Evolution of sedimentation rates
The interpretation of the sliding window plot based on the U data (see Fig. 5) shows that the intensities and positions of the correlated Milanković cycles are not constant in the Lake Van record. Our correlations suggest a gradual evolution in interval I from being dominated by the E cycle in conjunction with weak obliquity to an increase in O at 140.3 mblf. In the lower part of interval II, the E cycle appears diffuse, and the energy begins to increase significantly at 64.9 mblf. E is strong in both intervals, but it is even more distinct in the upper interval II. The precession cycle appears to be of minor importance because it was not detected. However, the additional high amplitudes at the bottom of interval I (λ = 16.5 m ≡ 75 ka cycle) may have been caused by precession (75 ka: multiple of the 19-ka-long P1 cycle).
As mentioned in the “Cyclostratigraphic analysis” section, paleoclimate studies of the Quaternary have shown that the E signal is the dominant Milanković cycle and is particularly pronounced from 900 ka to the present (Berger and Loutre 2010). However, higher-frequency cycles (O and P) have been interpreted in several records from Quaternary terrestrial archives (Berger and Loutre 2010; Bogota-A et al. 2011; Kashiwaya et al. 2010; Prokopenko et al. 2006).
The results for interval I (up to 340 ka) and the recovery of E starting at 225 ka (interval II) confirm that the 100-ka cycle has a strong influence in the Lake Van record. The amplitude of the E cycle prevails over the other signals for the older part (interval I). However, the O cycle appears to be equivalent to the E cycle for the youngest 215 ka.
In general, when fewer waves are contained in a time series, the likelihood for detection of the signal is lower. For interval II (sedimentation rate of 33 cm/ka), the low-frequency 100-ka cycle (λ = 33 m) could only be recorded with a maximum of three times in the sediment section (interval length: 93 m). The O signal appears to have a strong effect as well, particularly in the shallower part of the section. In contrast to E, a maximum of eight O cycles (λ = 13.2 m) can be present in interval II, which could contribute to the higher spectral energy and the equal energy level compared to E.
A disturbed unit was recognized at 185–168 mcblf (the reference depth of the composite profile) and is likely associated with discontinuities in the sedimentary record (Stockhecke et al. 2014a). This might cause the less well-defined spectral peaks in interval I compared to the younger section.
Our evaluation showed that the sedimentation rates increased significantly from 22 cm/ka in interval I to 33 cm/ka in interval II; the change occurred at approximately 120 mblf. The exact depth cannot be determined because the spectral analysis was calculated for a 66-m-long window. The results at a specific depth are therefore affected by the 33 m above and 33 m below the center of the window. However, our results suggest that a change occurred in the sedimentary system and that the intervals can be divided at 120 mblf (282 ka).
The sedimentation rates were used to calculate the time span of the system, though we are aware that ages are needed to determine an absolute chronology. Several aspects of the data preparation process (Fig. 3a) might produce artifacts that affect the calculated sedimentation rates and time span:
-
1.
The procedure was based on a cluster analysis. Tephra layers that are thinner than the minimum bed resolution (<15–20 cm) cannot be resolved. Hence, we cannot preclude the occurrence of thin tephra layers that might affect the spectral analysis and their interpretation. An incomplete removal of the event stratification would lead to an overestimation of the thickness of the lacustrine sediments and hence to an overestimation of the sedimentation rates.
-
2.
Minor portions of the intercalated lacustrine sediments could also be cut out of the record, which might introduce small gaps that would result in a loss of the original sediment thickness and hence an underestimation of the sedimentation rates.
