Application of the pseudo–Thellier technique to a paleomagnetic record from lake El Trébol (Patagonia, Argentina)

C.S.G. Gogorza^1,2*, S. Torcida³, M. A. Irurzun^1,2, M. A. E. Chaparro^1,2 and A. M. Sinito^1,2

¹ IFAS. Instituto de Física Arroyo Seco, Universidad Nacional del Centro, Pinto 399, 7000, Tandil, Argentina. * Corresponding author: cgogorza@exa.unicen.edu.ar

² CONICET. Consejo Nacional de Investigaciones Científicas y Técnicas, Rivadavia 1917, 1033, Buenos Aires, Argentina. E–mail: claudia_gogorza@hotmail.com

³ Instituto Multidisciplinario de Ecosistemas y Desarrollo Sustentable, Universidad Nacional del Centro, Pinto 399, 7000, Tandil, Argentina

Resumen

La técnica de pseudo–Thellier se aplicó con la finalidad de obtener registros de paleointensidades relativas utilizando un testigo de sedimentos del Lago El Trébol (Patagonia, Argentina).

En un conjunto de muestras, se llevaron a cabo mediciones de la magnetización remanente natural residual (NRM_left) luego del proceso de desmagnetización por Campos Alternos en función de la intensidad de la magnetización remanente an–histérica (ARM _gained) en el mismo valor de campo. Dos versiones del método de re–muestreo jackknife se usaron para estimar el error de los registros de paleointensidades.

Los resultados obtenidos mediante el método de pseudo–Thellier se compararon con trabajos previos de los autores en los cuales la magnetización remanente a 20mT (NRM_20mT) fue normalizada usando la magnetización remanente anhistérica a 20mT (ARM_20mT), la magnetización remanente isotérmica de saturación a 20mT (SIRM_20mT) y la susceptibilidad magnética (k) (Gogorza et al., 2006). Los registros obtenidos a partir del método de pseudos–Thellier muestran un acuerdo razonable con los obtenidos mediante el método standard de normalización (NRM_20mT/ARM_20mT).

Palabras clave: Pseudo–Thellier, paleointensidad relativa, lago El Trébol, re–muestreo jacknife.

Abstract

Measurements of intensity of natural remanent magnetization left (NRM_left) after AF demagnetization versus intensity of anhysteric remanent magnetization gained (ARM _gained) at the same peak were carried out on a set of samples. Two versions of a jackknife resampling scheme were used to get error estimates on the paleointensity.

The pseudo–Thellier paleointensity records were compared with the authors previous results where the remanent magnetization at 20mT (NRM_20mT) has been normalized using the anhysteric remanent magnetization at 20mT (ARM_20mT), the saturation of the isothermal remanent magnetization at 20mT (SIRM_20mT) and the low field magnetic susceptibility (k) (Gogorza et al., 2006). The pseudo–Thellier record shows a reasonable agreement with the standard method of normalization (NRM_20mT/ ARM_20mT).

Key words: Pseudo–Thellier, relative paleointensity, lake El Trébol, jackknife resampling, robust–resistant slope estimation.

Introduction

Many efforts have been made to recover the paleointensity from sedimentary records taking into account the potential that lake sediments have for providing continuous and high–quality records of the past geomagnetic field.

The conventional method of extracting geomagnetic field information from the signal consists in normalizing the "cleaned" natural remanent magnetization (NRM) at some demagnetization level with some normalizer. Different normalizers have been proposed, such as the anhysteretic remanent magnetization (ARM), the saturation isothermal remanent magnetization (SIRM) and the magnetic susceptibility k (Tauxe, 1993). It is assumed that the normalizers minimize the effects of magnetic grain–size distribution and variation of magnetic input, for example, as determined by environmental effects.

