Feasibility of applying viscous remanent magnetization (VRM) orientation in the study of palaeowind direction by loess magnetic fabric

1 Introduction

The orientation distribution of magnetic minerals in rock or sediment is referred to as its magnetic fabric. From the macroscopic perspective, the magnetic susceptibility of rock or sediment is usually different when distributed in different directions, which is known as the anisotropy of magnetic susceptibility (AMS).

The AMS can be described by the sizes and directions of the maximum, intermediate and minimum axes of the AMS ellipsoid. As a geological analysis method, magnetic fabric analysis was first used by Graham [1] to study the fabric of rocks, and has since been widely applied in research on sedimentary processes and to distinguish the transport power direction of sediments [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,].

Studies of the AMS of modern aeolian sediments have proved that there is a correlation between the direction of the maximum axes’ declination of the anisotropic ellipsoid and the wind direction [11, 12, 13, 14, 15, 16, 17,]. To use the magnetic fabric of loess to trace its palaeowind direction, it is necessary to obtain information about the original direction or attitude of the loess samples, and, in addition, a sample mode of the basset section is usually taken to identify the field orientation [11].

In recent years, there have been large numbers of drilling technology applications in loess sampling, which has resulted in the availability of many core samples of loess. However, due to missing information about the original direction, those core samples cannot be used in magnetic fabric tracing research. Viscous remanent magnetization (VRM) is the remanent magnetization of magnetic minerals from the long-term presence of an applied magnetic vector field below the Curie temperature. The direction of the geomagnetic field has remained generally stable for about 0.78 Ma, so the VRM has had enough time to record the northern direction of the geomagnetic field. Measuring the VRMto reconstruct the direction of the core samples has long been proposed by researchers [18, 19, 20, 21, 22, 23,], but the accuracy of this method in matching the precision of the magnetic fabric technique when tracing the palaeowind direction has not yet been reported. In this research, our objective was to conduct some preliminary experiments regarding this issue.

Regarding the principle of VRM for determining the orientation of a drilling core, Yue et al [18] has considered this issue in detail and this will not be repeated here. We emphasize that loess is very different from hard rock, in having a certain plastic deformation capacity, so there may be a bigger VRM error at some core-sample levels than others along the length of the core [20]. To obtain more reliable directions for each sample, we suggest that each sample taken from the loess drilling core be measured with respect to their AMS and VRM.

Will there be any obvious difference if we use the VRM direction rather than the geographic direction of a sediment to calculate the palaeowind direction? Which palaeowind direction distribution calculated in which temperature range will be closest to the palaeowind direction distribution obtained from the geographic direction? To answer these questions, we collected samples from the basset section to examine the feasibility of using the VRM orientation instead of the geographic direction in determining the palaeowind direction. (We did not use drilling core samples in this experiment, because the geographical location of the core samples is difficult to obtain, and there is no way to determine the difference between the magnetic fabric positioned by VRM and that by field orientation.)

2 Sampling and measuring methods

The Duanjiapo section is located in the northern Duanjia village of Lantian County, which is about 25 km southeast of Xi’an in the province of Shaanxi (109.2° E, 34.2° N, see Figure 1), which is in a semi-humid monsoon climate zone. The loess formation in this section of the slope is continuous and complete, and the basset sediments are clear with a thickness of about 132 m. The loess sequence was divided into three formations—the Wucheng (S₁₅-L₃₄, ~1.3–2.8 Ma), Lishi (S₁-L₁₅, ~0.07–1.3 Ma) and Malan (L₁, ~0.01–0.07 Ma) [24, 25, 26, 27, 28, 29, 30,]. The bottom of Wucheng is in contact with red clay that belongs to the Neogene system and the Malan Loess is overlain by the Holocene soil (S₀). Yue et al [31], Sun et al [32] and Zheng et al [33] have conducted detailed studies on this section. Their research had great significance in the stratigraphic identifications of this study.

Figure 1

Location of Duanjiapo section, Lantian, China.

The sampling sequences of the Duanjiapo section totals 15 m in length, including L₃₂, S₃₂, L₃₃, S₃₃, L₃₄ and the upper layers of red clay. These samples were oriented using a geologic compass in the field, and were taken to the laboratory to be cut into cubes (2 cm × 2 cm × 2 cm). We performed AMS and remanent magnetization (thermal demagnetization) measurements. We measured the AMS using a Kappa Bridge MFK1A instrument with a frequency of 976 Hz, a magnetic field strength of 200 A/m and an accuracy of 2×10⁻⁸ SI.

