887% and Th = 11.60–14.74 ppm. Assuming that K2O = 1.609%, Th = 11.60 ppm, and the mean values of γ-rays K and Tl are linearly proportional to these values, γ-rays K and Tl were converted to each concentration datum.

Figure 28 shows the correlation of K₂O-Th for each rock body in the Tanagura fracture zone, except for the Hanawa rock body which belongs to the Abukuma belt. Table 2 shows regression equations and clay mineral species for each rock body using Table 1. According to the Th-/K cross-plot (Figure 3), clay minerals in each rock mass are plotted between the kaolinite line (K/Tl = 12) and the mixed layer clay line (K/Tl = 3.5). Therefore, the main clay minerals in all rock bodies are mixed layer clays. However, parts of the Neogene, Ashio sedimentary rocks, Metamorphic rocks, and Ohonukari rock bodies are plotted in the montmorillonite region. From Table 1, kaolinite and chlorite may be found in the Ashio belt origin granite (K/Tl = 9.8), and illite may be found in the Abukuma belt granite (K/Tl = 3.8). Summarizing the results from the figure and table, the most weathered rock body is the Ashio belt origin granite distributed in the Tanagura fracture zone. The least weathered rock is the Hanawa rock body of the Abukuma belt.

Figure 28.

The correlation of K₂O-Th between Neogene sedimentary rocks, Ashio granites, Ashio sedimentary rocks, Ohonukari rocks, Higashidate rocks, metamorphic rocks, and Abukuma Hanawa granites in the Southern Abukuma area.

Geologic province	Rock facies	Regression equation	R^2	Clay mineral
Tanagura zone	Ashio granite	y = 9.7654x-3.5928	0.93	Kaolinite and chlorite
	Neogene Sediment	y = 7.630 Sx- 0.4579	0.84	Mixed-layer clays: Montmorillonite
	Ashio sediment Rocks	y = 7.2368x-0.4683	0.53	Mixed-layer clays: Montmorillonite
	Metamorphic Rocks	y = 6252 5x + 2.0278	0.81	Mixed-layer clays: Montmorillonite
	Ohonurai Rock Mass	y = 5353 5x + 2.5684	0.60	Mixed-layer clays: Montmorillonite
	Higashidate Rock Mass	y = 4.587 7x + 4.1897	0.41	Mixed-layer clays
Abukuma Belt	Abukuma Hanawa Granite	y = 3.7866x + 5.8611	0.58	Mixed-layer clays: Illite

Table 2.

Regression equations of K₂O-Th and clay mineral species estimated for each rock body in the Southern Abukuma area.

5. Discussion

5.1 Characteristics of distribution patterns of nuclides around faults and fault detection indices

Due to the ease of exploration, the anomaly defined by the Bi/Tl increased rate and the radon gas concentration peak were selected as indicators for fault trace detection. We evaluated whether these indices can detect four existing buried faults. The results are summarized in Table 3. Figures in the table show the distance from the locations of the already-known fault. The three kinds of faults presumed by radioactivity prospecting are: (1) Fault presumed from the continuity of anomaly points with an increasing rate Bi/Tl above the threshold defined for each study area. (2) Fault estimated from the continuity of peak points of non-diffusive radon (more precisely, the radon has to coincide with CO₂ gas peak points). (3) Faults inferred from the continuity of conversion points where the average value changes in the stepped fluctuation pattern of nuclide and nuclide ratio. Point (3) was newly added based on field survey results.

Table 3.

Summary table of fault detection by radioactivity prospecting.

