Geochemistry, Geophysics, Geosystems

Contribution of aeolian dust in Japan Sea sediments estimated from ESR signal intensity and crystallinity of quartz

Authors


Abstract

[1] Late Quaternary hemipelagic sediments in the Japan Sea contain aeolian dust from East Asia which potentially records past variations in the Asian monsoon and the westerly jet. However, extracting information about aeolian dust from Japan Sea sediments is difficult because the sediments also contain detrital material from the Japanese Islands. Here we present a method for extracting the aeolian dust components from Japan Sea hemipelagic sediments using provenance analysis of different size fractions. Hemipelagic sediments of the Japan Sea can be described by a combination of two populations characterized by lognormal grain size distributions with different median grain sizes, which are called the coarser (2–15 μm) and the finer (<10 μm) populations. We characterized these populations on the basis of provenance analyses, using the electron spin resonance (ESR) signal intensity and crystallinity of quartz. The results suggest that quartz in the coarser population has been mostly supplied from Northern China and Siberia/northeastern China by aeolian transport, whereas quartz in the finer population has been dominantly supplied by rivers draining the Japanese Islands.

1. Introduction

[2] The Japan Sea is a semi-enclosed marginal sea off the eastern coast of Eurasia that is located beneath the westerly jet and along the downwind trajectory of the Asian winter monsoon blowing from East Asia, where large-scale aeolian dust emissions have been observed. Meteorological observations show large quantities of dust are transported from East Asia to the Japan Sea and beyond during spring [e.g., Sun et al., 2001]. Irino and Tada [2000] demonstrated that hemipelagic sediments from the Japan Sea contain significant amounts of aeolian dust by a Q-mode factor analysis combined with a multiple-regression analysis of chemical and mineral compositions of the detrital component. Therefore Japan Sea sediments might record past variations in Asian winter monsoon and westerly jet intensities as variations in the grain size, flux, and provenance of the aeolian dust component of the sediments. However, it is difficult to extract such information because the sediments also contain detrital components derived from the Japanese Islands [Irino and Tada, 2000, 2002]. Tada [2004] demonstrated grain size variability of the coarse population in the Japan Sea sediments, which he assumed to be of aeolian origin although its aeolian origin is supported by limited evidences available at that time.

[3] In this study, we extracted the grain population of aeolian dust components in the hemipelagic Japan Sea sediments by evaluating the aeolian dust contribution in different sized populations on the basis of analyses of the total grain size distribution and provenance of the detrital component. Clay minerals (e.g., kaolinite, smectite, illite) and the long-lived radiogenic isotopes (e.g., 87Sr/86Sr, 143Nd/144Nd) [e.g., Grousset et al., 1988; Nakai et al., 1993; Zhang et al., 1996] are commonly used tracers for provenance studies of detrital materials. However, these parameters may not appropriately represent the source of bulk detrital materials since they are minor elements and tend to be held in the minerals of minor components of the detrital materials. In this study, we focused on detrital quartz because it is the major component of aeolian dust and resistant to alteration during weathering, transport, and diagenesis. We examined electron spin resonance (ESR) signal intensity and crystallinity index of detrital quartz to determine its provenance. These two parameters are useful for determining the provenance of aeolian dust because the ESR signal intensity of quartz reflects the age of the host rock [Toyoda, 1992; Toyoda and Hattori, 2000], and its crystallinity reflects the physical conditions of its formation such as temperature and crystallization rate [Murata and Norman, 1976]. We also estimated the size dependence of quartz provenance from the ESR signal intensity and crystallinity of quartz in the clay and silt fractions of the sediments. Taken together, the results yield information on the provenance and dominant grain size of aeolian dust in the Japan Sea sediments.

2. Locations of the Studied Sites

[4] Site MD01-2407 (37°04′00″N, 134°42′11″E; water depth = 932 m), which is located near the top of Oki ridge approximately 200 km north of southern Honshu, mainland of Japan, was selected for evaluating the contribution of aeolian dust in different size fractions of hemipelagic sediments (Figure 1). Site KT94-15-9 (39°34′31″N, 139°24′05″E; water depth = 807 m), which is located on a small terrace on the upper continental slope, approximately 50 km west of northern Honshu, was also selected. This site was expected to be strongly influenced by suspended detrital materials from Honshu, as it is close to the mouths of several major rivers that drain into the Japan Sea from northern Honshu. Both sites are away from submarine valleys and are unaffected by turbidite deposition.

Figure 1.

