2024 Volume 58 Issue 4 Pages 155-168
Riverbed sediments from 19 Hayakawa River sites in Hakone Caldera Japan were analyzed for 37 elements, categorized into three groups by location: Group-1 (nine mainstream samples), Group-2 (five tributary samples with somma upstream), and Group-3 (five tributary samples with central volcanic cones upstream). Principal component analysis (PCA) showed the 3PCs explain approximately 77% of the total variance. The first principal component (PC1) is derived from the inverse correlation between Al and a group including Na, K, and rare earths elements (REE), likely due to their weathering, and their acidic alteration. Groups 1 and 2 sediments might be more influenced by this weathering or acidic alteration than Group-3. The PC2 was characterized by elements enriched in volcanic ash, such as Sc, V, Mn, Fe, and Cu, showing no significant score differences among the three groups. The PC3 was characterized by the correlation among Li, Al, Cu, Ga, Rb, and Sr, with no significant score variations among the three groups. Arsenic showed loadings below 0.5 in PC1 to PC3, indicating its limited contribution to the overall sample characteristics. However, a specific site with upstream volcanic hot spring discharge exhibited high As content, implying possible sediment contamination by As-bearing minerals like pyrite. This suggests the possibility of using As content in riverbed sediments to detect volcanic activity. This study’s results indicate that the elemental composition of riverbed sediments could be associated with the weathering and acidic alteration of volcanic rocks from Hakone Volcano, indicating the potential to identify areas influenced by geothermal activity.
Japan has many active volcanoes such as Mt. Hakone, Mt. Nasu, Mt. Motohirane, Mt. Aso, and Mt. Ontake (Fig. 1a). In this paper, “active”, “dormant”, and “extinct” are defined as follows, based on the descriptions from the United States Geological Survey (USGS). “Active” refers to volcanoes or volcanic regions that have erupted during the Holocene geological epoch (which started about 11,650 years ago) or have the potential to erupt in the future. “Dormant” refers to volcanoes that are not currently erupting but are considered “active” in terms of their potential to erupt in the future. “Extinct” refers to volcanoes that are considered unlikely to erupt again. However, since volcanoes once considered “extinct” can occasionally erupt, this definition is not absolute. Geothermal area refers to an area with the discharges of fumarolic gas, hot springs, and the presence of altered rocks. Volcanoes bring many blessings as geothermal and tourist resources, but they can also cause serious disasters due to eruptions. Furthermore, volcanic activity has physical, social, and economic impacts, as exemplified by the following cases. The following seven examples of volcanic activity are active volcanoes that have had significant physical, social, and economic impacts in Japan. As a physical impact, at Mt. Unzen, eruption activities resumed in November 1990, and the largest pyroclastic flow since the eruption began occurred on June 3, 1991. This resulted in 43 fatalities and missing persons. The eruption activity became prolonged, causing extensive damage to houses, roads, agricultural fields, etc., due to mudflows and pyroclastic flows (Central Disaster Management Council, 2007). As a social and economic impact, the eruption of Mount Ontake in September 2014 dealt a significant blow to the local tourism industry. Although the volcanic activity has calmed down, entry restrictions due to the danger of devastated trails and buildings have led to a decrease in tourists, which has been a major blow to the local industry. The majority of tourists to the Kiso region surrounding Mt. Ontake were domestic travelers, many of whom were repeat visitors. However, since the eruption of Mt. Ontake, there has been a decrease in domestic tourists (Japan National Tourism Organization, 2019). In 2014, a phreatic eruption at the Ontake Volcano, Japan, was claiming the lives of at least 58 hikers (Yamaoka et al., 2016). Therefore, predicting volcanic eruptions is crucial to reducing the damage caused by them. Prior to the 2014 eruption, the Ontake Volcano erupted in 1979, 1991, and 2007. Before the 1979 eruption, Ontake Volcano was dormant. Until 1979, the Ontake Volcano was recognized as an “Extinct Volcano”. The 1979 eruption was surprising for Japanese volcanologists. Following the 1979 eruption, fumarolic gases were strongly discharged near the summit. The 2014 eruption occurred in a geothermal area with a fumarolic gas discharge. Although the 2014 eruption at Ontake Volcano could not be forecast in advance, it was possible to envision the geothermal area near the summit as a potential location for the coming eruption.
(a) Location of some volcanoes on Honshu Island in the Japanese archipelago. (b) Sampling points (a to r and x). The red curve connects the peak of the somma of caldera. The Hayakawa River (blue curve), a major river in Hakone, originates from Lake Ashi. It flows clockwise from the northern part of the caldera to the eastern part and empties into Sagami Bay. The green curves indicate the tributaries of the Hayakawa River. Nine samples (Group-1; g, k, l, m, n, o, p, q, and r) were collected from mainstream (red circle), five samples (Group-2; b, c, f, h, and x) were collected from tributary with somma upstream (red square), and five samples (Group-3; a, d, e, i, and j) were collected from tributary with central volcanic cones upstream (red triangle). The yellow circles represent the geothermal areas called Owakudani and Sounzan.
The 2018 phreatic eruption of the Motoshirane Volcano, Japan, was also a surprise for Japanese volcanologists. The eruption location was not anticipated (Kataoka et al., 2021; Terada et al., 2021). A ski slope was established near the eruption location. Several skiers were hit by the debris thrown during the eruption, and one skier died. Prior to the 2018 phreatic eruption at the Motoshirane Volcano, no geothermal area developed near the location of the eruption. Even if it is difficult to forecast the eruption timing, disaster prevention is possible with the prediction of a potential eruption location.
The 2018 phreatic eruption of the Ebinokogen-Iwoyama Volcano in the Kirishima volcanic area and the 2015 phreatic eruption of the Hakone Volcano occurred within the geothermal area (Mannen et al., 2018; Tajima et al., 2019; Ohba et al., 2019). In these cases, phreatic eruptions occur in geothermal areas, prohibiting people from approaching the geothermal areas and preventing human damage. Hakone Volcano is still a popular tourist destination and fumarolic activity is still taking place.
This study focuses on the riverbed sediments of the Hayakawa River flowing through the Hakone Caldera and its tributaries, measuring the content of 37 types of elements contained in these sands to elucidate their compositional characteristics. Specifically, by comparing the chemical composition of riverbed sands of rivers originating from the central cone and the outer rim, we analyze how the characteristics of geothermal areas affect the surrounding environment. In particular, it investigates how the elemental composition of the sediments is affected by variations in the composition of volcanic rocks, weathering, or acid alteration, and how these sediments can be utilized for monitoring volcanic activity. Descriptions regarding the Hakone volcano will be elaborated in the Subsection “Geological setting”.
