Coarse versus fine-grain quartz optical dating of the sediments related to the 1985 Ms7.1 Wuqia earthquake, northeastern margin of the Pamir salient, China
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.
Introduction
Dating paleo-earthquake events is very important for estimating earthquake hazards. High-resolution and high-precision dating of seismic events that help to calculations of recurrence intervals on timescales of decades to several thousand years is challenging, particularly in semi-arid to arid settings, where organic debris for 14C dating is uncommon. Recent studies (e.g., Banerjee et al., 2001; Lu et al., 2007; Porat et al., 2009; Fattahi et al., 2010; Liu et al., 2010; Wang et al., 2012) have shown that optical dating of earthquake-related deposits is capable of producing robust chronologies of paleo-earthquake events. However, a potential problem in such settings is that light exposure of earthquake-related sediments may have been insufficient to fully reset (bleach) the OSL signal prior to deposition (Cunha et al., 2009; Yang et al., 2012).
Incomplete resetting of the OSL signal will lead to age overestimation. Fuchs et al. (2005) and Olley et al. (1998) examined the dependence of bleaching on grain size. It is now well known that quartz OSL signals bleach more easily than feldspar IRSL (infrared stimulated luminescence) signals (Yang et al., 2012), and in many cases it appears that coarse grain (74–250 μm) bleaches better than fine grain (4–11 μm) quartz (Yang et al., 2012). But in fact, samples from different depositional environments may have different bleaching characteristic.
The August 23, 1985 Ms7.1 Wuqia Earthquake ruptured the eastern segment of the seismically active Pamir Front Thrust at the leading northeastern edge of the Pamir salient of the Tibetan Plateau, producing a 15-km long surface fault scarps, and destroying the old Wuqia town (Feng et al., 1986). Data on the recurrence period of such devastating earthquakes is a vital input for seismic hazard evaluation and risk assessment in this region. In this study, we test one modern sample and three samples from a trench excavated across the 1985 Wuqia Earthquake surface rupture (Fig. 1) using a variety of techniques on fine grain quartz (FQ) and coarse grain quartz (CQ). The purpose is to test assumptions regarding the zeroing of the deposition after the Wuqia earthquake, to assess which grain sizes are most appropriate and which dating technique can provide reliable ages for earthquake related rapidly-deposited sediments. Finally we also applied time-saving protocols such as small aliquot standardised growth curve (SA-SGC) (Roberts and Duller, 2004; Lai et al., 2007, Li et al., 2015) to check its potential for dating heterogeneously reset samples.
Fig. 1
Aster image showing regional tectonic structure, the 1985 Wuqia earthquake surface rupture and trench location. Modified from Li et al. (2012).
Materials and methods
Samples and instruments
A trench (Fig. 2b) was excavated across a small sag pond of the 1985 Ms7.1 Wuqia Earthquake surface fault scarp at a fluvial terrace and exposed a sequence of slope deposited silts and sands mainly sourced from the Pliocene sandstone and siltstone rocks 200 m to the north (Fig. 1). Four earthquake events including the 1985 event were identified in the trench. Three samples deposited before the 1985 earthquake were collected from trench, and one sample deposited after the 1985 earthquake (Fig. 2 Left) were selected as a known age sample. Cosmogenic 10Be depth-profile ages on detrital sand and pebble from a trench on the same terrace several km to the west from our study area (Fig. 1 CRN sample) are 8.0 ± 1.1 ka and 8.8 ± 1.5 ka, respectively (Thompson, 2013). The OSL samples we collected are deposited on top of the same terrace surface, should be younger than 8.0 ka.
Fig. 2
Location of the modern sample LED11-7 (Left); Trench wall showing the location of samples LED11-1, 11-2 and 11-6 (Right).