Studied core sites (solid circles) in the Japan Sea and sampling points for river sand (open triangles) on Honshu. Also shown are the distribution of loess and major deserts in central and eastern Asia, adapted from Sun [2002] and Laurent et al. [2005], and the atmospheric circulation patterns associated with the Asian winter monsoon and the westerly jet in East Asia.

[5] Because Honshu is a possible source of detrital material in the Japan Sea sediments [Irino and Tada, 2000], sediments in the estuaries of the Agano and Mogami rivers, which drain into the Japan Sea from Honshu (Figure 1), were used as examples of detrital material discharged from northern Honshu.

3. Studied Cores, Samples, and Analytical Methods

3.1. Studied Cores

[6] Piston cores, 50.28 m and 5.0 m long, were taken at sites MD01-2407 and KT94-15-9, respectively. Both cores are composed mainly of clay to silty clay with occasional intercalations of thin ash layers. The clay and silty clay intervals are characterized by centimeter- to decimeter-scale alternations of dark and light layers [Kido et al., 2007]. Most of the dark layers are laminated, whereas the light layers are bioturbated to homogeneous [Watanabe et al., 2007]. Basal contacts of the dark layers are generally sharp, whereas the upper boundaries are gradual, owing to bioturbation, and occasionally mottled.

3.2. Samples and Preparation

[7] Prior to the analyses, samples were pretreated in order to isolate the detrital component; carbonate was removed from samples with a 20% acetic acid solution, organic matter with a 10% hydrogen peroxide solution, and opal with a 2 mol/L sodium carbonate solution to isolate the detrital component [Tada et al., 2000; Mortlock and Froelich, 1989]. Forty-six 100 mg samples covering the last 150 kyr [Kido et al., 2007; Yokoyama et al., 2007] were subsampled from the MD01-2407 core, and two 100 mg samples covering the last 11 kyr [Crusius et al., 1999] were subsampled from the KT94-15-9 core and used for the grain size analysis after pretreatment. Eighty-three 500 mg samples covering the last 150 kyr, including all of the samples used for the grain size analysis, were subsampled from the MD01-2407 core and subjected to ESR and X-ray powder diffraction (XRD) analyses after pretreatment in order to identify the source areas of the quartz in the sediments. Two 500 mg samples covering the last 11 kyr, which were also used for the grain size analysis, were subsampled from the KT94-15-9 core and also subjected to ESR and XRD analyses after pretreatment. Several kilograms of sediments were scooped from sandbars near the mouths of the Agano and Mogami rivers and used to examine the ESR signal intensity and crystallinity of quartz derived from Honshu. These samples were sieved through a 32 μm sieve, and the <32 μm fractions were recovered. After pretreatment, 500 mg samples of the <32 μm fraction of the river samples and of four selected subsamples from the MD01-2407 core and two subsamples from the KT94-15-9 core were separated into clay (<4 μm) and silt (>4 μm) fractions using the pipette method [Krumbein and Pettijohn, 1938]. The clay fraction of the river samples and the clay and silt fractions of the MD01-2407 and KT94-15-9 core samples were subjected to ESR and XRD analyses.

3.3. Analytical Methods

3.3.1. Grain Size Analysis and Data Processing

[8] Grain size distributions of the bulk detrital component of the MD01-2407 and KT94-15-9 core samples were analyzed using a laser-diffraction-scattering grain size analyzer (Horiba LA-920) equipped with tungsten and He-Ne laser light sources. Approximately 50 mg of each sample was suspended in 50 cc of filtered deionized water, and poured into the analyzer with approximately 300 cc of 0.2% sodium pyrophosphate to aid dispersion of the particles. The dispersed sample solution was circulated in a closed transport circuit while being sonicated for 8 min to further enhance the sample dispersion. Measurement was conducted for diameter range between 0.02 and 2000 μm with 85 grain size classes, each Δlog10 (μm) = 0.06 in size. The number of iterations of the arithmetic calculation for transforming the detected laser signal into the grain size distribution was set at 50. This number of iterations was selected because it gives precise estimation of median diameter of the clay- to silt-sized populations [Nagashima, 2005]. The results were expressed as volume percent in each class. The reproducibility was better than ±2% for each grain size class.

[9] Grain size distributions of the bulk detrital samples were split into finer and coarser distributions using the peak split program Igor pro, which assumes that the finer and coarser distributions are lognormal shapes [Nagashima et al., 2004]. The fractional area of finer (coarser) distribution in total grain size distribution is defined as the finer (coarser) population content. However, this method causes underestimation of clay-sized (=finer) population content compared to the content estimated by the traditional technique of pipette method [Nagashima, 2005]. Therefore we corrected the finer population content estimated by laser-diffraction using Horiba LA-920 to the values comparable to the content estimated by pipette method based on the experimental equation between the finer component contents estimated by two different methods [Nagashima, 2005].