PrincipleFumarolic gases and hot spring water are transported to the surface through underground channels from the hydrothermal reservoir. Therefore, geothermal areas are expected to be potential locations of phreatic eruptions because a phreatic eruption is the explosion of a hydrothermal reservoir, and the fluid channels hosting the hydrothermal reservoir could be the weak point on the crust (e.g., Ohba et al., 2007; Aizawa et al., 2022).
The area covered by altered rock and soil is considered a signature of fumarolic gas and hot-spring water discharge in the past and present. Generally, rocks and soils alter due to their contact with hot spring water and fumarolic gases (e.g., Nogami and Yoshida, 1993; Sanemasa et al., 1972; Sanemasa and Katsura, 1973). In dormant volcanoes, altered rocks and soils record the influence of past volcanic activity. Therefore, the localization of altered rocks and soil could contribute to the estimation of potential eruption locations in the future (e.g., John et al., 2008). The altered rocks and soil are eroded by precipitation; some of the eroded material is carried by surface water, contaminating the riverbed with altered rock- and soil-derived particles. Riverbed sediment represents the concentration of elements in the surface layer of the crust in the upstream region, which is the principle of geochemical mapping (e.g., Imai et al., 2004; Ohta et al., 2005). The geochemical map is a concentration distribution map of elements that shows the chemical composition of basement rocks on the crust surface. Geochemical maps have been used to discover local heavy metal anomalies on the surface and explore the surrounding ore deposits (Webb et al., 1978). If the signature of altered rocks is found in the riverbed sediment, the activity of fumarolic gas and hot spring water discharges that occurred in the past within the upstream area can be estimated. Alteration zones covered with vegetation can be localized even if the altered zone cannot be found using aerial photographs, which is an advantage of examining riverbed sediment.
In this study, the riverbed sediments from the Hayakawa River (Fig. 1b), which flows within the caldera floor of the Hakone Volcano, Japan, were collected and analyzed. Hakone Volcano is currently active, and fumarolic gas and hot spring water are discharged into the central cones of the caldera. A geochemical signature of volcanic activity is expected to be found in the sediment composition of the Hayakawa River.
Geological settingHakone is an active volcano located in the center of Honshu Island, Japan. The volcano has a caldera structure with several central cones (Fig. 1b). Several geothermal areas developed as foothills of the central cone. The history of caldera formation was first proposed by Kuno (1950) and then revised by Mannen (2008). Eruptive activity began 50 ka ago at the central cones and continued until 3 ka ago (Kobayashi et al., 1997). The phreatic eruption 3 ka ago at the central cone broke a section of its western flank. The debris from the eruption dammed the Hayakawa River, forming Lake Ashi (Fig. 1b). No historical eruption was recorded at the Hakone Volcano until a small phreatic eruption in 2015, which occurred in the geothermal area on the central cone. Owakudani and Sounzan, which are circled in yellow (Fig. 1b), are geothermal areas of Hakone Volcano, where fumarolic activity is taking place. The Hayakawa River (blue curve), a major river in Hakone, originates from Lake Ashi. It flows clockwise from the northern to the eastern part of the caldera, joining the Sukumo River, which flows through the southern part of the caldera at Yumoto, and the Hayakawa River tributary, which has somma and central volcanic cones upstream, as shown by the green curve, and then drains into Sagami Bay. The catchment area of the Hayakawa River is 107.4 km2 (Horiuchi and Kodera, 2020). According to Horiuchi and Kodera (2020), due to water rights issues, a weir is installed at the outlet of Lake Ashi to the Hayakawa River, and water is released into the Hayakawa River only when the water level of Lake Ashi rises due to heavy rain.
Fumarolic activity and hot-spring water discharge have been observed in the central cone of the caldera. No geothermal area has been identified in the somma of the caldera. Some tributaries of the Hayakawa River have headwaters at the somma of the caldera, whereas others have headwaters at the central cones. By comparing the composition of the riverbed sand with a water source in the central cone and somma, it is possible to investigate how the signature of geothermal area is reflected in the riverbed sediment. Descriptions of the effects of volcanic activity on riverbed sediments are summarized in Major (2003).
Geochemical studies of the Hakone volcano have been conducted from various perspectives. In particular, the whole-rock composition of the Hakone Volcano ejecta has been reported in detail (e.g., Kuno, 1972; Nagai and Takahashi, 2007; Takahashi et al., 2006). At Hakone Volcano, geological surveys of volcanic products from rocks (e.g., Nagai and Takahashi, 2007) and studies of volcanic gas and hot spring water (e.g., Oki and Hirano, 1970; Ohba et al., 2008) have been conducted. In particular, arsenic (maximum of 3.50 mg/L) has been detected in hot spring waters discharged from the central cones of the Hakone caldera (Awaya et al., 2002). Nagai and Takahashi (2007) conducted an analysis of the whole-rock chemical compositions of nearly 900 rock samples from Hakone Volcano. The results revealed that the outer rim of Hakone Volcano is composed of small to medium-sized stratovolcanoes and monogenetic volcanoes. The study detailed the history of volcanic activity and the evolution of whole-rock chemical compositions. Notably, it was discovered that both calc-alkaline (hypersthene) and tholeiitic (pigeonite) series rocks are present, with significant temporal variations in the K2O content of the rocks. The classification of rock series is based on the total Alkali-total FeO-MgO (AFM) triangulation by Kuno (1968). In addition, Kuno (1968) classified the rock series into the pigeonite series and the hypersthene series, corresponding to the tholeiitic series and the calc-alkaline series, respectively, based on the combination of lithic pyroxene. Ohba et al. (2008) presented an analysis of volcanic gases collected from the Owakudani geothermal area of Hakone Volcano after 2001, demonstrating a decrease in gas temperature coinciding with the occurrence of earthquake swarms. Furthermore, it was revealed that specific chemical compositions (such as the SO2/H2S ratio, HCl concentration, stable isotopes of H2O, C/S ratio, and CO2/H2O ratio) underwent changes. Awaya et al. (2002) compared the impact of hot spring water on river water in the Hakone and Yugawara regions by examining the arsenic load. The study confirms that the presence of arsenic in river water can be sufficiently explained by the influence of surrounding hot springs.
Riverbed sediments were collected from 19 sampling points, as shown in Fig. 1b. Ten sampling points (a, b, c, d, e, f, h, i, j, and x) were located along the tributaries of the Hayakawa River. The other nine sampling points (g, k, l, m, n, o, p, q, and r) were located in the mainstream of the Hayakawa River. Among the nine tributary sampling points, points x, b, c, f, and h have their water sources on the outer rim, and points a, d, e, i, and j have their water sources on the central cone. Sampling points d and e had characteristics different from those of the other sampling points. Upstream points d and e were the Sounzan and Owakudani geothermal areas, respectively, where fumaroles and hot spring water were discharged.