The OSL samples were collected in stainless steel tubes (length 15–20 cm, diameter 6 cm). Both ends were immediately sealed with aluminium foil and taped to prevent light leakage and loss of water during transport and storage. The samples were extracted from the inner part of the tubes and quartz was purified using conventional sample preparation techniques (dissolving carbonates with 10% HCl, oxidizing organic matter with 10% H2O2, settling using Stokes’ Law, dissolving feldspars by 30% fluorosilicic acid, followed by immersion in 10% HC) for the 4–11 µm fine grain quartz (FQ) fraction (Aitken, 1998; Liu et al., 2010), and 180–250 µm coarse grain quartz (CQ) fractions (sieving, 10% HCl, 10% H2O2, etching with 40% HF, followed by immersion in 10% HCl). The purity of the quartz extracts was confirmed by the absence of an IRSL signal for both FQ and CQ. Subsamples of FQ had IR signals close to background levels (Fig. 3a), with an IR-OSL depletion ratio between 0.9 and 1.1. Nevertheless, an IR-OSL depletion test (Duller, 2003) was carried out on every aliquot for small aliquot CQ and single grain (SG) quartz. All the De values of samples were measured on a Risø TL/OSL Reader model DA-20 equipped with a calibrated 90Sr/90Y beta radiation source (dose rate 0.1125 Gy/s for CQ and 0.1045 Gy/s for FQ in the standard configuration), blue (470 ± 30 nm; ~50 mW/cm2) and infrared (880 ± 40 nm, ~145 mW/cm2) light sources, and detection through a 7-mm thick U-340 glass filter.
Fig. 3
Luminescence characteristics. a – Quartz purity IR check; b – Dose recovery as a function of preheat temperature for coarse grains quartz; c – OSL decay curves for single quartz grains from sample LED11-1, the inset shows a corresponding dose response curve (DRC), the red point is recycling point; d – A histogram of dose recovery ratios for single quartz grains using a 10 s preheat at 260°C for and a 220°C cutheat.
Luminescence characteristics
We carried out dose recovery tests on sample LED11-6 (180–125 μm quartz). Eighteen aliquots were first beached by sunlight in groups of three, then given a known laboratory beta dose (10 Gy), after which the standard SAR-protocol (Murray and Wintle, 2000) was applied using six different preheat and cutheat temperatures (preheat for 10 s at 180–280°C) followed by cutheat 40°C lower than preheat). The measured dose is indistinguishable from the known given dose at preheat temperatures of 260°C and cutheat 220°C (Fig. 3b). In addition, we applied a dose recovery test to 24 single quartz grain using a 260°C preheat for 10 s and a 220°C cutheat, with results showing that the average dose recovery ratio is 0.99 (Fig. 3d).
Based on the above results, a preheat of 260°C and a cutheat of 220°C were selected for all the samples.
Equivalent dose measurements
Fine grain and single grain quartz
The FQ were measured using a sensitivity-corrected (by a test dose) multiple aliquot regenerative-dose (SMAR) protocol (Lu et al., 2007; Liu et al., 2010). At least six aliquots were measured to determine the natural OSL intensity, and the dose response curve (DRC) was constructed from five aliquots that brackets the natural (normalized) OSL signal (Fig. 4a). In general, the De values of the fine grain quartz using SMAR protocol are similar to those of the SAR protocol (Lu et al., 2007; Liu et al., 2010; Yang et al., 2012). We have compared the results of SMAR and SAR protocol of the fine grain quartz for one sample. Both results are the same. But the SMAR protocol can save much more measurement time. So we used the SMAR protocol for fine grain size quartz.
Fig. 4
a – Sensitivity-corrected multiple aliquot regenerative-dose growth curve (SMAR) of sample LED11-1; b – OSL decay curves of sample LED11-1; c – A SAR growth curve; d – Coarse grain quartz small aliquot standardized growth curve (SA-DRC). The red diamonds are sample-specific and were projected on the SA-DRC.
Small aliquot coarse grain quartz standardised growth curve (SA-SGC)
Single grain results show that only 1.5–3.7% of the grains yield a detectable OSL signal, a very time consuming endeavour considering that >95% of the grains were without any signal. To improve efficiency, we measured very small aliquots (~80–220 grains per disc as observed under the microscope), as most likely only one or two quartz grains on each disc give a signal, mimicking single grain measurements (Arnold et al., 2012; Medialdea et al., 2014; Faershtein et al., 2016). The SA-SGC was used to measure a large number of De values: first a well constrained SAR DRC was constructed for each sample using 5–7 large aliquots from each sample (Fig. 4c), then the Ln/Tn, a zero dose point and one regenerative dose point were measured for a large number (up to 144) of small aliquots (Fig. 4d). The Ln/Tn of each individual small aliquot was projected on the sample-specific DRC to calculate the De for that particular aliquot. Aliquots were rejected if the single regenerative dose point did not lie on the SA-SGC within SA-SGC errors. This procedure is advantageous for samples with heterogeneous resetting of the OSL signal prior to deposition, as it allowed rapid measurements of a large number of small aliquots that could be later used for calculating a meaningful De (Table 2).