3.3.2. Quartz Content and Crystallinity Analyses

[10] The quartz contents in MD01-2407 and KT94-15-9 core samples and Agano and Mogami river samples were determined by the internal standard method [Klug and Alexander, 1974], with silicon (Wako Co., Ltd.) as the internal standard, using the MAC Science MXP-3 X-ray diffractometer, with tube voltage of 40 kV and tube current of 20 mA. The scanned interval was 5° to 40° 2θ, the scanning speed was 4° 2θ/min, and the sampling step was 0.02° 2θ. The crystallinity index (CI) of quartz was originally defined by Murata and Norman [1976] on the basis of the degree of resolution of the d (212) reflection of quartz at 1.3820 Å on the XRD profile. In this study, the scaling factor, which was introduced to adjust the CI of automorphic quartz to 10, was set at 1.32, using clear automorphic quartz of an industrial standard sample (20–28 mesh granular quartz; Wako Co., Ltd.) as the reference sample. Quartz crystallinity was measured with a MAC Science MXP-3 X-ray diffractometer as mentioned above with scanned 2θ interval of 66° to 69°, scanning speed of 0.5° 2θ/min, and sampling step of 0.006° 2θ. Reproducibility of the CI for five repeated XRD measurements was ±0.3.

3.3.3. ESR Analysis of Quartz

[11] The ESR signal intensity of the E′1 center in quartz, an unpaired electron in a single silicon sp3 orbit oriented along a bond direction into an oxygen vacancy [Feigl et al., 1974], is used to estimate the relative number of oxygen vacancies in quartz. Oxygen vacancies in quartz have been formed by natural radiation, and are known to increase with the age of the host rock [Toyoda, 1992; Toyoda and Hattori, 2000]. First, approximately 120-mg pretreated samples were irradiated with γ-radiation, for a total dose of 2.5 kGy, using a 60Co source at the Institute of Scientific and Industrial Research, Osaka University. The samples were then heated at 300 °C for 15 min to convert the oxygen vacancies to E1′ centers [Toyoda and Ikeya, 1991]. ESR signal intensity measurements were conducted at room temperature with an X-band ESR spectrometer (JEOL, PX-2300) at the Okayama University of Science under 0.01 mW of microwave power, and 0.1 mT magnetic field modulation (100 kHz), 5 mT scan range, 2 min scan time, and 0.03 s time constant. Intensity of the E1′ centers was normalized to the quartz content of each sample to estimate the ESR signal intensity of pure quartz. The ESR signal intensity of quartz is expressed in spin units: one spin unit is equivalent to 1.3 × 1015 spins/g [Toyoda and Naruse, 2002]. The reproducibility of ESR signal intensity was ±1.5 spin units.

4. Results

4.1. Grain Size Distribution of Detrital Materials in the Japan Sea Sediments

[12] The grain size distributions of the MD01-2407 core samples have a negatively skewed unimodal shape, with most grains <20 μm in diameter (Figure 2). Nagashima et al. [2004] reported that grain size distributions of detrital materials for samples of hemipelagic sediment derived from the northern part of the Japan Sea could be described by the combination of two lognormal distributions of grains with clay- and silt-sizes. Lognormal fitting was preferred for the Japan Sea sediments on the basis of the observation that grain size distribution of aeolian quartz in the North Pacific sediments is close to lognormal distribution [Okamoto et al., 2002]. This idea is also supported by our own observation on aeolian dust that fell in the northern part of Japan on May 2002 shows grain size distribution close to lognormal distribution [Nagashima et al., 2007]. All of the grain size distributions of the MD01-2407 core samples could also be explained by the combination of two lognormal distributions of grains with median diameters of 3–4 μm and 5.5–8 μm, respectively. We refer to the two grain populations characterized by finer and coarser lognormal grain size distributions as finer and coarser populations, respectively. The grain size distributions of the two KT94-15-9 core samples also had a negatively skewed unimodal shape with most grains <20 μm. This distribution could similarly be explained by finer and coarser lognormal distributions of grains with median diameters of 3.2–3.6 μm and 6.0–6.1 μm, respectively.

Figure 2.

Examples of grain size distributions of the detrital components of MD01-2407 and KT94-15-9 core samples. Each bulk distribution can be split into two lognormal distributions, composed of finer and coarser populations. The MD01-2407 core sample is from 1312 cm depth, and the KT94-15-9 core sample is from 106 cm depth.