Nineteen riverbed sediment samples were collected from riverbanks. To investigate whether the collected samples represented the chemical composition of the river sediment at each collection point, multiple samples (1 to 10) were collected from the point x (Fig. 1b). The detailed sampling points at x are shown in Fig. 2.
Sampling points of riverbed sediment at the point x. A total of five points were sampled in quadrants of the stream width (horizontal: 6.53 m), and five points were sampled every 2 m along the stream flow from the center of the stream width. Red points indicate the samples with a positive PC2 score.
Sediment collection and pre-treatment were performed according to the method described by Imai et al. (2004). Approximately 1 kg of riverbed sediment was collected from the riverbanks using a shovel and transported to the laboratory. The sand particles with a size of 0.30–0.85 mm were extracted from the sediment using sieves. Generally, sand particles, including magnetite, have high density and tend to accumulate at specific locations on the riverbed. As the contamination of particles with magnetite has a considerable influence on the bulk composition of sediments, particles containing magnetite are usually removed from the samples used for geochemical maps (Imai et al., 2004). In this study, magnetic particles were removed by using a strong magnet. After the removal of the magnetic particles, the sample was ultrasonically cleaned with pure water. The washed samples were then air-dried and crushed using an agate mortar and pestle, with approximately 1 g of the sample being pulverized.
A portion of the powdered sample was decomposed into a solution using the following method (Uchida, 1986). The chemical reagents used for decomposition were based on Yajima and Fujimaki (2002). A 20–30 mg powdered sample was weighed on a chemical balance and moved to a Teflon beaker with an inner volume of 25 ml. In this container, 0.58 mL of 60% HNO3 solution, 0.25 mL of 60% HClO4 solution, and 0.25 mL of 50% HF solution were added. The Teflon beaker was then sealed with a Teflon lid and inserted into a stainless-steel outer cylinder for pressure resistance. A Teflon beaker confined in a stainless-steel cylinder was placed in an oven at 150°C for 3 h. After heating for 3 h, the Teflon beaker was removed from the oven and cooled. To mask the unreacted HF in the reaction solution, the vessel was opened, and 3.75 mL of saturated boric acid water and 1.25 mL of pure water were added. The vessel was sealed again and heated at 150°C for 3 h. After the second heating, the Teflon beaker was opened, and the reaction solution containing the decomposed sample was transferred to a 25-ml plastic flask. Pure water was then added to the flask to obtain a total volume of 25 ml.
We analyzed all elements that could be measured with the analytical equipment we used. We measured the recovery rate of elements using standard rock samples JA-3. The JA-3 data were taken from Imai (1995). In this paper, we discuss elements whose recoveries fall within the range of 70–120% (ICP-MS) and 120–140% (MP-AES). As a result, we could not obtain sufficient recovery rates for some elements. For example, Ca is one such element. We excluded elements with poor recovery rates as analytical items, and finally selected 37 elements for evaluation. Elements with relatively high concentrations (Na, Mg, Al, Si, K, and Fe) in the prepared solutions were analyzed using a microwave plasma atomic emission spectrometer (MP-AES) (MP-AES 4210, Agilent Technologies, Inc.). The precision of MP-AES analysis was within approximately ±5%, with a detection limit of 0.01 ppm. XSTC-22 (SPEX CertiPrep Inc.) with an original concentration of 100 μg/mL was used as a standard solution to quantify the concentrations of these elements. MP-AES analysis was conducted at the Meteorological Research Institute of the Japan Meteorological Agency. Elements with relatively low concentrations (Li, Sc, V, Mn, Ni, Cu, Zn, Ga, As, Rb, Sr, Y, Cs, Ba, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Pb, Th, and U) were analyzed using inductively coupled plasma mass spectrometry (ICP-MS) (iCAP-Q, Thermo Fisher Scientific Inc.). The precision of ICP-MS analysis was within approximately ±5%, with a detection limit of less than 10 ppt. XSTC-13 (SPEX CertiPrep Inc.) with an original concentration of 10 μg/mL was used as a standard solution to quantify the concentrations of these elements. A manganese standard solution (FUJIFILM Wako Chemicals Inc.) with an original concentration of 1002 mg/L was used as standard solutions to quantify the concentrations of Mn. ICP-MS analysis was conducted at Tokai University.
In this study, data analysis was conducted using software called R for principal component analysis. The R software (R Core Team, 2018) can be downloaded for free from the CRAN (Comprehensive R Archive Network) website (http://cran.r-project.org/). The principal component analysis (PCA) was applied to 28 samples with 37 elemental concentrations listed in Table 1. PCA is a multivariate analysis method that is useful for identifying similar or characteristic variance among samples with large dimensions of analytical values. PCA enables the aggregation of information contained in multidimensional data, extraction of a small number of principal components that represent their characteristics, and interpretation of the information contained in multidimensional data. Eigenvalues, eigenvectors, and loadings play an important role in PCA analysis. Eigenvalues are a measure of the degree to which the principal components explain the variability of the data set. The higher the eigenvalue, the more variability in the data is captured by the principal components. Eigenvectors indicate the directionality of the transformation of the data into the new component space. Each eigenvector defines the direction of the principal components in a principal component analysis and points to the direction of maximum variance of the data. Loadings indicate the strength of the relationship between the original variables and the extracted components (principal components). It can be understood as the correlation coefficient between the original variable and each principal component, and indicates the degree to which the variable loads on the principal components. Geochemical data usually contain characteristic variance in elemental composition and a small number of groups found in a large number of samples (e.g., Iwamori et al., 2017). By identifying the characteristic elemental compositional variance in a large amount of geochemical data, it is possible to infer the causes and processes that gave rise to these variance (e.g., Anazawa and Yoshida, 1996). PCA can identify variance in multiple elemental compositions and can handle geochemical data from a large number of samples (e.g., Anazawa and Yoshida, 1996).