Several studies have been carried out to recover the relative paleointensity from marine and terrestrial records (Brachfeld and Banerjee, 2000; Sagnotti et al., 2001; Nowaczyk et al., 2001; Pan et al., 2001; Stoner et al., 2002; Brachfeld et al., 2003; St–Onge et al., 2003; Gogorza et al., 2004, 2006). Each study holds that their results represent the true geomagnetic field intensity and are free of environmental effects. Although the criteria summarized by Tauxe (1993) are strict, one must be aware that there might still be some climate influence left in the normalized NRM record and that we might not be looking at a purely geomagnetic signal (Kruiver et al., 1999).

Another way to obtain paleointensity records from sediments is the pseudo–Thellier method, based on AF demagnetization and ARM or IRM acquisition (Tauxe et al., 1995). The goal of this method is to diminish the environmental contamination of the paleointensity signal better than the conventional normalizing methods (Kruiver et al., 1999).

Geologic Setting

The sediment core studied in this work is core lt98–2 from Lake El Trébol (41°04'S 71°29'W). Lake El Trébol, a closed basin, is an oligotrophic, small lake (surface area: 0.4km², maximum depth=11m), located at 758m a.s.l. on the east side of the Andean Patagónica Cordillera, in a wooded area with moderate human influence. Presently, there are no perennial stream discharges into the lake and the hydrological budget is dominated by groundwater influx and losses by evaporation (Bianchi et al., 1999; Gogorza et al., 2006).

Lake El Trébol sedimentary stratigraphy shows strong similarities to others obtained by the authors from surrounding lakes (Gogorza et al., 1999; 2001; 2002). The push corer reached the basement (or erratic blocks), passing through a sedimentary column, which is represented by the cores (Irurzun et al., 2006; Gogorza et al., 2006).

Material and Equipment

The core investigated in this paper (lt98–2) was recovered at water depths of about 10 m from Lake El Trébol in 1998 using a push corer installed on a raft with a central hole (Gogorza et al., 2006).

One half of each core was subsampled with cubic plastic boxes of 8 cm³. A total of 281 samples were obtained.

A Spinner Fluxgate Magnetometer Minispin, Molspin Ltd. was used for measurements of remanent magnetization and a Shielded Demagnetizer Molspin Ltd. was used to isolate the magnetization components. A Pulse Magnetizer IM–10–30 AC Scientific and Shielded Demagnetizer Molspin Ltd. with an ARM device were used for IRM and ARM acquisition experiments, respectively.

Experimental Methods

Paleomagnetic Analysis

In order to test the stability of the NRM and to remove any viscous overprints, progressive alternating field (AF) demagnetization was carried out. In the orthogonal demagnetization (Zijderveld) diagrams, we generally observed straight lines that passed through the origin of the demagnetization diagram, indicating a single component and stable characteristic remanent magnetization. A limited number of samples show a small viscous remanent magnetization (VRM) that is progressively destroyed at 10 or 15mT. Further steps revealed a high stability of the magnetization directions (Irurzun et al., 2006).

Paleointensity determinations were carried out according to Tauxe et al. (1995) with the pseudo–Thellier method on a group of 40 selected pilot samples from core lt98–2. The NRM is demagnetized using an alternating field (AF) in 13 steps up to 95mT using steps of 5–10mT. Then the ARM is imparted at the same field steps as the NRM demagnetization. The ARM acquired at 95mT AF was demagnetized in the same way as the NRM. Finally, the sample is subjected to an increasing field to analyze acquisition of IRM using the same field steps as the initial demagnetization of the NRM, continuing to saturation. This SIRM was also demagnetized (Tauxe et al., 1995). The results obtained after this procedure for sample 280 are plotted in Fig. 1. The average median destructive field (MDF) is 23+2mT for the NRM, 25.7+0.5mT for ARM and 26±1mT for SIRM. ARM acquisition and demagnetization curves show a crossing at about 37% and a comparable crossing for IRM data at about 27%. The ARM data are, thus, more stable against demagnetization than the IRM (Tauxe et al., 1995).