4.2. ESR Signal Intensity and Crystallinity of Quartz

4.2.1. Quartz in the Japan Sea Sediments

4.2.1.1. Crystallinity Index

[13] The CI of quartz in the 83 bulk MD01-2407 core samples ranged between 7.3 and 10.6, with an average value of 8.9 and a standard deviation of 0.65 (Table 1), whereas those of the two bulk KT94-15-9 core samples were lower (7.1 and 7.5). The CI of quartz in the clay fractions of the MD01-2407 core samples ranged between 7.4 and 8.7 (average, 8.0) (Table 2); these values are comparable to values in the lower end of the range for the bulk samples from the core. The CI of quartz in the silt fractions of the MD01-2407 core samples ranged between 8.5 and 9.8 (average 8.9) and was therefore comparable to values in the high end of the range for the bulk samples. The CI values of quartz in the clay fractions of the KT94-15-9 core samples were 5.6 and 5.8, which are lower than those of the bulk samples of both cores. The CI of quartz in the silt fractions of the KT94-15-9 core samples was 7.7 and 7.8; these values are comparable to the lowest CI values in the bulk MD01-2407 core samples, and they are higher than those of the bulk KT94-15-9 core samples.

Table 1. Electron Spin Resonance Signal Intensities and Crystallinity Index Values of Quartz in Bulk Samples From the MD01-2407 and KT94-15-9 Coresa
CoreDepth, cmAge,b kaESR Signal Intensity (Spin Unit)CI
MD01-2407110.28.79.8
MD01-2407410.714.28.4
MD01-2407611.112.98.1
MD01-2407911.612.59.7
MD01-24071122.512.49.6
MD01-24071765.216.08.7
MD01-24072569.912.58.9
MD01-240727611.310.68.9
MD01-240728612.113.49.5
MD01-240729112.613.88.8
MD01-240729312.814.19.6
MD01-240730313.711.99.4
MD01-240731715.414.19.5
MD01-240731915.714.78.8
MD01-240733317.413.28.8
MD01-240733717.913.29.4
MD01-240734619.012.48.7
MD01-240735320.320.18.6
MD01-240735620.816.99.2
MD01-240736121.616.28.8
MD01-240736622.513.28.1
MD01-240736822.915.68.3
MD01-240738125.116.88.5
MD01-240738625.918.68.3
MD01-240742630.615.18.3
MD01-240742730.915.98.4
MD01-240743131.515.28.8
MD01-240744233.312.210.6
MD01-240747137.915.18.7
MD01-240748139.510.29.7
MD01-240748940.912.19.1
MD01-240749341.514.78.7
MD01-240751143.515.39.6
MD01-240752144.716.67.7
MD01-240752445.113.68.3
MD01-240753145.916.88.9
MD01-240754147.116.38.9
MD01-240755148.217.69.2
MD01-240755949.313.210.3
MD01-240756750.212.19.2
MD01-240758452.212.68.4
MD01-240758652.49.29.4
MD01-240758952.810.99.6
MD01-240760254.310.39.1
MD01-240760654.715.28.9
MD01-240761155.316.48.7
MD01-240762957.614.99.8
MD01-240764158.916.68.9
MD01-240766162.020.18.2
MD01-240767163.616.89.3
MD01-240768165.219.68.5
MD01-240769166.717.08.6
MD01-240770168.319.59.1
MD01-240773673.816.99.2
MD01-240774174.416.08.9
MD01-240774675.016.18.5
MD01-240776277.012.29.3
MD01-240776777.514.98.6
MD01-240776877.716.98.8
MD01-240780682.010.19.9
MD01-240782183.68.67.4
MD01-240783284.98.78.1
MD01-240784486.314.99.7
MD01-240786188.113.58.0
MD01-240788192.010.07.7
MD01-240789795.211.88.1
MD01-240790196.017.48.2
MD01-240790296.214.59.9
MD01-240790696.912.09.3
MD01-2407941103.88.88.4
MD01-2407951105.810.79.1
MD01-2407999112.313.88.7
MD01-24071019113.712.38.7
MD01-24071054116.012.88.7
MD01-24071085118.113.87.5
MD01-24071113119.912.97.3
MD01-24071147122.212.28.4
MD01-24071156122.813.68.1
MD01-24071204127.111.88.6
MD01-24071304137.714.18.5
MD01-24071312138.515.89.2
MD01-24071349142.813.19.0
MD01-24071362144.413.99.9
KT94-15-9106104.97.5
KT94-15-9113114.57.1
Table 2. Quartz Content, ESR Signal Intensity, and CI of Quartz in the Clay and Silt Fractions of Samples From the MD01-2407 and KT94-15-9 Cores
CoreDepth, cmAge, kaClay FractionSilt Fraction
Quartz Content, %ESR Signal Intensity (Spin Unit)CIQuartz Content, %ESR Signal Intensity (Spin Unit)CI
MD01-240735621.01716.87.73215.58.5
MD01-240782183.7207.07.42914.78.5
MD01-2407941103.9158.28.12711.48.8
MD01-24071192125.9148.68.7389.09.8
KT94-15-91069.9175.15.8224.67.8
KT94-15-911311.4126.75.6227.07.7
4.2.1.2. ESR Signal Intensity