Elemental content of riverbed sediment (mg/kg)
| Li | Na | Mg | Al | Si | K | Sc | V | Mn | Fe | Ni | Cu | Zn | Ga | As | Rb | Sr | Y | Cs | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| a | 13 | 18200 | 11200 | 71700 | 317800 | 8400 | 12 | 124 | 398 | 29000 | 11 | 21 | 49 | 9.8 | 3.1 | 20 | 205 | 9.9 | 1.5 |
| b | 6.0 | 14800 | 20400 | 102000 | 268000 | 4700 | 10 | 96 | 463 | 31400 | 25 | 19 | 21 | 10 | — | 10 | 250 | 5.9 | 0.49 |
| c | 9.0 | 17800 | 9700 | 73000 | 296600 | 7800 | 8.4 | 82 | 320 | 23400 | 10 | 13 | 36 | 8.2 | 0.47 | 18 | 181 | 7.5 | 0.77 |
| d | 17 | 19500 | 12600 | 71300 | 306700 | 8300 | 9.4 | 84 | 359 | 25800 | 13 | 14 | 77 | 8.5 | 28.4 | 20 | 165 | 7.7 | 4.8 |
| e | 2.6 | 10200 | 15500 | 54900 | 302000 | 3500 | 19 | 174 | 454 | 37400 | 3.8 | 11 | 10 | 8.1 | 2.8 | 4.9 | 199 | 8.9 | 0.35 |
| f | 6.4 | 15900 | 10800 | 82400 | 304400 | 5000 | 12 | 114 | 311 | 27700 | 6.1 | 14 | 19 | 9.3 | 2.6 | 10 | 198 | 6.3 | 0.76 |
| g | 7.8 | 15800 | 16000 | 90200 | 272900 | 3900 | 12 | 103 | 413 | 33700 | 9.7 | 16 | 20 | 9.5 | 3.6 | 7.2 | 200 | 6.0 | 1.2 |
| h | 9.9 | 17700 | 15500 | 72800 | 298200 | 9800 | 10 | 103 | 456 | 33000 | 15 | 12 | 52 | 7.9 | 0.67 | 21 | 169 | 10 | 0.87 |
| i | 23 | 32800 | 12000 | 70900 | 396300 | 13400 | 9.1 | 82 | 391 | 27900 | 7.8 | 12 | 147 | 8.3 | 4.5 | 30 | 364 | 8.7 | 1.3 |
| j | 13 | 22500 | 19200 | 74700 | 344000 | 10300 | 9.5 | 77 | 418 | 32400 | 18 | 9.7 | 34 | 8.4 | 2.0 | 24 | 183 | 7.6 | 1.7 |
| k | 8.5 | 17000 | 5400 | 83300 | 331900 | 8000 | 6.0 | 54 | 209 | 19400 | 4.7 | 11 | 13 | 7.9 | 1.1 | 15 | 177 | 5.3 | 0.77 |
| l | 7.9 | 16200 | 6500 | 74200 | 319200 | 9200 | 7.0 | 65 | 205 | 22500 | 5.5 | 9.5 | 12 | 7.5 | 1.7 | 17 | 154 | 5.3 | 0.75 |
| m | 8.2 | 18000 | 13300 | 88500 | 369600 | 9100 | 9.6 | 85 | 292 | 26400 | 10 | 9.1 | 13 | 8.2 | 1.3 | 17 | 185 | 6.8 | 0.78 |
| n | 9.2 | 15500 | 11800 | 83100 | 285300 | 5300 | 20 | 130 | 473 | 31100 | 15 | 20 | 21 | 14 | 2.8 | 19 | 268 | 10 | 0.79 |
| o | 14 | 16600 | 4300 | 92600 | 274800 | 5700 | 10 | 94 | 271 | 18100 | 6.6 | 22 | 25 | 13 | 3.7 | 23 | 279 | 7.0 | 1.9 |
| p | 14 | 15500 | 6900 | 98900 | 258900 | 4100 | 15 | 101 | 372 | 24800 | 8.2 | 24 | 25 | 14 | 7.9 | 15 | 301 | 9.2 | 2.2 |
| q | 11 | 15200 | 8900 | 83500 | 240300 | 4100 | 15 | 120 | 423 | 23400 | 12 | 18 | 21 | 12 | 3.1 | 15 | 256 | 9.9 | 1.3 |
| r | 14 | 17700 | 6700 | 90600 | 280200 | 5300 | 13 | 111 | 372 | 24000 | 7.6 | 23 | 29 | 14 | 6.0 | 20 | 286 | 9.9 | 2.2 |
| Average | 11 | 17606 | 11483 | 81033 | 303728 | 6994 | 11 | 100 | 367 | 27300 | 11 | 15 | 35 | 10 | 4.5 | 17 | 223 | 7.9 | 1.3 |
| Standard Deviation | 4.5 | 4383 | 4499 | 11372 | 37892 | 2684 | 3.6 | 27 | 81 | 5038 | 5.2 | 4.9 | 32 | 2.4 | 6.3 | 5.9 | 56 | 1.7 | 1.0 |
| 1 | 10 | 12300 | 23100 | 92900 | 257900 | 4200 | 22 | 152 | 580 | 41500 | 35 | 37 | 35 | 13 | 2.4 | 17 | 259 | 6.3 | 0.66 |
| 2 | 14 | 12100 | 13400 | 102400 | 257500 | 4100 | 20 | 165 | 476 | 37700 | 18 | 47 | 34 | 15 | 3.6 | 16 | 288 | 8.2 | 0.73 |
| 3 | 14 | 10600 | 9100 | 105600 | 255800 | 3900 | 19 | 168 | 397 | 35200 | 10 | 45 | 33 | 14 | 4.6 | 15 | 255 | 5.3 | 0.71 |
| 4 | 11 | 12600 | 5600 | 108800 | 258800 | 3900 | 15 | 123 | 268 | 27300 | 7.3 | 34 | 25 | 14 | 3.3 | 15 | 298 | 5.9 | 0.64 |
| 5 | 12 | 12600 | 4400 | 112500 | 265000 | 4200 | 14 | 114 | 289 | 24900 | 6.1 | 35 | 27 | 14 | 3.2 | 16 | 309 | 5.7 | 0.70 |
| 6 | 11 | 13400 | 4500 | 102300 | 258300 | 4500 | 12 | 121 | 251 | 22100 | 6.3 | 28 | 22 | 13 | 2.5 | 17 | 288 | 5.6 | 0.67 |
| 7 | 11 | 13900 | 5600 | 116100 | 279900 | 4500 | 14 | 115 | 292 | 27100 | 7.1 | 34 | 25 | 14 | 3.4 | 17 | 304 | 6.3 | 0.71 |
| 8 | 9.1 | 12500 | 11000 | 97100 | 265700 | 4600 | 17 | 117 | 377 | 30700 | 12 | 27 | 26 | 12 | 2.5 | 18 | 248 | 6.5 | 0.67 |
| 9 | 11 | 18300 | 9100 | 107200 | 324300 | 7100 | 14 | 105 | 307 | 24200 | 8.5 | 16 | 21 | 12 | 1.3 | 27 | 297 | 7.3 | 0.87 |
| 10 | 13 | 16900 | 5400 | 98100 | 301700 | 7100 | 8.6 | 65 | 207 | 17700 | 5.7 | 12 | 16 | 11 | 0.38 | 27 | 248 | 5.5 | 0.87 |
| Average | 12 | 13520 | 9120 | 104300 | 272490 | 4810 | 15 | 125 | 344 | 28840 | 12 | 32 | 26 | 13 | 2.7 | 19 | 279 | 6.3 | 0.72 |
| Standard Deviation | 1.5 | 2218 | 5473 | 6817 | 21908 | 1167 | 3.9 | 29 | 108 | 7033 | 8.7 | 11 | 5.7 | 1.1 | 1.1 | 4.1 | 23 | 0.85 | 0.079 |