The index of (1) was near all four faults. In particular, it should be emphasized that the location of the Bi/Tl index was within 0–30 m of the known fault location. Anomalies of Bi/Tl are not due to an increase in Bi alone. They are caused by relative changes in Bi and Tl. All fault-related anomalies were formed by local maxima in Bi and local minima or decreasing trends of Tl. The geological meaning of the index (1) is considered to be related to a fault gouge, although only the Adela fault has been confirmed. The mechanism (1) is explained as follows: Uranium salts in groundwater were adsorbed on clay minerals (e.g., montmorillonite: see Sections 2.2.4 and 2.2.5) in the fault gouge. On the other hand, the concentration of Th in the fault gouge tends to preserve the composition of original rock (see Section 2.2.4). Given the above mechanism, we can understand the limitations of radioactivity prospecting. Bi/Tl anomalies cannot detect existing fault locations, such as when fault gouging is insufficient in fault fracture zones.

Index (2) was confirmed for all faults. The locations of the (2) index were within 10–180 m of the known fault location. In this paper, the CO₂ survey results are shown only for two faults, but radon concentration peaks that coincide with the CO₂ peaks were shown for these two faults. These prove the existence of non-diffusive radon. In a typical fracture zone structure, the open cracks develop around the central fault gouge (Figures 6 and 7). The peak position of (2) may indicate an open crack (open fault). Therefore, the peak position in the fracture zone does not match the Bi/Tl anomaly position located in the fault gouge. In the Notori fault of Abashiri Lake East Fault Group, several lines of the non-diffusive radon peak linking faults were distributed 100–250 m apart to the east and west of the fault (Figure 17).

The fault estimated by the index (3) was confirmed at three locations, except for the Atera fault. This phenomenon occurs when strata with different nuclide concentrations are in contact with each other at the fault boundary, so it is thought that it was not detected at the boundary between granodiorite and granite on the Atera fault. The positions of the turning points of the three faults were observed within 0–100 m from the known faults.

Since the Honguu fault and the Tanagura EM fault contact different rock bodies (stratums) with their faults, the fault detection by the index (3) is expected. However, for the Abashiri Lake Toho Fault (Notori Fault), the fault could be detected with this index, even though the Kusharo pumice flow deposits are distributed on both sides of the fault. This phenomenon was explained by the difference in the depth of thin layers (for example, sand layers) sandwiched between pumice flow deposits, but it is necessary to clarify the cause in the future.

The radioactivity fluctuation pattern of the fault at the F dam site, which has a fracture zone width of about 20 m, is considered to be the simplest fluctuation pattern. However, it is worth noting that fracture zones with a width of 50 m or more, such as the Tanagura electromagnetic fault, also have a similar deformation pattern to the fault at the F dam site (Figure 21). This phenomenon is the basis for detecting faults with the Bi/Tl index, even in running carbon with wide measurement intervals (Figure 24). However, as shown on the Notori fault (Figure 17), large shatter zones cannot be detected by single gamma-ray anomalies or radon anomalies alone. It should be kept in mind that in general, large fracture zones have a complex distribution of closed and open cracks.

In summary, it is possible to detect fault traces with three indices. However, each index has an error of 0–30 m, 0–180 m, and 0–100 m from the published fault position, respectively. In particular, index (2) may exist in multiple locations within the fracture zone, and even outside the fracture zone (Section 2.2.5.1).

5.2 Relationship between car-borne survey anomalies, faults, and lithofacies boundaries

Car-borne survey is an exploration technology that bridges the gap between man-borne and air-borne [4]. Even if car-borne survey is carried out in off-road areas, geological mapping using car-borne survey may not be able to investigate the original geological situation because the roads are heavily contaminated with materials used in road construction [4]. Therefore, the main purpose of a car-borne survey is recommended to be applied to environmental problems such as searching for lost radioactive sources and mapping radioactive fallout [4]. However, Kimura’s car-borne system (Section 3.1.2) was able to extend the limits of car-borne survey by giving direction to the measurement efficiency of the detector (Fault/lineament detection: Section 4.1, 4.2.1, geological boundary mapping: Section 4.2.2, weathering assessment: Section 4.2.3). Directivity to the measuring efficiency of the detector is given by installing the detector on one side of the cargo hold and shielding one side of the detector.