[14] The ESR signal intensity of quartz in the 83 bulk MD01-2407 core samples ranged between 8.2 and 20.1 spin units (average, 13.9; standard deviation, 2.75), whereas signal intensities in the two bulk KT94-15-9 core samples, 4.5 and 4.9, were lower. The ESR signal intensity of quartz in the clay fractions of MD01-2407 core samples ranged between 7.0 and 16.8 (average, 10.2) (Table 1), comparable to the low-end values in the bulk MD01-2407 core samples except one sample, whereas that in the silt fractions ranged between 9.0 and 15.5 (average, 12.6), which is comparable to the signal intensity in the bulk samples. In the clay fractions of the KT94-15-9 core samples, the signal intensities were 5.1 and 6.7 (Table 1), values which are lower than those in the bulk MD01-2407 samples. ESR signal intensities of quartz in the silt fractions of the KT94-15-9 core samples were 4.6 and 7.0, values which are lower than those in the bulk MD01-2407 core samples and comparable to those in the bulk and clay fraction of the KT94-15-9 core samples.

4.2.2. Quartz From Honshu

[15] The CI values of quartz in the clay fractions of the two river samples from Honshu were 4.2 and 6.2, which were the lowest values among the samples analyzed in this study. The ESR signal intensities of quartz in the clay fractions of the two river samples were 1.5 and 4.2, which were also the lowest among the samples analyzed.

4.3. ESR Signal Intensity Versus Crystallinity Plot

[16] We plotted the ESR signal intensity against the CI of quartz for the bulk and the silt and clay fractions of samples from the MD01-2407 and KT94-15-9 cores and for the clay fraction of the two river samples in order to visualize the difference in source areas of the quartz in Japan Sea sediments obtained from different sedimentological settings and size dependency of quartz provenance (Figure 3). The bulk MD01-2407 core samples plot within a triangular area spanned by corner points, referred to as 1, 2 and 3, at (x = 23.5, y = 8.3), (8.8, 11), and (8.4, 6.9), suggesting that the bulk samples are mixtures of three end-members which plot on corner points of the triangle. Quartz in the two bulk KT94-15-9 core samples plotted relatively close to the corner point 3 but slightly outside the triangle toward lower ESR signal intensities. Quartz in the clay fractions of three of the MD01-2407 core samples plotted between corner points 2 and 3 with positions closer to corner point 3, whereas one sample plotted between corner points 1 and 3. Quartz in the clay fractions of the KT94-15-9 core samples plotted outside the triangle, toward the lower ESR signal intensity and crystallinity direction. Quartz in the clay fraction of the Agano and Mogami river samples also plotted outside the triangle within the area similar to quartz in the clay fractions of the KT94-15-9 core samples. Quartz in the silt fractions of the MD01-2407 core samples plotted inside the triangle along the line parallel to, but approximately 1.5 crystallinity index units below, the line connecting corner points 1 and 2. Quartz in the silt fraction of the two KT94-15-9 core samples plotted relatively close to corner point 3 but slightly outside the triangle toward smaller ESR signal intensities and slightly larger CIs.

Figure 3.

The relationship between ESR signal intensity and crystallinity of quartz in bulk, silt, and clay fractions of samples from the MD01-2407 and KT94-15-9 cores and samples from Honshu Island rivers. Data for Tengger Desert and Vladivostok (Y. Sun, unpublished data, 2006) are also shown as shaded areas. A thick triangle formed by the three corner points numbered 1, 2, and 3 represents the area in which bulk MD01-2407 core samples plot. Three corner points labeled A, B, and C represent probable end-members estimated on the basis of the plots of probable source areas.