Rock Magnetic Analysis

In order to study the nature of the magnetic minerals in the sediments, a set of experiments were carried out by Gogorza et al. (2006). The following measurements were performed for all samples: NRM; magnetic susceptibility at low frequency (specific, X and volumetric, k); isothermal remanent magnetization (IRM) in increasing steps up to 1.2T, reaching the SIRM; back field, in growing steps until cancelling the magnetic remanence; anhysteric remanent magnetization (ARM_100mT), with a direct field of 0. 1mT and a peak alternating field of 100mT. Associated parameters calculated by Irurzun et al. (2006) were also used: S–ratio (IRM_–300mT JSIRM), remanent coercitive field (B_CR), SIRM/ k, ARM_100mT /k and SIRM/ARM_100mT. In addition to these studies, hysteresis curves and temperature dependence of SIRM were obtained for a set of discrete samples. These studies show that the dominant remanence carriers are magnetite and titano–magnetite within the pseudo–single domain (PSD) grain size and that the concentration of these minerals varies between 0.01 to 0.1%. A more detailed description of these items is given in Gogorza et al. (2006) and Irurzun et al. (2006).

Fig. 2 graphically displays the linear behavior of ARM left (ARM_left) versus ARM gained (ARM _gained) at the same peak field ("Arai Plot" of Nagata et al., 1963) and the marked curve behavior of IRM left (IRM_left) versus IRM gained (IRM _gained) (sample 108). The slight deviation of the ARM . versus ARM . plot from the linear behavior may be due to the existence of a magnetic interaction. IRM acquisition curves and corresponding DC and AF demagnetization curves were analyzed to test the magnetic interaction (Henkel, 1964; Cisowski, 1981; Kruiver et al., 1999). Stepwise acquisition was carried out up to fields of 1.2T; then a progressive demagnetization was applied up to 1.2T for the DC experiment (Fig. 3a) and up to 95mT for the alternating field (AF) (Fig. 3b). Fig. 3a shows that about 90% of the SIRM is obtained between 200 and 300mT. From Fig. 3b we can see that the intersection of the acquisition and AF demagnetization curves is at about 30% of the SIRM. According to Cisowski (1981), if the point of intersection is at 50% of the SIRM, there is no magnetic interaction. It is necessary to take into account that the Cisowski (1981) investigation is focused on magnetic interaction for SD particles and his results might not be applicable to the PSD particles present in our samples (Kruiver et al., 1999). From the Henkel plot (Henkel, 1964), IRM_gained vs. IRM_left, the occurrence of magnetic interaction can be verified (Wohlfarth, 1958). Non–linearity in this plot is typically motivated by interparticle dipolar interactions in fine–particle systems (Cisowski, 1981). The behavior observed in the samples of Lake El Trébol (Fig. 3.c) suggests that no magnetic interaction would be present in our samples. Finally, all samples show similar behavior in the stepwise acquisition experiments (Fig. 3d), independent of the absolute IRM intensities. According to Kruiver et al. (1999), the degree of interaction between magnetic particles is dependent on concentration; for this reason, we would have to find different behaviors in samples which have different concentrations. However, all samples deviate to the same extent from the non–interaction line (slope –0.5) in the Henkel plot and the point of intersection in the Cisowski plot (app. 30%) is the same for all samples.

In summary, it is unlikely that magnetic interaction has occurred in the studied sediments. This is supported by the fact that magnetic interaction takes place when the concentration is higher than >0.1 –1 % and the concentration in our sediments varies between 0.01–0.1% (Gogorza et al., 2006).

Age Control

The sediments of lake El Trébol are dated by correlation between the Declination and Inclination curves obtained in Irurzun et al. (2006) and the PSV curves from lake Escondido (Gogorza et al., 2002). The dating is based on visual inspection of the curves (Fig. 10 in Irurzun et al., 2006). On the basis of this correlation, 44 tie points provide age estimates. Ages of the most distinct declination peaks are transferred to the lake El Trébol record and inclination features are matched. Minor changes in inclination and declination are then correlated.

The resultant age scale was tested by three accelerator mass spectrometer (AMS) radiocarbon dates obtained for that work by the AMS Laboratory of the University of Arizona, which were converted into calendar years using the calibration curves of Stuiver and Reimer (1993). The information about each sample, including radiocarbon years before present (RCYBP) and calibrated ages, is listed in Table 1 in Irurzun et al. 2006.