| Ba | La | Ce | Pr | Nd | Sm | Eu | Gd | Tb | Dy | Ho | Er | Tm | Yb | Lu | Pb | Th | U | ||
| a | 329 | 5.5 | 17 | 1.3 | 5.5 | 1.3 | 0.46 | 1.4 | 0.21 | 1.5 | 0.28 | 0.75 | 0.12 | 0.90 | 0.12 | 6.1 | 1.3 | 0.18 | |
| b | 154 | 1.9 | 6.8 | 0.70 | 3.3 | 0.82 | 0.32 | 0.89 | 0.11 | 0.98 | 0.16 | 0.58 | 0.058 | 0.53 | 0.065 | 1.3 | 0.36 | — | |
| c | 242 | 4.3 | 12 | 1.0 | 4.5 | 1.1 | 0.37 | 1.1 | 0.15 | 1.2 | 0.21 | 0.72 | 0.047 | 0.29 | 0.087 | 11 | 1.7 | 0.40 | |
| d | 267 | 4.9 | 14 | 1.1 | 4.7 | 1.1 | 0.37 | 1.1 | 0.15 | 1.2 | 0.21 | 0.76 | 0.084 | 0.73 | 0.094 | 5.1 | 1.3 | 0.16 | |
| e | 209 | 0.87 | 5.3 | 0.64 | 3.3 | 0.95 | 0.30 | 1.1 | 0.16 | 1.4 | 0.25 | 0.91 | 0.10 | 0.89 | 0.12 | 1.8 | — | — | |
| f | 213 | 2.0 | 6.5 | 0.69 | 3.3 | 0.81 | 0.33 | 0.90 | 0.12 | 1.0 | 0.18 | 0.64 | 0.068 | 0.62 | 0.078 | 5.4 | 0.25 | — | |
| g | 160 | 1.3 | 5.3 | 0.60 | 3.0 | 0.78 | 0.34 | 0.90 | 0.12 | 1.1 | 0.19 | 0.66 | 0.071 | 0.63 | 0.081 | 2.4 | 0.11 | — | |
| h | 275 | 6.0 | 17 | 1.4 | 5.9 | 1.4 | 0.41 | 1.5 | 0.19 | 1.6 | 0.27 | 0.96 | 0.11 | 0.92 | 0.12 | 4.4 | 1.5 | 0.22 | |
| i | 485 | 8.8 | 32 | 1.7 | 7.0 | 1.6 | 0.44 | 1.7 | 0.22 | 1.7 | 0.30 | 1.1 | 0.12 | 1.00 | 0.13 | 21 | 2.3 | 0.52 | |
| j | 326 | 6.5 | 18 | 1.4 | 5.7 | 1.3 | 0.40 | 1.3 | 0.17 | 1.3 | 0.23 | 0.82 | 0.089 | 0.77 | 0.10 | 5.5 | 1.5 | 0.20 | |
| k | 235 | 3.3 | 8.7 | 0.81 | 3.5 | 0.79 | 0.35 | 0.81 | 0.10 | 0.84 | 0.14 | 0.52 | 0.053 | 0.48 | 0.062 | 20 | 0.72 | — | |
| l | 298 | 5.8 | 14 | 1.1 | 4.5 | 0.96 | 0.35 | 0.95 | 0.12 | 0.74 | 0.16 | 0.58 | 0.060 | 0.55 | 0.070 | 2.7 | 1.6 | 0.029 | |
| m | 243 | 3.3 | 9.7 | 0.84 | 3.7 | 0.87 | 0.36 | 0.94 | 0.13 | 1.0 | 0.13 | 0.66 | 0.071 | 0.44 | 0.079 | 3.5 | 0.95 | — | |
| n | 233 | 6.2 | 15 | 1.3 | 5.3 | 1.2 | 0.42 | 1.2 | 0.17 | 1.3 | 0.24 | 0.78 | 0.10 | 0.75 | 0.11 | 2.8 | 1.1 | 0.19 | |
| o | 242 | 3.5 | 9.4 | 0.89 | 3.8 | 0.92 | 0.40 | 0.94 | 0.13 | 0.99 | 0.19 | 0.59 | 0.073 | 0.56 | 0.075 | 35 | 0.45 | 0.23 | |
| p | 206 | 2.7 | 7.4 | 0.78 | 3.6 | 0.93 | 0.41 | 1.0 | 0.15 | 1.1 | 0.21 | 0.69 | 0.083 | 0.64 | 0.089 | 10 | 0.05 | 0.15 | |
| q | 175 | 2.9 | 8.4 | 0.86 | 3.9 | 0.99 | 0.40 | 1.1 | 0.16 | 1.2 | 0.22 | 0.71 | 0.088 | 0.67 | 0.092 | 2.6 | 0.58 | 0.19 | |
| r | 220 | 3.7 | 9.2 | 0.98 | 4.4 | 1.1 | 0.43 | 1.2 | 0.18 | 1.3 | 0.25 | 0.81 | 0.10 | 0.59 | 0.11 | 3.8 | 0.19 | 0.20 | |
| Average | 251 | 4.1 | 12 | 1.0 | 4.4 | 1.0 | 0.38 | 1.1 | 0.15 | 1.2 | 0.21 | 0.74 | 0.083 | 0.66 | 0.093 | 8.1 | 0.94 | 0.22 | |
| Standard Deviation | 74 | 2.0 | 6.3 | 0.30 | 1.1 | 0.22 | 0.044 | 0.22 | 0.033 | 0.24 | 0.047 | 0.15 | 0.021 | 0.18 | 0.020 | 8.7 | 0.65 | 0.12 | |
| 1 | 215 | 2.9 | 8.2 | 0.84 | 3.7 | 0.96 | 0.33 | 1.0 | 0.15 | 1.1 | 0.21 | 0.54 | 0.083 | 0.64 | 0.090 | 2.1 | 0.30 | 0.17 | |
| 2 | 211 | 4.9 | 14 | 1.4 | 5.7 | 1.3 | 0.34 | 1.2 | 0.17 | 1.2 | 0.21 | 0.67 | 0.080 | 0.61 | 0.085 | 2.7 | 1.4 | 0.18 | |
| 3 | 215 | 3.3 | 8.5 | 0.81 | 3.5 | 0.86 | 0.29 | 0.89 | 0.12 | 0.95 | 0.18 | 0.57 | 0.070 | 0.54 | 0.074 | 2.1 | 0.39 | 0.16 | |
| 4 | 179 | 2.0 | 5.2 | 0.59 | 2.7 | 0.65 | 0.26 | 0.68 | 0.10 | 0.75 | 0.14 | 0.45 | 0.054 | 0.43 | 0.059 | 1.6 | 0.39 | 0.13 | |
| 5 | 205 | 2.2 | 7.1 | 0.64 | 2.9 | 0.68 | 0.26 | 0.69 | 0.10 | 0.55 | 0.14 | 0.44 | 0.052 | 0.41 | 0.057 | 1.6 | 0.15 | 0.11 | |
| 6 | 194 | 2.8 | 7.2 | 0.70 | 3.1 | 0.69 | 0.27 | 0.69 | 0.10 | 0.72 | 0.14 | 0.42 | 0.051 | 0.40 | 0.055 | 16 | 0.66 | 0.16 | |
| 7 | 196 | 2.5 | 7.0 | 0.68 | 3.0 | 0.71 | 0.26 | 0.73 | 0.10 | 0.78 | 0.15 | 0.47 | 0.056 | 0.44 | 0.061 | 2.1 | 0.57 | 0.14 | |
| 8 | 174 | 3.0 | 8.3 | 0.80 | 3.5 | 0.84 | 0.29 | 0.86 | 0.13 | 0.93 | 0.17 | 0.55 | 0.067 | 0.51 | 0.071 | 3.6 | 0.39 | 0.17 | |
| 9 | 241 | 5.9 | 14 | 1.2 | 4.6 | 0.99 | 0.32 | 0.93 | 0.13 | 0.94 | 0.17 | 0.55 | 0.066 | 0.51 | 0.069 | 3.5 | 1.6 | 0.24 | |
| 10 | 209 | 4.2 | 12 | 0.94 | 3.8 | 0.82 | 0.27 | 0.77 | 0.11 | 0.78 | 0.15 | 0.47 | 0.056 | 0.44 | 0.060 | 14 | 1.2 | 0.25 | |
| Average | 204 | 3.4 | 9.1 | 0.85 | 3.6 | 0.85 | 0.29 | 0.85 | 0.12 | 0.87 | 0.16 | 0.51 | 0.064 | 0.49 | 0.068 | 5.0 | 0.70 | 0.17 | |
| Standard Deviation | 18 | 1.2 | 2.9 | 0.23 | 0.88 | 0.19 | 0.029 | 0.17 | 0.022 | 0.18 | 0.027 | 0.074 | 0.011 | 0.081 | 0.011 | 5.2 | 0.48 | 0.042 |