Dose rate results for the samples.
Sample No.
Depth (m)
U (ppm)
Th (ppm)
K (%)
Water content
Water contents were estimated at 5 ± 5% or 2 ± 2%
(%)
Dose rate (Gy/ka)
CQ
FQ
LED11-1
2.4
1.65
5.3
1.37
5 ± 5
2.1 ± 0.1
2.9 ± 0.2
LED11-2
2.8
1.74
4.84
1.15
5 ± 5
2.0 ± 0.1
2.6 ± 0.1
LED11-6
0.6
1.96
7.75
1.49
2 ± 2
2.7 ± 0.2
3.4 ± 0.2
LED11-7
0.2
1.35
4.67
1.2
2 ± 2
2.1 ± 0.1
2.5 ± 0.1
De values of single grains, small aliquot SA-SGC and fine grains for the samples shown on Fig 2.
Sample No.
Single grains
Small aliquot SA-SGC
Fine grains
n/N
n represents the number of accepted single grains or small aliquots that passed the SAR rejection criteria and used for De determination, N represents total number of grains or aliquots measured;
P
Proportion (%) of the grains or aliquots that have a detectable OSL signal;
(%)
OD
over-dispersion;
(%)
De
De obtained using the minimum age model.
(Gy)
Age (ka)
n/N
n represents the number of accepted single grains or small aliquots that passed the SAR rejection criteria and used for De determination, N represents total number of grains or aliquots measured;
P
Proportion (%) of the grains or aliquots that have a detectable OSL signal;
(%)
OD
over-dispersion;
(%)
De
De obtained using the minimum age model.
(Gy)
Age (ka)
De (Gy)
Age (ka)
LED11-1
158/4800
3.3
45
$14.3_{-1.7}^{+1.9} $
$6.7_{-0.9}^{+1.0} $
43/78
55
36
$14.3_{-3.0}^{+3.3} $
$6.7_{-1.5}^{+1.6} $
65.7 ± 5.4
22.9 ± 1.9
LED11-2
55/2300
2.4
62
$9.6_{-1.9}^{+2.5} $
$4.9_{-1.0}^{+1.3} $
63/142
44
59
$9.8_{-1.8}^{+3.0} $
$5.0_{-1.0}^{+1.6} $
157.6 ± 14.3
60.5 ± 5.4
LED11-6
55/3700
1.5
45
$5.2_{-1.0}^{+1.4} $
$2.0_{-0.4}^{+0.5} $
52/144
40
33
$5.4_{-0.5}^{+1.1} $
$2.0_{-0.2}^{+0.6} $
66.8 ± 3.1
19.5 ± 0.9
LED11-7
53/1800
2.9
80
$0.2_{-0.1}^{+0.2} $
$0.1_{-0.05}^{+0.1} $
42/88
47
59
$0.3_{-0.2}^{+0.3} $
$0.15_{-0.1}^{+0.15} $
7.0 ± 0.3
2.8 ± 0.1
Data analysis
For FQ, De values were calculated using the sum of photons detected in the first 1 s of the stimulation minus the average of the last 10 s (background) of the OSL decay curve.
OSL data from single grains and small aliquot were analyzed using the Risø Analyst software V4.31 and DRC was fitted by the exponential function for all the datasets. The De for small aliquots was calculated using early background subtraction (i.e., the signal integrated from the first 0.32 s and background integrated from the next 0.96 s). The De for a single grain was calculated by integrating channels 4–10 (of 250 channels, 1 s stimulation) and subtracting the background from channels 40–50. Aliquots or grains were rejected if any of the following criteria applied: (1) no natural signal was present; (2) test dose error were >10%; (3) the response to the test dose was less than 3 standard deviations above background; or (4) the recycling ratio or IR-OSL depletion ratio (Duller, 2003) were not consistent within 10% (15% for single grains) of unity. A 3.0% and 2.0% measurement error was incorporated for single grain data (Duller, 2006) and small aliquot data (estimated from Armitage et al., 2000; Duller, 2007), respectively. We consider the uncertainty sources for every aliquot or grain final De included counting statistics, curve fitting error, instrument reproducibility uncertainty of 1.5% (Risø Analyst Manual).