The "pseudo– Thellier" normalization

Fig. 4 shows the pseudo–Thellier results of the sample 176. Fig. 4a shows stepwise AF demagnetization of NRM and acquisition of ARM, with the intensities normalized to their peak values. The best fit slope (m_a) is obtained from the Arai plot (Fig. 4b), the NRM intensity left after demagnetization versus the ARM acquisition intensity (at the same peak field), by using two versions of a jackknife resampling procedure (Efron, 1982). Fig. 4c represents a histogram of the calculated slopes (ma). The solid line represents the best–fit slope; dashed lines point out the upper and lower limits of 90% of possible m_a values.

In one of the estimation procedures (Method 1), we followed the methods of Kok (1998) and Kruiver et al. (1999). The proposed method produced in every sample several regression slope estimates based on combinations of Jackknife subsamples (all, all but one, all but two and all but three, if possible) of field values lying between 20mT and 60mT (a minimum of three points in each subsample was required by the slope formulae) in most of the cases (Kok, 1998). In the other estimation procedure (Method 2), a modified version of that method was used to produce the slope estimates. In method 2, the Jackknife resampling was combined with Theil's complete method of robust–resistant regression slope estimation (Glaister, 2005).

In both methods, the best slope estimate for each sample was identified as the slope using a specific optimizing criterion: for the first one, the slope estimate with minimum relative error (Kok, 1998); for the latter, the slope estimate achieving the minimum sum of absolute deviations between experimental values and the slope estimates.

In addition, two measures of estimation goodness were calculated for each method, to compare the results: i) a 90% bandwidth is obtained, as in Kok (1998), and the percentage of points lying inside that band was given (37.2% applying the robust–resistant estimation method and 8% if the other method is used); ii) the sum of absolute deviations between the line joining best slopes and the experimental values is calculated. Based on both measures, it was observed that Method 1 (robust–resistant estimation method) performed better.

In order to compare the results, the conventional paleointensity records obtained by Gogorza et al. (2006) taking NRM_20mT/ARM_20mT and the pseudo–Thellier paleointensity are shown in Fig. 5a using Method 1 and in Fig 5b using Method 2. With the aim of comparing them, both are normalized by their mean values. We have taken NRM_20mT /ARM_20mT as a conventional paleointensity estimate because the pseudo–Thellier method uses ARM as well (Kruiver et al., 1999).

It is necessary to note that both methods applied to the pseudo–Thellier data provide an estimate of errors in the paleointensity. This error constitutes the information enclosed in the sample about NRM and ARM acquisition (Kruiver et al., 1999).

Results

Relative Paleointensity Results

Here, we have found sufficient similarities between the pseudo–Thellier data and the average NRM_20mT /ARM_20mT profile (according to Brachfeld et al. (2000)) which is not surprising for young sediments unaffected by viscous overprint. The most conspicuous differences occur in features at 2600–3000¹⁴C years BP, 10450–11800¹⁴C years BP and 16800–18350 ¹⁴C years BP, which coincide with prominent dissimilarities in the grain size proxies (Figure 3, Gogorza et al., 2006). Changes in the ratios ARM_100mT/k and ARM_100mT/SIRM; higher ratios indicate smaller grain size and a higher proportion of single–domain (SD) grains (Hunt et al., 1995). If, for example, there is an increase in the proportion of finer grains, the NRM._20mT/ARM _20mT will decrease but the NRM_20mT /SIRM_20mT will increase, which is noticed at about 11800–12800 ¹⁴C years BP; on the other hand, if there is an decrease in the proportion of finer grains, the NRM_20mT/ARM_20mT will increase but the NRM_20mT /SIRM_20mT will decrease, as is observed at about 16800–18100 ¹⁴C years BP. There are other intervals where differences are observed, but further studies would be necessary in those cases (for ex. 2600–3100, 4700–5200, 6700–7500, 13500–14350 and 14400–15400 ¹⁴C years BP). Unfortunately we do not have data in this respect in the pseudo–Thellier profile.