The hyphen indicates a value less than the detection limit.
The elemental concentrations of the collected samples are listed in Table 1 and presented in Fig. 3. In Fig. 3, the elements are classified into three groups according to their content: Na, Mg, Al, Si, K, and Fe as high-content group; V, Mn, Sr, and Ba as intermediate-content group; and elements other than the above groups as low-content group. In Fig. 3, there are large differences in the dispersion of the content exhibited by each element: the dispersions of Si, Al, Fe, Ga, Sr, and Y are relatively small, whereas the dispersions of Li, Zn, As, Rb, Cs, Pb, Th, and U are relatively large. The dispersion of REE was relatively small, whereas medium dispersion was observed for La and Ce. REE are elements in the lanthanide series. These elements are numbered 57 to 71 on the periodic table and are classified as light rare earth elements (LREE) and heavy rare earth elements (HREE). In this paper, LREE in the lanthanide series is defined as up to La, Ce, Pr, Nd, Sm, and Eu, and HREE as up to Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu.
Content of each element in the sediment. The ordinary logarithm of the content (mg/kg) is shown on the vertical axis. The black bars indicate the elemental content of the sediment collected at each point (a to r). Red crosses indicate the elemental content of the sediment collected at the point x. The red dashed lines in the figure are classified by elemental content. Na, Mg, Al, Si, K, and Fe as high-content group; V, Mn, Sr, and Ba as intermediate-content group; and elements other than the above groups as low-content group.
The correlation coefficient between the elemental contents was obtained by the programming language “R” (http://cran.r-project.org/). In Fig. 4, the values in the cells highlighted in red, green, and blue have absolute correlation coefficients greater than 0.8, whereas those of 0.7 to 0.8 are highlighted in yellow.
Correlation coefficients among the content of 37 elements. The values in the cells highlighted in red, green, and blue have absolute correlation coefficients greater than 0.8, whereas those of 0.7 to 0.8 are highlighted in yellow.
Lanthanide elements show a strong positive correlation between themselves. Here, the number of lanthanide elements that show a correlation coefficient of 0.8 or more is defined as N. For example, Pr shows a correlation coefficient of 0.8 or more with La, Ce, Nd, Sm, and Gd, so N is 5 for Pr. The values of N are 3, 4, 5, 6, 10, and 4, respectively for La, Ce, Pr, Nd, Sm, Eu (LREE). The values of N are 10, 9, 7, 8, 7, 7, 3, and 9, respectively for Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu (HREE). There is a tendency for N values to be small for LREE relative to the values of HREE.
Principal componentsFor elements with high concentrations, such as Na, Mg, Al, Si, K, and Fe, the raw concentrations were input into the PCA calculation. For elements other than those mentioned above, the logarithm of the concentration with a base of 10 was input to the calculation of PCA analysis. However, for As, Th, and U, the raw concentrations were input into the calculation of PCA because there were samples with zero concentrations, namely, below the detection limit. When implementing PCA, the data was standardized. In R standardization the data were standardized so that the mean was 0 and the variance was 1.
Table 2 shows the eigenvalues of each principal component obtained from PCA and the contribution ratio of each component to the total amount of information. In the PCA of this study, with 28 independent variables, 28 principal components were obtained theoretically. Hereafter, the first, second, and 28th components are denoted as PC1, PC2, and PC28, respectively. As the contribution ratios of the sixth and subsequent components were very small, the contribution ratios up to PC5 and the cumulative contribution ratios are listed in Table 2. Table 2 shows that approximately 77% of the total variance can be explained by the sum of the contribution values of PC1, PC2, and PC3. Since the contribution ratio of PC4 decreased significantly relative to that of PC3, PC4 and the following principal components were not considered.