5. Discussion

5.1. Provenance of Quartz in the Japan Sea Sediments Based on Its ESR Signal Intensity and Crystallinity

[17] Irino and Tada [2000] reported that the Japanese Islands and arid areas of the Eurasian continent are the probable source areas of detrital materials in Japan Sea sediments. From their statistical analyses using chemical and mineral compositions of the detrital materials in Japan Sea sediments, detrital subcomponents which are attributed to “weathered” and “not weathered” aeolian dust were found. However, they did not specify the area(s) from where aeolian dust was transported to the Japan Sea. At present, observations based on satellite images, a network of surface stations, and estimations based on size-dependent soil dust emission and transport model (NARCM) [e.g., Zhang et al., 2003; Sun et al., 2001] suggest that arid areas such as the Taklimakan Desert and the Qaidam Desert in western China, the Badain Juran Desert and the Tengger Desert in northern China, and the Otindag and Horqin sandy lands in northeastern China, are considered major source areas of dust emitted from East Asia. On the basis of meteorological observation data, Sun et al. [2001] suggested that aeolian dust emitted from the Taklimakan Desert was lifted to altitudes higher than 5000 m due to its geomorphological setting and transported by the westerly jet stream to the remote Pacific Ocean, whereas aeolian dust emitted from northern China was lifted to altitudes less than 3000 m in most cases, and transported to proximal areas such as the Loess Plateau, southern China, and offshore regions of the western margin of North Pacific.

[18] To explore the potential source area(s) of detrital materials in the Japan Sea sediments, we examined the provenance of detrital quartz based on its ESR signal intensity and crystallinity. Quartz in the fine silt fraction of loess samples from Vladivostok, located close to northeast China, is characterized by high ESR signal intensities of 16–20 spin units and moderate crystallinity with CIs of 8.4–8.7 (Y. Sun, unpublished data, 2006). The high ESR signal intensities of 16–20 are consistent with observation by Toyoda and Naruse [2002] who reported ESR signal intensity of 17 spin units for quartz in the fine silt fraction of loess samples derived from northeastern China. Quartz in the fine silt fraction from Tengger Desert in northern China is characterized by moderate ESR signal intensities of 7–13 spin units and high CIs of 9.6–9.9 (Y. Sun, unpublished data, 2006). On the other hand, quartz in the clay fraction of the samples from the Agano and Mogami rivers on Honshu analyzed in this study was characterized by low ESR signal intensities of 1.5 and 4.2 and low CIs of 4.2 and 6.2.

[19] Quartz in bulk samples of Japan Sea hemipelagic sediments plots within a triangular area on the ESR signal intensity versus CI diagram (see section 5.2; Figure 3). Corner point 1 (ESR signal intensity = 23.5, CI = 8.3) represents quartz with a high ESR signal intensity and a moderate CI. The high ESR signal intensity of >20 spin units implies that the quartz was derived from source rocks of Precambrian age [Toyoda, 1992; Toyoda and Hattori, 2000]. The quartz in the fine silt fraction of loess samples from Vladivostok plot close to the end-member 1, but slightly inside the triangle with approximately the same CI values and ESR signal intensities approximately 4 spin units lower than corner point 1. However, the difference is within twice the measurement error, and we interpret corner point 1 to represent fine silt-sized quartz derived from northeastern China, although the true end-member is probably located inside of the corner point 1 with a distance of twice the measurement error. Therefore we plot the end-member, named end-member A, to the area where quartz in the fine silt fraction samples of Vladivostok plots (Figure 3). The quartz in samples from northeastern China might have been transported from Siberia, where Precambrian bedrock is widely exposed (Figure 1), by the winter monsoon winds from Siberia.

[20] Corner point 2 (x = 8.8, y = 11) represents quartz with a moderate ESR signal intensity and a high CI of quartz, which was probably derived from igneous rocks of Paleozoic age [Toyoda and Naruse, 2002; Murata and Norman, 1976]. Quartz in fine silt fractions from the Tengger Desert is also characterized by moderate ESR signal intensity (7–13 spin units) and high crystallinity (CIs of 9.6–9.9) (Y. Sun, unpublished data, 2006), which is close to corner point 2 but located slightly inside the triangle with approximately the same ESR signal intensity and CI values of approximately 0.8 units lower than corner point 2. However, the distance is approximately twice the measurement error, and we interpret corner point 2 to represent fine silt-sized quartz particles derived from northern China although the true end-member is probably located inside the corner point 2 with a distance of twice the measurement error. Therefore we plot the end-member, named end-member B, to the area where quartz in the fine silt fraction samples of Tengger Desert plots (Figure 3).