Conclusions

In this paper, we have applied the pseudo–Thellier method to 40 samples from Lake El Trébol and present the results. This technique not only conveniently separates VRM from DRM, but offers a means of assessing the uncertainty of the relative paleointensity estimate (Tauxe et al., 1995). The pseudo–Thellier record and the conventional method of normalization (NRM_20mT/ ARM_20mT) show a reasonable agreement. This similarity would support the conclusion that any technique for relative paleointensity of sediment samples produces reliable results, providing that the sediments meet the criteria for magnetic uniformity necessary for estimates of the relative paleointensity of the geomagnetic field.

The advantage of using a robust–resistant method of regression slope estimation in this type of research should not be surprising, as the influence of outliers, often present due to different kind of perturbations, is minimized.

Bibliography

Bianchi, M. M., J. Masaferro, G. Roman Ross, A. J. Amos and A. Lami, 1999. Late Pleistocene and early Holocene ecological response of Lake El Trébol (Patagonia, Argentina) to environmental changes, Journal of Paleolimnology, 22, 137–148.        [ Links ]

Brachfeld, S. A. and S. A. Banerjee, 2000. A new high–resolution geomagnetic relative paleointensity record for the North American Holocene: A comparison of sedimentary and absolute intensity data, J. Geophys. Res., 105, B1, 821–834.        [ Links ]

Brachfeld, S. A., E. Domack, C. Kisel, C. Laj, A. Lev enter, S. Ishman, R. Gilbert, A. Camerlenghi and L. B. Eglinton, 2003. Holocene history of the Larsen–A Ice Shelf constrained by geomagnetic paleointensity dating, Geology, 31, 9, 749–752.        [ Links ]

Cisowski, S., 1981. Interacting vs. non–interacting single domain behaviour in natural and synthetic samples, Phys. Earth and Planet. Inter. 26, 56–62.        [ Links ]

Efron, B., 1982. The Jackknife, the Bootstrap and other resampling Plans., Society for Industrial and Applied Mathematics, Philadelphia, PA.        [ Links ]

Glaister, P., 2005. Robust Linear Regression Using Theil's Method. J. Chem. Educ., 82, 1472–1474.        [ Links ]

Gogorza, C. S. G., A. M. Sinito, I. Di Tomasso, J. F. Vilas, K. M. Creer and H. Nuñez, 1999. Holocene Secular variation recorded by sediments from Lake Escondido (South Argentina), Earth Planets Space, 51, 93–106.        [ Links ]

Gogorza, C. S. G., A. M. Sinito, I. Di Tomasso, J. F. Vilas, K. M. Creer and H. Nuñez, 2001. Geomagnetic Secular Variations 0–12000 yr. as recorded by sediments from Moreno Lake (South Argentina), J. of South American Earth Sci., 13, 7, 627–645.        [ Links ]

Gogorza, C. S. G., A. M. Sinito, J. M. Lirio, H. Nuñez, M. A. E. Chaparro and J. F. Vilas, 2002. Paleosecular Variations 0–19,000 years recorded by sediments from Escondido Lake (Argentina), Phys. Earth and Planet. Inter., 133, 35–55.        [ Links ]

Gogorza, C. S. G., J. M. Lirio, H. Nuñez, M. A. E. Chaparro, H. R. Bertorello and A. M. Sinito, 2004. Paleointensity studies on Holocene–Pleistocene sediments from Lake Escondido, Argentina, Phys. Earth and Planet. Inter., 145, 219–238.        [ Links ]

Gogorza, C. S. G., M. A. Irurzun, M. A. E. Chaparro, J. M. Lirio, H. Nuñez, P. G. Bercoff and A. M. Sinito, 2006. Relative paleointensity of the geomagnetic field over the last 21,000 years BP from sediment cores, Lake El Trébol (Patagonia, Argentina), Earth Planet and Space, 58, 1323–1332.        [ Links ]