Contribution ratio and cumulative contribution ratio of the principal component to the data
| PC1 | PC2 | PC3 | PC4 | PC5 | |
|---|---|---|---|---|---|
| Eigenvalue | 16.2 | 7.51 | 4.81 | 2.31 | 1.66 |
| Contribution ratio (%) | 40 | 18 | 12 | 6 | 4 |
| Cumulative contribution ratio (%) | 40 | 58 | 70 | 75 | 79 |
The loadings of each element for PC1, PC2, and PC3 are listed in Table 3 and Figs. 5a, b, and c. The loadings of PC1 in Fig. 5a exhibit two notable features: the first feature is the large negative loading of Na, Si, and K, with a large positive loading of Al; and the second feature is the negative loading of all lanthanide elements. Furthermore, the absolute values of the loadings of the lanthanide elements decreased gradually from Gd to Yb. The third feature was the negative loadings of Y, Ba, and Th. The features found in PC1 correspond to the correlations indicated by the red cells in the correlation matrix shown in Fig. 4.
Loadings of PC1, PC2, and PC3 for 37 elements
| Li | Na | Mg | Al | Si | K | Sc | V | |
|---|---|---|---|---|---|---|---|---|
| PC1 | –0.36 | –0.75 | –0.36 | 0.68 | –0.58 | –0.74 | 0.2 | 0.17 |
| PC2 | –0.32 | –0.47 | 0.55 | –0.02 | –0.51 | –0.58 | 0.84 | 0.86 |
| PC3 | 0.81 | 0.03 | –0.35 | 0.63 | –0.35 | –0.15 | 0.36 | 0.26 |
| Mn | Fe | Ni | Cu | Zn | Ga | As | Rb | |
| PC1 | –0.4 | –0.15 | –0.3 | 0.42 | –0.65 | 0.42 | –0.21 | –0.44 |
| PC2 | 0.86 | 0.82 | 0.47 | 0.47 | 0.06 | 0.36 | 0.03 | –0.53 |
| PC3 | 0.05 | –0.11 | 0.22 | 0.72 | 0.48 | 0.78 | 0.15 | 0.63 |
| Sr | Y | Cs | Ba | La | Ce | Pr | Nd | |
| PC1 | 0.19 | –0.71 | –0.5 | –0.82 | –0.71 | –0.86 | –0.89 | –0.94 |
| PC2 | 0.18 | 0.41 | –0.19 | –0.37 | –0.41 | –0.33 | –0.17 | –0.07 |
| PC3 | 0.76 | 0.08 | 0.27 | 0.02 | 0.4 | 0.22 | 0.22 | 0.13 |
| Sm | Eu | Gd | Tb | Dy | Ho | Er | Tm | |
| PC1 | –0.96 | –0.83 | –0.95 | –0.9 | –0.85 | –0.84 | –0.87 | –0.78 |
| PC2 | 0.16 | 0.05 | 0.26 | 0.38 | 0.4 | 0.44 | 0.28 | 0.48 |
| PC3 | 0.06 | –0.06 | –0.08 | 0.04 | –0.16 | –0.01 | –0.28 | –0.03 |
| Yb | Lu | Pb | Th | U | ||||
| PC1 | –0.7 | –0.86 | –0.32 | –0.71 | –0.57 | |||
| PC2 | 0.43 | 0.41 | –0.61 | –0.46 | –0.19 | |||
| PC3 | –0.17 | –0.15 | 0.12 | 0.1 | 0.58 |
(a) Loadings for each element in the first principal component (PC1). (b) Loadings for each element in the second principal component (PC2). (c) Loadings for each element in the third principal component (PC3).
Takahashi et al. (2006) analyzed volcanic rocks from the Hakone Volcano to determine their elemental composition. They indicated that Na, K, and Y contents were positively correlated with the Si content. In contrast, Al content was negatively correlated with Si content. Although Takahashi et al. (2006) did not analyze the lanthanide elements because Y and lanthanides show similar characteristics, the lanthanide content was also expected to be positively correlated with Si in Hakone’s volcanic rock. These characteristics of the Hakone volcanic rock are consistent with the features of PC1 shown in Fig. 5a.
In general, with the weathering of rocks, Na and K contents in the rocks decrease and the relative Al content increases (e.g., Andrews et al., 2004). As shown in Fig. 5a, Al is positive, whereas Na and K are negative, which is consistent with the effects of rock weathering.
The negative correlations in the lanthanide series weaken from Gd to Yb (Fig. 5a). When rocks undergo acidic alteration, LREE tends to migrate toward the solution, whereas HREE tends to remain in the rock (Kikawada et al., 1995). Therefore, the characteristics of the REE shown in Fig. 5a are consistent with the behavior of the elements during acidic alteration of the rocks. Based on this discussion, PC1 appears to be influenced by both the intrinsic chemical composition of Hakone’s volcanic rock and/or the effects of weathering or acidic alteration of volcanic rock.
For PC2, the loadings for each element showed the following features (Fig. 5b): The first feature was the large positive loadings for Mg, Sc, V, Mn, and Fe, corresponding to the correlations indicated by the green cells in Fig. 4; and the second feature was the negative loadings for Si, K, Rb, and Pb, with notable negative loading of Pb owing to its lack of strong correlation with any particular element (Fig. 4).
According to Nanzyo et al. (2002), Sc, V, Mn, Fe, and Cu tend to remain concentrated in volcanic ash soil compared to natural soil. Furthermore, Fe forms insoluble salts, such as phosphates, and remains in volcanic ash soils (Nanzyo, 2018). Among the compositions that tend to remain concentrated in volcanic ash soil, Sc demonstrated a strong positive correlation with V and Cu, while Mn showed a strong positive correlation with Fe. Additionally, Sc and V each exhibited a weak positive correlation with Mn and Fe, respectively. The formation of insoluble salts could be attributed to the mechanism by which Sc, V, Mn, Fe, and Cu are concentrated in the volcanic ash. Sc, V, Mn, Fe, and Cu exhibited positive loadings in PC2. The positive loadings in PC2, seems to indicate the formation of insoluble salts.
The loadings of the elements for PC3 are shown in Fig. 5c. One feature of PC3 is the high positive loading for Li, Al, Cu, Ga, Rb, Sr, and U. This feature is produced by the high correlations between Li-Rb and Cu-Ga, shown in the blue cells in Fig. 4. Another feature of PC3 was that there were no elements with large negative loadings. In principal component analysis, the component that best represents the variation in the dataset (PC1) is found first, followed by the component (PC2) that is orthogonal to PC1 and best represents the remaining variation. On the other hand, PC3 represents the residuals after extracting PC1 and PC2 based on orthogonal coordinates. This indicates that PC3 carries less information compared to PC1 and PC2. Therefore, the lack of significant features in PC3 can be considered due to the limited additional information or insights it provides.