Dose rates
The U, Th, and K contents were measured by inductively coupled plasma mass spectrometer in Genalysis Laboratory Service, Australia, and the values are given in Table 1. Water content is from fresh sample we measured 5 ± 5% for the bottom sample LED11-1, 11-2 and 2 ± 2% for the above sample LED11-6, 11-7. A relative uncertainty of 100% was assigned to all moisture content measurements, to accommodate the effects of any potential variations during antiquity. The samples are above groundwater level and the current water content may have been constant over the whole burial span of the sediment. The dose rate for each sample is calculated directly by using the observed radionuclide concentrations and attenuation values from Aitken (1998). All the samples are above groundwater level and the water content was assumed to be 5 ± 5% for the bottom sample LED11-1, LED11-2 and 2 ± 2% for sample LED11-6 and LED11-7. α efficiency compared to beta irradiation of 0.04 ± 0.02 for fine-grained quartz was used, as suggested by Rees-Jones (1995) and Lu et al. (2007). Cosmic dose rates were evaluated using present day burial depths. The dose rates are given in Table 1 and are similar for all samples, ranging from 2.0 to 2.7 Gy/ka for coarse grain and 2.5 to 3.4 Gy/ka for fine grains (Table 1).
Results
As Table 2 shows, the De values of FQ range from 7.0 Gy to 158 Gy and the ages from 2.8 ka to 60.5 ka. For SG, we measured a sufficient number of grains so that there were more than fifty accepted for every sample. The percentage of grains with detected signal is lower than 4%. The over-dispersion values range from 45% to 80%. They are higher than average value 20% comparing with the global overdispersion datasets (Arnold and Roberts, 2009). It may be caused different by degrees of the signal resetting. Therefore we used the four parameter version of the minimum age model (Galbraith et al., 1999) which can be used by a likelihood ratio test such as the maximum log likelihood (Lmax higher than 1.92) to isolate the best bleached grains and calculate the most representative De values. In the four-parameter version (MAM-4), δi is drawn from a population in which a proportion ρ of values are equal to ɤ and the remaining values are larger, having a normal distribution with parameters μ and σ truncated at ɤ. Therefore, exp(ɤ) represents the minimum true dose (in Gy) of the population (Arnold et al., 2009). Prior to running the MAM-4, we added, in quadrature, an overdispersion value of 20% to each individual measured De error, to represent the underlying dose overdispersion typically present in well-bleached and undisturbed samples (Arnold and Roberts, 2009; Arnold et al., 2009). The De
values and ages range from
$\begin{array}{}
0.2_{-0.1}^{+0.2}
\end{array} $ Gy to
$\begin{array}{}
14.3_{-1.7}^{+1.9}
\end{array} $ Gy and from
$\begin{array}{}
0.1_{-0.05}^{+0.1}
\end{array} $ ka to
$\begin{array}{}
6.7_{-0.9}^{+1.0}
\end{array} $ ka, respectively. For the same samples we measured at least forty accepted small aliquots that passed the SAR rejection criteria for each sample using CQ SA-SGC. The proportion of the accepted aliquots, 40–55%, is much higher than for single grains, and the over-dispersion values range from 33% to 59%, lower than for SG. The De values and ages range from
$\begin{array}{}
0.3_{-0.2}^{+0.3}\text{ Gy to}14.3_{-3.0}^{+3.3}
\end{array} $ Gy and from
$\begin{array}{}
0.15_{-0.1}^{+0.15}\text{ Ka to }6.7_{-1.5}^{+1.6}
\end{array} $ ka, respectively.