All of the available data should be used to generate a robust age model. Litt et al. (2009) proposed a sedimentation rate of 50 cm/ka for the AR site. Because that estimate was based on short cores (3–9 m overall length), it does not contradict our results of 33 cm/ka for the interval from 120 mblf to the lake floor (interval II). Stockhecke et al. (2014b) determined that the Lake Van record covers a time span of approximately 600 ka. These estimates are based on correlations of TOC trends with marine isotope stages (MIS) and nine 40Ar/39Ar ages (Litt et al. 2014) and thus are based mainly on proxy data, the interpretation of which is ambiguous. The estimated sedimentation rates must be verified with dating methods. A chronology from radiometric dating (e.g., 40Ar/39Ar-dating of the tephra) is pending, and the few preliminary ages from below 130 mblf provide only a rough estimate. The tephra deposits on the land surrounding Lake Van are well dated, but the correlation with the tephra layers in the lake’s stratigraphy is pending (Sumita and Schmincke 2013b). Therefore, further dating is of great importance for verification of the age model.
Nevertheless, the age model of Stockhecke et al. (2014b) shows a change in the sedimentation rates from ~23 cm/ka to ~36 cm/ka at approximately 121–131 mcblf (~270–290 ka). This is consistent with the results of the spectral analysis of the U
res data (Fig. 4) and both age–depth relationships are displayed in Fig. 7. The age–depth model from core analysis (Stockhecke et al. 2014b) was simplified (averaged) for comparability purposes.
Changes in the sedimentation rates in a lacustrine system can be affected by lake-level fluctuations. Regressive trends in conjunction with basinward movement of the lakeshore could have increased the supply of terrestrial material. However, the lake level of Lake Van was likely subjected to repeated changes of up to hundreds of meters (Cukur et al. 2013). Therefore, we do not consider this process as a possible explanation for the distinct change at approximately 282 ka, whereas the average sedimentation rate remained approximately constant during the deposition of interval I (305 ka) and interval II (282 ka).
The change in the sedimentation rates indicates a change in the sedimentary system. Its occurrence coincides with the >9-m-thick tephra layer (V-206; Stockhecke et al. 2014a) with a top depth of 120 mblf (Baumgarten et al. 2014), which suggests a period of very strong volcanic activity. The large amount of tephra deposited in the lake sediments suggests large-scale subaerial deposits in the catchment. This additional terrestrial material could have contributed to an increase in erosion and deposition into the lake by surface runoff.
The cyclic signal with a wavelength of ~4 m (136.7–62.9 mblf; see (Fig. 4) could not be correlated with an orbital signal, and we can only speculate about the cause. The SGR tool was run though the casing, so an attenuation effect caused by joint connections, which have a greater wall thickness, could have caused the signal. However, the drillpipes are approximately 6.24 m long and their connectors are >6 m apart; thus, we can preclude their influence. For this depth section (136.7–62.9 mblf), the removed tephra layers tend to occur at intervals of 4 and 2 m. We propose that the high amplitude (λ = 4 m) might be caused by the data preparation process that was performed on the record, whereas the removal of frequent tephra layers at intervals of 4 and 2 m (multiples that may contribute to the 4 m signal) could have generated this artificial signal. Further, it might have been generated by boundary effects at the transition from tephra to lacustrine deposits. At these boundaries, parts of the tephra layers may still be present and the GR would be increased at their vicinity.
The role of compaction
The compaction and associated reduction in sediment thickness need to be taken into account to estimate the sedimentation rates and perform further calculations. However, the role of compaction of the Lake Van sediments, which were likely influenced by the deposition of thick tephra layers, is difficult to determine. The compaction is expected to differ from “normal” compaction in lake systems, which are controlled by the overburden pressure of pelagic sedimentation. The deposition of large amounts of tephra (e.g., the >9-m-thick tephra layer at 120 mblf) had an impact on the compaction of the underlying sediments. Further, large-scale lake-level changes and associated changes in water depth and hydrostatic pressure have probably influenced the compaction as well.
To estimate the effect of compaction of sediments and calculate the original thickness (decompacted), several parameters are required. Essential are the initial porosity (surface porosity) and the compaction coefficient (Brunet 1998), the latter can be determined by evolution of porosity with increasing depth [Eq. (1)]. The porosity reduction with greater depth depends strongly on sediment characteristics [texture, grain size, grain shape and sorting (Serra and Serra 2003).