To evaluate the influence of PC1 and PC2 on each sample, Fig. 6 shows the scores of PC1 and PC2 for each sample and the eigenvectors of PC1 and PC2 for each element. The 19 riverbed sediments were classified into the following three groups based on the location of the river source where they were collected:
Biplot of the first principal component (PC1) and the second principal component (PC2). The upper horizontal and right vertical axes show the PC1 and PC2 scores for each location, respectively. The lower horizontal axis and left vertical axis show the PC1 and PC2 components of the eigenvectors for each element, respectively. “a” to “r” denotes the sampling point, and “1” to “10” denote the sample at the point x. The scores of the point x are the average scores of the points 1 to 10. The red arrows indicate the eigenvector of each element. The sampling points of Group-1, -2, and -3 are circled in yellow, green and blue, respectively.
Group-1: Nine samples (g, k, l, m, n, o, p, q, and r) were collected from the main stream of the Hayakawa River. These samples were circled in yellow (Fig. 6).
Group-2: Five samples (b, c, f, h, and x) were collected from tributaries with water sources in the somma. These samples were circled in green (Fig. 6).
Group-3: Five samples (a, d, e, i, and j) were collected from tributaries with water sources in the central cone. These samples were circled in blue (Fig. 6).
The mean and standard deviation of the PC1 scores for Groups 1, 2, and 3 were 0.0 and 1.1, 0.0 and 2.3, and –2.9 and 2.4, respectively. The average score for the individual ten points at x was 1.6, which was used for the calculation in Group-2. There was a clear difference between the Group-2 and Group-3 averages, suggesting either an intrinsic difference in composition between the sediments originating from the somma and the central cone or that the somma sediments have undergone more rock weathering or acidic alteration. The standard deviation of the PC1 score for Group-1 was 1.1, which was smaller than the standard deviations for Group-2 and Group-3, suggesting that the sediment composition in the mainstream was homogenized relative to Groups 2 and 3.
The mean and standard deviation of the PC2 scores for Groups 1, 2, and 3 were –0.4 and 2.5, 0.1 and 1.4, and 0.7 and 2.3, respectively. No statistical differences were found among the three groups with respect to the mean and standard deviation of the PC2 scores, suggesting that sampling location did not affect PC2. The mean and standard deviation of the PC3 scores for Groups 1, 2, and 3 were –0.8 and 2.9, –1.4 and 2.2, and –1.4 and 4.0, respectively. No statistical differences were found among the three groups with respect to the mean and standard deviation of the PC3 scores, suggesting that the sampling location did not affect PC3.
Local heterogeneity at the point xThere was a large discrepancy among the ten samples collected from the point x with respect to PC2 scores (Fig. 6). The scores for the points 1, 2, 3, and 8 were positive, whereas those for the other points were negative (Fig. 2). Positive scores were distributed mainly near the left bank of the river, whereas negative scores were distributed in the center of the river and on the right bank. This heterogeneity may be related to the influence of the sampling location on PC2, as described earlier. For the samples collected in this study, the positive and negative PC2 score components were mixed for each site, and the PC2 score at each site was determined according to the mixing ratio.
ArsenicAccording to PCA analysis, the absolute value of loading for As was less than 0.5 for all of PC1, PC2, and PC3, indicating that As does not contribute to the overall characteristics of the samples. Since As is associated with volcanic and geothermal activities, some volcanic hot springs and geothermal waters contain relatively high concentrations of As (Ballantyne and Moore, 1988; Webster and Nordstrom, 2003). Therefore, this study investigated whether anomalies existed in specific samples. As shown in Fig. 7, high concentrations of As were specifically detected in samples from the point d. Upstream of the point d, the Sounzan geothermal area is located, and volcanic hot spring water has been discharging. According to Awaya et al. (2002), As was detected in hot spring waters discharging within the Hakone caldera (average 0.27 mg/L), and especially high concentrations of As (maximum 3.50 mg/L) were detected in hot spring waters in the Sounzan geothermal area. The anomalous As concentrations detected at the point d suggest the influence of volcanic hot spring water. Arsenic is present as an impurity in pyrite (Ballantyne and Moore, 1988). The sediment at the point d may contain pyrite. Furthermore, the reason As concentration was not associated with PCA is that the PC axes are based on the criteria of maximum variance, so the exceptionally high As concentrations detected at the point d were not reflected in the PCA.
Arsenic content in the riverbed sediment (mg/kg) at the points a to r and 1 to 10.
The sediment deposited on the riverbed at 19 locations (points a to r and x) in the main and tributary streams of the Hayakawa River, which flows within the Hakone Caldera, was collected, and the contents of 37 elements (Li, Na, Mg, Al, Si, K, Sc, V, Mn, Fe, Ni, Cu, Zn, Ga, As, Rb, Sr, Y, Cs, Ba, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Pb, Th, and U) were measured. The collection sites were categorized as mainstream (Group-1), tributaries with upstream somma (Group-2), and tributaries with upstream central cones (Group-3).
Upon applying PCA to the analytical results, the first, second, and third principal components (PC1, PC2, and PC3, respectively) explained 77% of the overall variance. PC1 was characterized by a strong inverse correlation between Al and the group of elements (Na, K, and REE). This feature was estimated to be related to the compositional variation of the volcanic rocks in Hakone Volcano and/or the weathering or acidic alteration of volcanic rocks. Groups 1 and 2 were found to be more strongly affected by rock weathering or acidic alterations than Group-3. Furthermore, the standard deviation of the scores for PC1 was smaller in Group-1 compared to Group-2 and Group-3, indicating a higher uniformity in Group-1 concerning the features of PC1.
PC2 was characterized by elements found in high concentrations in volcanic ash (Sc, V, Mn, Fe, and Cu), with no significant differences in PC2 scores observed among the three groups. PC3 was characterized by Li, Al, Cu, Ga, Rb, and Sr, with no significant differences in PC3 scores observed among the three groups. At the point x, ten samples were collected from slightly varying locations on the riverbed to investigate the compositional heterogeneity at the sampling point. Heterogeneity was found in the PC2 scores, suggesting that the individual scores of PC2 at points other than the point x are inappropriate to represent each sampling point.
The absolute value of loading for As was consistently below 0.5 in all three principal components, suggesting that As did not significantly influence the overall characteristics of the samples. However, an anomalously high As content was detected in the sample at the point d. Upstream from the point d, there are hot springs discharging As, suggesting the possibility that mineral particles containing As, like pyrite, may have mixed with the sediment sample. This highlights the potential to detect volcanic activity by measuring the As content in the riverbed sediments.
This research was partially supported by the Integrated Program for Next-Generation Volcano Research and Human Resource Development funded by the Ministry of Education, Culture, Sports, Science and Technology, Japan.