Henkel, O., 1964. Remanenzverhalten und Wechsel–wirkungen in hartmagnetischen Teilchenkollektiven, Phys. Stat. Sol., 7, 919–929.        [ Links ]

Hunt, C. P., S. K. Banerjee, J. Han, P. A. Solheid, E. Oches, W. Sun and T. Liu, 1995. Rock–magnetic proxies of climate change in the loess–paleosol sequences of the western Loess Plateau of China. Geophys. J. Int. 123, 232–244.        [ Links ]

Irurzun, M. A., C. S. G. Gogorza, A. M. Sinito, J. M. Lirio, H. Nuñez and M. A. E. Chaparro, 2006. Paleosecular variations recorded by sediments from Lake El Trébol, Argentina, Phys. Earth and Planet. Inter., 154, 1–17.        [ Links ]

Kok, Y. S., 1998. Relative paleointensities of the Earth's magnetic field derived from deep–sea sediments. Reading the muddy compass (PhD Thesis Utrecht University), Geologica Ultraiectina, 165, pp. 131.        [ Links ]

Kruiver, P. P., Y. S. Kok, M. J. Dekkers, C. G. Langereis and C. Laj, 1999. A pseudo–Thellier relative paleointensity record, and rock magnetic and geochemical parameters in relation to climate during the last 276 kyr. in the Azores region,. Geophys. J. Int., 136, 757–770.        [ Links ]

Nowaczyk, N. R., S. Harwart and M. Melles, 2001. Impact of early diagenesis and bulk particle grain size distribution on estimates of relative geomagnetic palaeointensity variations in sediments from Lama Lake, northern Central Siberia, Geophys. J. Int., 145, 300–306.        [ Links ]

Pan, Y., R. Zhu, J. Shaw, Q. Liu and B. Guo, 2001. Can relative paleointensities be determined from the normalised magnetisation of the wind–blown loess of China?, J. Geophys. Res., 106, B9, 19, 221–19, 232.        [ Links ]

Sagnotti, L., P. Macri, A. Camerlenghi and M. Rebesco, 2001. Environmental magnetism of Antarctic Late Pleistocene sediments and interhemispheric correlation of climatic events, Earth Planet. Sci. Lett., 192, 65–80.        [ Links ]

Stoner, J. S., C. Laj, J. E. T. Channell and C. Kissel, 2002.  South Atlantic and North Atlantic geomagnetic paleointensity stacks (0–80ka): implications for interhemispheric correlation, Quat. Sci. Rev., 21, 1141–1151.        [ Links ]

St–Onge, G., J. S. Stoner and C. Hillaire–Marcel, 2003. Holocene paleomagnetic records from St. Lawrence Estuary, eastern Canada: centennial– to millennial–scale geomagnetic modulation of cosmogenic isotopes, Earth Planet. Sci. Lett., 209, 113–130.        [ Links ]

Tauxe, L., 1993. Sedimentary records of relative paleointensities of the geomagnetic field: theory and practice, Rev. Geophys., 31, 319–354.        [ Links ]

Tauxe, L., T. Pick and Y. S. Kok, 1995. Relative paleointensity in sediments: a pseudo–Thellier approach, Geophys. Res. Lett., 22, 2285–2888.        [ Links ]

Wohlfarth, E. O., 1958. Relations between different modes of acquisition of the remanent magnetisation of ferromagnetic particles, J. Appl. Phys., 29, 595–596.        [ Links ] ]]> 1999 22 22 137-148 2000 105 B1 B1 821-834 2003 31 9 9 749-752 1981 26 26 56-62 1982 2005 82 82 1472-1474 1999 51 51 93-106 2001 13 7 7 627-645 2002 133 133 35-55 2004 145 145 219-238 2006 58 58 1323-1332 1964 7 7 919-929 1995 123 123 232-244 2006 154 154 1-17 1998 165 165 131 1999 136 136 757-770 2001 145 145 300-306 2001 106 B9 B9 232 2001 192 192 65-80 2002 21 21 1141-1151 2003 209 209 113-130 1993 31 31 319-354 1995 22 22 2285-2888 1958 29 29 595-596