In the thermal demagnetization process, we used an ASC TD-48 thermal demagnetization instrument at temperatures of 25°C, 50°C, 100°C, 125°C, 150°C, 200°C, 250°C, 300°C, 350°C, 400°C, 450°C, 500°C, 525°C, 550°C, 570°C and 585°C. Most of the samples’ remanent magnetizations can be reduced to below 10% of the natural remanent magnetization (NRM) at 585°C, and the heating temperatures of the other samples went higher to 600°C, 610°C, 620°C, 630°C, 640°C, 650°C, 660°C, 670°C and 680°C. The results showed that all of the samples’ remanent magnetizations went below 10% of the NRM. We measured the remanent magnetism of samples using a 2G-755 U-channel superconducting magnetometer, which has a measurement scope of 2.0×10⁻¹² – 2.0×10⁻⁴ Am² and a sensitivity of 2.0×10⁻¹² Am² (2.0×10⁻⁹ emu). We performed demagnetization and remanent measurement in a zero-magnetic-field environment (<150 nT).

The remanence measurement results show that the VRM of the sample was mostly removed at 150°C, and was almost removed at 300°C. The VRM declinations from the various demagnetization temperatures can be obtained by calculating the space coordinates. As a benchmark, we selected the declinations of NRM (hereinafter abbreviated as D_NRM. The NRM is the measured remnant magnetism of a natural sample with no demagnetization, which includes the depositional remanent magnetization, VRM and so on. Typically, the direction of NRM is not consistent with geographic north. D _NRM is the angle between the NRM direction and geographic north.) and declinations of VRM (hereinafter abbreviated as D_VRM) from some of the temperature ranges (25–50°C, 50–100°C, 100–150°C, 150–200 °C, 200–300°C) instead of the geographic north. Given these declinations, we determined whether there were significant differences in the declinations of the maximum axes of the AMS ellipsoid (hereinafter abbreviated as D _AMS) obtained from the field orientation, VRM orientation and NRM orientation.

3 Results and analysis

Figure 2 shows the distributions of D _NRM and D_VRM at various thermal demagnetization temperature ranges, with a distribution bin of 5°. The distributions of Figures 2a, 2c, 2d and 2e appear to be normal, so we fitted a Gaussian curve to the frequency distribution of the bar graph. In the fitting, to ensure that the distribution of the fitting curve was close to that shown in the bar chart, we eliminated declination values that deviated significantly from the average and would significantly increase the standard deviation. We can see that D_VRM in the demagnetization temperature range of 50–200°C is relatively concentrated, especially from 100–150°C. The distribution of D _VRM in the demagnetization temperature range of 100–150°C is most concentrated and had the smallest standard deviation (Figures 2c, 2d, 2e), so we predicted that the D _VRM from 100–150°Cmay be the most promising choice for replacing the samples’ original north direction.

Figure 2

(a) is frequency distribution chart of declination of NRM, and (b), (c), (d), (e), (f) are frequency distribution charts of VRM from different temperature ranges which are noted in the charts. The bin size of all the charts are 5°, and (a), (c), (d), (e) are fitted with gaussian curve with parameters.

As reported by Yue et al [18], the angle difference between the VRM direction of the drilling core samples and geographic north can be corrected by performing the same measurements on samples from the basset section near the drilling location. In this research, we used oriented samples from the basset section to simulate drilling core samples, and explored how the VRM orientation changed the AMS result when assuming that information about the direction of the basset samples was missing. Because our samples were taken from the basset section, we can correct the D_VRM value of the basset section itself. For instance, when we take the VRM direction as the north direction for re-calculating the D_AMS of the samples, we can directly use the average D_VRM value as the correction term δ, using the following formula:

where:

D_AMS0 is the declination of the maximum axes of the measured AMS ellipsoid, using geographic north as the benchmark direction.

D_VRM is the VRM declination.

δ is the average angle between geographic north and the VRM direction of samples from the basset section near the drilling location. Here, instead, we use the average value of each fitted Gaussian curve. Due to the disordered D_VRM distribution in the temperature ranges 25–50°C and 200–300°C (Figures 2b, 2f), the average D_VRM values of these two temperature ranges are not a good reflection of the distribution. The D_VRM of these two ranges cannot be corrected by the average D_VRM value, which is believed to be δ = 0.

The calculated D_AMS value may be negative, and if so, must be converted (plus 360°) to guarantee that all D_AMS values are distributed in the 0–360° range.

Based on the NRM orientation, we can follow formula (1) to obtain the following formula:

where:

D_AMS0 is the declination of the maximum axes of the measured AMS ellipsoid, using geographic north as the benchmark direction.

D _NRM is the NRM declination.

δ is the average angle between geographic north and the NRM direction of samples from the basset section near the drilling location. Here, we used the average value of the fitted Gaussian curve rather than this value.

The calculated D_AMS value is likewise converted to guarantee that all D _AMSvalues fall within the 0–360° range.

After comparing the D_AMS values calculated using formulas (1) and (2) with the original D_AMS0 value, we can draw the D_AMS distribution diagram as shown in Figure 3.

Figure 3

The rose diagrams show the distribution of D_AMS, (a) is distribution of D_AMS0 based on field orientation, (b) is distribution of D_AMS1 calculated from D_NRM, (c) is distribution of D_AMS2 calculated from 25–50°C D_VRM, (d) is distribution of D_AMS3 calculated from 50–100°C D_VRM, (e) is distribution of D_AMS4 calculated from 100–150°C D_VRM, (f) is distribution of D_AMS5 calculated from 150–200°C D_VRM.