The distribution of abnormal points, which are a fault signature, changes depending on the threshold set arbitrarily by humans. Therefore, if the threshold value is set too small, noise anomalies other than faults and geological information will occur. Therefore, noise processing methods are one of the future challenges. Artificial Intelligence (AI) machine learning can be expected to become one of the processing methods.

Machine learning models such as random forests can draw decision boundaries for linearly inseparable distributed anomalous and normal points. Using random forest classification (OpenCV), which is a built-in function of SAGA-GIS, an attempt will be made to classify the car-borne survey anomalies in the Atera fault zone.

We already have location information of anomaly point distribution with R > =10% in GIS. In order to draw abnormal point areas in GIS, location information of normal areas without faults is required. For simplicity, the distribution of normal points without faults was created using the QGIS plugin “QChainage.” The QChainage can create points at specified intervals on the lines of the car-borne survey route on QGIS. Here, points were generated every 200 m on the car-borne survey routes (254 points). Among these points, points that overlapped with the anomaly points and points within 200 m of the anomaly points were deleted to obtain the location information of normal points (167 points) (Figure 29).

Figure 29.

Distribution map of anomaly and normal points in the Yamaguchi village area. The anomaly points were defined as R > =10%. The distribution of normal points was created using the QGIS plugin “QChainage.” see text for detailed explanation.

Training data for Random Forest classification (OpenCV) require raster data. The point data of abnormal points and normal points were converted to raster data using the IDW function (Inverse Distance Weighted). Figure 30 shows the classification results of anomaly and normal areas. The anomaly points area is indicated by a black polygon. In this classification, only the location information of abnormal points and normal points was used as a covariate, so only the areas of abnormal and normal points could be classified. Therefore, in this figure, anomaly areas related to faults (brown areas) and anomaly areas related to geological boundaries (purple areas) were manually colored. In the figure, there are still noise anomaly areas that cannot be classified. In the future, it is expected that fault areas will be automatically extracted using a machine learning model that also uses fault distribution maps and geological maps as covariates.

Figure 30.

Results of classification of anomaly and normal regions using the SAGA-GIS built-in function random Forest (CV) in the Yamaguchi village area. Anomaly areas related to faults (brown areas) and anomaly areas related to geological boundaries (purple areas) are manually colored.

6. Conclusions

Gamma-ray spectrometry is an excellent exploration method that can obtain information at a depth of 30 to 50 cm below the surface, which cannot be obtained by other remote sensing methods. The usefulness of this technology will be even greater if we can detect buried fault locations that cannot be directly observed due to recent sediments and soil. In order to clarify the possibility of fault detection by GRS, radioactivity prospecting including soil radon gas survey, etc., was carried out on already-known faults of four areas: active faults (the Adera fault, the Abashiri lake east coast fault group, and the Honguu fault) and non-active faults (Tanagura east marginal fault). As a result, it was shown that it is possible to detect fault traces with the following three indices: (1) Fault presumed from the continuity of anomaly points with an increasing rate Bi/Tl above the threshold defined for each study area. (2) Fault estimated from the continuity of peak points of non-diffusive radon. (3) Faults inferred from the continuity of conversion points where the average value changes in the stepped fluctuation pattern of nuclide and nuclide ratio. However, each index has an error of 0–30 m, 0–180 m, and 0–100 m from the published fault position, respectively. Published fault locations are compiled from literature maps and may have errors of up to 100 m or more depending on location. Therefore, it is concluded that the error of the fault position estimated by the three indices is within the allowable error range. Fault detection using radioactivity exploration should be based on the three indicators listed above. It is difficult to detect faults using only total gamma-ray measuring equipment such as survey meters, which has been done so far. In order to continue to develop analysis technology for radioactivity prospecting, in addition to accurate GRS and radon gas concentration measurements, it is also necessary to develop analysis technology using Artificial Intelligence (AI) technology, etc.