[21] Corner point 3 (x = 8.4, y = 6.9) represents quartz with low ESR signal intensity and low crystallinity. Low ESR signal intensity implies that the quartz was derived from relatively young source rocks [Toyoda, 1992; Toyoda and Hattori, 2000], whereas the low CI of quartz implies that the quartz was formed under low temperature or rapid cooling conditions [Murata and Norman, 1976]. Quartz in the clay fraction of river samples from Honshu is also characterized by low ESR signal intensity (1.5–4.2 spin units) and low CI (4.2–6.2), which seems to reflect Tertiary to Mesozoic age rocks exposed there. In addition, quartz in the clay fraction of KT94-15-9 core samples is characterized by low ESR signal intensity (5.1–6.7 spin units) and low CI (5.6–5.8). Since core KT94-15-9 was recovered from the upper continental slope close to the mouth of the Agano and Mogami rivers draining northern Honshu, and the core shows high sedimentation rates probably resulting from a relatively large contribution of detrital materials from rivers, quartz in the KT94-15-9 core samples is considered to be mainly derived from Honshu. Thus it is reasonable that the clay fraction of the KT94-15-9 core samples and river samples from Honshu plot close to corner point 3, although they plot outside the triangle toward smaller CI and ESR intensity values. We interpret corner point 3 to represent clay-sized quartz particles derived from Honshu although the true end-member seems to be located outside the third corner point with a difference nearly three times the measurement error. Therefore we plot the end-member, named end-member C, to the area where quartz in clay fraction samples of the Agano river and KT94-15-9 core plots (Figure 3).

[22] Thus it is reasonable to consider that the quartz in the hemipelagic sediments of the Japan Sea is a mixture of quartz derived from three sources; northern China, Siberia/northeastern China, and Honshu.

5.2. Size Dependence of Quartz Provenance

[23] The size dependence of the quartz provenance in the MD01-2407 core further supports the inference that quartz in hemipelagic sediments of the Japan Sea is a mixture of quartz derived from Honshu and the Eurasian continent. Namely, quartz in the clay fraction of the MD01-2407 core samples plot inside the triangle close to the end-member indicating Honshu (end-member C) when the measurement error is taken into account (Figure 3), indicating that it is dominantly supplied from Honshu with a subordinate contribution from the Eurasian continent. Quartz in the silt fractions of the MD01-2407 core samples also plot inside the triangle, relatively close to end-members indicating northern China (end-member B) or Siberia/northeastern China (end-member A) sources, suggesting that the silt-sized quartz can be regarded as a mixture of quartz from northern China and Siberia/northeastern China, with only a minor (approximately 10 to 20%) contribution from Honshu. Quartz in the silt fraction of the MD01-2407 core samples plots in the area different from that of the KT94-15-9 core (Figure 3). Since quartz in the silt fraction of core KT94-15-9 is considered to have been dominantly derived from Honshu because of its close location to the river mouths, provenance of quartz in the silt fraction of core MD01-2407 should be different from Honshu. This finding also supports the inference that silt-sized quartz in the MD01-2407 core is derived from the Eurasian continent. Size dependence of quartz provenance between the silt and clay fractions of core MD01-2407 strongly suggests that coarser and finer populations of the core samples represent different source areas. On the basis of this idea, the ESR signal intensities and CI values of quartz in the finer and coarser populations can be estimated from those in the clay and silt fractions and quartz contents in the two fractions (Table 2) by assuming that quartz contents and ESR signal intensities and CIs of the quartz in finer and coarser populations show end values of those in each sample. On the basis of a deconvolution result of the grain size distributions that extract the finer and coarser populations from the MD01-2407 core samples and estimated quartz contents of the finer and coarser populations, quartz in the clay fractions of four MD01-2407 core samples are regarded as a mixture of 56–77 volume% of the quartz in the finer population and 23–44 volume% of the quartz in the coarser population, whereas the quartz in the silt fractions of four MD01-2407 core samples are regarded as a mixture of 8–24 volume% of the quartz in the finer population and 76–92 volume% of the quartz in the coarser population. The ESR signal intensities and CI values of quartz in the finer and coarser populations were calculated from those in the clay and silt fractions using the mixing ratio of quartz in each population within each size fraction. The estimated ESR signal intensities and CI values for the finer and coarser populations plot on the diagram in Figure 4. Quartz in the finer population plot relatively close to the end-member indicating a Honshu provenance (end-member C). Consequently, we consider that the quartz in the finer population was derived dominantly from Honshu, with a subordinate contribution from the Eurasian continent. On the other hand, quartz in the coarser population plot close to the line between the end-members A and B (indicating northern China and Siberia/northeastern China sources, respectively), suggesting that quartz in the coarser population of core MD01-2407 can be regarded as a mixture of quartz from Siberia/northeastern China and northern China, with a negligible contribution from Honshu.