In the D_AMS0 diagram (Figure 3a), we can see that the angle ranges 112.5–135.0° and 292.5–315.0° have the highest and second-highest distribution frequency, respectively. This indicates that the prevailing wind direction is almost NW–SE. Figures 3d, 3e and 3f show similar distributions to that inFigure 3a, and all show the highest distribution frequency at angles of 112.5–135.0°. These results indicate that taking the VRM direction as the north direction of the sample can obtain relatively reliable wind direction information. However, in Figures 3d, 3e and 3f, the frequency of the D_AMS distribution at angles of 112.5–135.0° is reduced, which indicates that the adoption of VRM orientation may result in weakening the advantage of the high-frequency wind direction. To quantitatively compare the impact of using the VRM orientation method on the D_AMS distribution at different demagnetization temperature ranges, we divided the circumferences into 16 equal bins, and Table 1 presents the statistical D_AMS0 and D_AMS results.

In the table, we can see that the D_AMS frequency distribution is a discrete distribution. The D_AMS0 distribution is the expected theoretical distribution, and each of the other distributions can be regarded as observational. Because the theoretical distribution reflects the distribution of the palaeowind direction, it cannot be described by any canonical distribution such as a Gaussian or Cauchy distribution. Therefore, we can only use a nonparametric test to determine the difference between these discrete distributions. We determined the differences between the observations and our theoretical results by performing a Pearson’s chi-square test of the goodness of fit (a kind of nonparametric test suitable for the distributions of categorical data analysis):

Null hypothesis H₀: the D_AMS distribution is consistent with the D_AMS0 distribution (or, we can say there is no significant difference between the two).

The formula for the Pearson’s chi-square test of the goodness of fit is as follows:

where:

n is the number of bins, and n = 16, according to the statistical grouping scheme shown in Table 1,

f_i is the D_AMS frequency for bin i (observation data) and

F _i is the D_AMS0 frequency for bin i (theoretical data).

The basic reason for performing the Pearson’s chi-square test of the goodness of fit is to determine how likely it is to find any observed differences between the sets of observation and theoretical results. The value of the chi-squared test, χ², depends on the difference between distributions of the observation and theoretical results.

The smaller is the χ² value, the smaller will be the difference between the distributions of the observation and theoretical results. In the opposite case, the null hypothesis is more likely to be rejected, and will indicate that there is significant difference between these distributions.

The last row of Table 1 shows the χ² results for D_AMS1−7. At a significance level α = 0.05, the chi-square table of critical values shows that for degrees of freedom of 15 (n-1, n = 16), the critical value is 25. Only the chi-square value of 21.8 from D_AMS4 is smaller than this critical value of 25, which allows us to accept the null hypothesis and state that there is no significant difference between the calculated and theoretical results. The above test shows that the D _AMS distribution obtained from the VRM orientation is consistent with the D_AMS0distribution obtained by field orientation when the VRM is from a demagnetization temperature range of 100–150°C.

We also tried to divide the circumferences into 32 equal angle ranges to perform the Pearson’s chi-square test of goodness of fit, and the results showed that there was no D _AMS distribution obtained from the VRM orientation that was consistent with the D_AMS0 distribution obtained by field orientation. Therefore, we found that by using the VRM orientation in the palaeowind direction study traced by AMS, the resolution can reach 22.5° (360°/16), but cannot reach 11.25° (360°/32). These findings indicate that there may be an upper limit in the resolution when using VRM orientation to replace the field orientation in palaeowind direction studies traced by AMS.

4 Conclusions and discussion

In this study, we obtained the following results.

1. For field-oriented samples from the Duanjiapo section in Lantian, the distribution of the VRM direction from a demagnetization temperature range of 100–150°C is more concentrated than those from other demagnetization temperature ranges, and has the smallest standard deviation.

2. The distribution of the declinations of the maximum axes of the AMS ellipsoid based on VRM(100–150°C) orientation is closest to that based on the field orientation. There is no significant difference between them when the bin angle is 22.5° and the significance level α = 0.05.

From the above results, when we consider the magnetic fabric to trace the palaeowind direction with a proper resolution of the wind direction, it is feasible to use the VRM orientation instead of the field orientation for loess samples for which the directions are difficult to mark during sampling (such as drilling core samples), but it may be necessary to pay close attention to the demagnetization temperature ranges when using VRM data.

Zeeden et al [15] reported that the direction of lineation in different layers can change by about 90°. We agree with these authors that magnetic fabric properties should be divided into three intervals for analysis. The relationship between D _AMS and D_VRM can appear too complex when considering the whole sequence, and this relationship can be better analyzed in different layers.

Our conclusions in this paper are based only on samples from the Duanjiapo section in the Lantian area on the Chinese Loess Plateau. Readers are advised that other regions must be further researched and validated before being assured of the same conclusions.