Discussion
Our study compared OSL ages from different grain-size fractions with different techniques based on the same sample. Ideally a sample should give similar ages for different grain-size fractions and different protocols, but this was not the case. We have one modern sample (LED11-7) deposited after 1985 as an independent age control. We did not find any carbon in the sequence. The depositional environments are fluvial and pluvial deposition in the arid mountainous regions. For FQ, the ages in the trench section are not in chronological sequence and all the ages are much older than the cosmogenic 10Be ages of 8–8.8 ka. The age of the modern sample LED11-7 is 2.8 ± 0.1 ka, overestimating the true modern age. De values of the FQ are 5 to 17 times higher than CQ. The apparent overestimation of OSL ages most likely results from the different aliquot size used for fine and coarse grains. For each of these different measurements, there is a very different number of grains on the disc that contribute to the luminescence signal. Roughly one million grains are present on a fine grain aliquot, an abundance that averages any grain-to-grain dose variations (Wallinga, 2002; Duller, 2008). This averaging masks the poorly bleached grains and results in an overestimation of the true depositional age of the sediment. Therefor fine grain quartz OSL dating is not suitable for such pluvial sediment in association with paleo-earthquakes.
SG ages show a good chronological sequence in the trench section. The MAM age of the modern sample LED11-7 is
$\begin{array}{}
0.1_{-0.05}^{+0.1}
\end{array} $ ka, sufficiently close to the zero. It shows that this modern analogue is representative of depositional environments, partial bleaching is not the cause using single grains and coarse grain SA-SGC. The lowest sample from the trench located on the terrace (LED11-1) gave a MAM age of
$\begin{array}{}
6.7_{-0.9}^{+1.0}
\end{array} $ ka, consistent with the cosmogenic 10Be. Using single grains can separate the best bleached grains from the entire grain population, implying that the single grain results are much closer to the true ages. We suggest that SG measurements combined with the appropriate model is suitable for optical dating of poorly bleached sediments associated with paleo-earthquakes, a conclusion similar to that reached by Porat et al. (2009).
The small aliquots of the SA-SGC have ~80–220 grains per disc; with the small fraction of luminescent grains (1.5–3.7% of the grains emit a luminescence signal according to the single grain results), only 1–2 grains give a luminescence signal on each small aliquot. In addition, the De values and ages obtained by the SA-SGC are, within errors, the same as those measured for single grains. In fact, the De value calculated for sample LED11-2 using 49 (2-mm) aliquots is
$\begin{array}{}
19.5_{-3.8}^{+6.4}
\end{array} $ Gy, however when additional smaller aliquots were measured (~50–120 grains, 1 mm disc), a De of
$\begin{array}{}
9.8_{-1.8}^{+3.0}
\end{array} $ Gy was obtained using MAM3, similar to the single grains results. Measuring coarse grain quartz using SA-SGC instead of SG provides a time-saving procedure when dating heterogeneously bleached sediments, with results similar to that of SG. But it is necessary to ensure that there is a small enough number of grains on a disc and enough accepted aliquots.
Conclusions
In this study, fine grains quartz measurements overestimated the expected De values and were much older than single grains and coarse grain SA-SGC. The residual dose
$\begin{array}{}
0.2_{-0.1}^{+0.2}
\end{array} $ Gy measured using single grains of the modern sample is insignificant when dating samples which are thousands of years old in paleo-earthquake settings. Only 1.5–3.6% of the coarse-grain quartz have a detectable OSL signal. The ages measured for three samples in the trench section using single grains are in stratigraphic sequence and are consistent with that of the cosmogenic 10Be dating, indicating that such samples might be suitable for optical dating of paleo-earthquakes. Coarse grains SA-SGC De values are similar to those measured for single grains. This is an efficient way to measure poorly bleached quartz which has good potential for future studies of paleo-earthquakes. Notably, enough accepted grains or aliquots are required for calculating De values. Detailed and thorough scientific studies should be carried out in order to improve accuracy for paleo-earthquake ages. We recommend using caution when interpreting ages derived from fine grain measurements.
Fig. 5
Radial plots and KDE plots from single grains (SG) and small aliquot standardised growth curve (SA-SGC) for samples LED11-1, LED11-2 and LED11-6. Grey bands on radial plots show the samples’ De values with 2-sigma error (Table 2). Only KDE plots are shown for the sample LED11-7.