Figure 4.

The relationship between ESR signal intensity and crystallinity of quartz in silt and clay fractions and estimated finer and coarser populations of samples from the MD01-2407 core.

5.3. Size Population Representing the Aeolian Dust Component

[24] In addition to the dust transported through the atmosphere, detrital components from the Eurasian continent discharged from the Yellow and Yangtze rivers could have been transported by the Tsushima Warm Current to site MD01-2407. At present, the Tsushima Warm Current occasionally transports fine detrital material in suspension from the Yellow and Yangtze rivers to the southern part of the Japan Sea [Otosaka et al., 2004]. Because only clay-sized particles can be transported for long distances of several hundred kilometers in suspension [Dobson et al., 2001], only clay-sized detrital components can be expected to have been transported to site MD01-2407. Therefore the only way silt-sized quartz of Eurasian continent origin could have been transported to the MD01-2407 core site is by aeolian transport, and the coarser population of MD01-2407 core samples must therefore represent the aeolian dust component from the Eurasian continent.

[25] This result is consistent with the study of Ishizaka [1991], who showed that aeolian dust transported from the Eurasian continent to the Japanese Islands during spring dust storm events consists of fine silt or smaller particles. On the other hand, the finer population of MD01-2407 core samples represents the detrital component from Honshu, with a subordinate contribution from the Eurasian continent, via either aeolian transport or transport by the Tsushima Warm Current. Consequently, the coarse population in hemipelagic sediments of the Japan Sea contains information on the grain size, flux, and provenance of aeolian dust. Similar results that silt-sized coarser populations in the sediments represent aeolian dust have been reported in other marginal seas off SW Africa and Arabian Sea [Prins and Weltje, 1999; Stuut et al., 2002]. In those studies, aeolian dust populations were extracted on the basis of end-member modeling which decompose grain-size distributions of those sediments into several populations that are interpreted as aeolian dust and fluvial sediment origin mainly on the basis of the spatial variation in grain size distributions of the pelagic and hemipelagic core samples in the seas and geological/meteorological settings there. Compared to those studies, our study is characterized by its direct approach to estimate the provenance area of each grain population in the sediments using the parameters of quartz, which could be adapted to the situations when only a small number of sediment cores are available or there is insufficient information of the geological/meteorological settings.

6. Conclusions

[26] We conducted a provenance study of detrital materials in the hemipelagic sediments of the Japan Sea using ESR signal intensities and crystallinity of quartz. From analyses of ESR signal intensities and crystallinity of quartz and comparison of their values with those of surface sediments in desert areas of East Asia, we estimated northern China, Siberia/northeastern China, and Honshu as the probable source areas of quartz in the sediments. Moreover, we demonstrated that quartz in the coarser population of the hemipelagic sediments was derived mostly from the Eurasian continent, whereas quartz in the finer population was derived from Honshu with a subordinate contribution from the Eurasian continent. Since the only way silt-sized detrital quartz of Eurasian continent origin could have been transported to the Japan Sea is through aeolian transport, and since the grain size of aeolian dust transported to the Japanese Islands has been reported to be silt sized or less, the coarser population in the hemipelagic sediments of the Japan Sea can be regarded as representing the aeolian dust component. Information on aeolian dust grain size and flux, which may record past variations in the Asian monsoon and the westerly jet, can thus be extracted through examination of the coarse population of detrital materials of hemipelagic sediments in the Japan Sea. In this study, we successfully demonstrate the utility of the combination of two parameters derived from quartz to discriminate different source areas of detrital materials in the Japan Sea sediments. Provenance studies using those parameters could be applied to other terrigeneous/hemipelagic sea sediments in East Asia, which increase efficiency and promote further understanding for dust-transport mechanisms in East Asia.

Acknowledgments

[27] We wish to express our thanks to Yusuke Yokoyama, Ayako Abe, and Mitsuo Uematsu for their suggestions and critical reviews. We also express our thanks to Hodaka Kawahata and Tadamichi Oba for providing us with materials from the MD01-2407 core and KT94-15-9 core, respectively. We also wish to express our sincere thanks to Laurent D. Labeyrie, Maarten A. Prins, Paul P. Hesse, and Jan-Berend Stuut for their helpful and pertinent comments. This work is a part of the doctoral thesis of K. Nagashima under the guidance of R. Tada. This research was supported by Grants-in-Aid for Scientific Research from the Japan Society for Promotion of Science (12308026, 16634012, and 15310008), awarded to R. Tada.

Ancillary