Porosity can be achieved by core analysis, e.g., the Archimedean method or nuclear magnetic resonance (NMR) or derivated indirectly from other physical properties as bulk density. The cores from Lake Van were measured before processing (splitting and sampling) by a multi-sensor core logger (MSCL) to estimate bulk density, but the data were not usable because of poor data quality. In general, the disadvantages of physical properties from core analysis are the non-in situ conditions due to relief of pressure, disturbance during drilling and core handling in particular for unconsolidated sediment cores. Therefore, no further approaches to estimate density or porosity were made subsequently.
In situ porosity can be achieved by downhole logging, by either direct measurements of neutron porosity or deviated from other parameters, e.g., bulk density. However, the employment of the density and neutron porosity logging tools from LIAG was not possible because these radioactive tools were not authorized for import into Turkey. After Erickson and Jarrard (1998), porosity can be derived from sonic data (vp). However, because of malfunctions of the sonic tool, vp data could only be recorded partly (section of 80 m; see Baumgarten et al. 2014).
Another approach for estimation of compaction would be the use of standard curves (Athy 1930; Brunet 1998) that provide values for, e.g., silts, which could be used as an approximate for compaction of the lacustrine clayey silts in Lake Van. However, no compaction curves are available for the tephra layers, which are in Lake Van additionally heterogeneous and differ strongly in sorting (fine ash to coarse pumice).
The effect of compaction seems not very strong in Lake Van as indicated by the following: (1) evaluation of the effect on the physical properties (see “Compaction”) and (2) the section of vp data (30–110 mblf; published in Baumgarten et al. (2014), which has mean values of 1,550 m/s and only a minor trend of increasing velocity with depth (approximately 100 m/s per 100 m) was observed.
The correlated orbital cycles (expressed as distinct wavelengths in the spectral analysis) are expected to be affected by compaction. If the effect of compaction was not compensated completely by the exponential regression, additional compaction would have reduced the sediment thickness, the wavelengths in the succession would also have undergone compression, and the sedimentation rates would be underestimated. Thus, the calculated sedimentation rates for interval I and II of 22 and 33 cm/ka, respectively, must be considered as minimum values.
(Sub-) Milanković cycles and small-scale fluctuations in the sedimentation rate
After the correlation of several Milanković cycles and the match between the U data and GIS from NGRIP, which was performed by visual correlation (Fig. 5) as well as objective spectral analysis (Fig. 6), we suggest that a climate signal is present in the U concentration data. The correlation of higher TOC during interglacial and interstadial periods (Stockhecke et al. 2014a) suggests the adsorption of U on organic matter and preservation under anoxic conditions. The current properties of the water column in Lake Van were investigated by Stockhecke et al. (2012), and the bottom water conditions in the deep Tatvan Basin (and AR) are characterized as anoxic.
The detection of Milanković cycles in the U data suggests a strong response of these sediments to glacial versus interglacial conditions. In addition, the U concentration appears to react to the higher-frequency interstadials (Dansgaard–Oeschger events; DO), which suggests strong sensitivity and response of the sedimentary system to changing climate conditions.
Even though sub-Milanković cycles were detected by visual correlation and spectral analysis between NGRIP and U for the time period of 13–75 ka, no indication of higher-frequency signals could be determined by the sliding window method (see Fig. 4).
With the selected window size of 66 m, the first spectrum is allocated at a depth of 33 m (≡38.9 mblf; see “Cyclostratigraphic analysis”). However, a good visual correlation was determined only for the upper part of the record (<75 ka/24.8 m). The lack of higher-frequency signals in the sliding window plot (Fig. 4) supports the observation that these signals are most pronounced in the last 75 ka of the record and that either the signals are absent in the older part or their energy level is too low to be detected by spectral analysis.
The potential of a possible climate signal for all spectral components was suggested (see “The role of spectral gamma ray data for cyclostratigraphic studies”), but the Th and K data do not appear to be suitable for cyclostratigraphic analysis in this case.
The tephra deposits from the area surrounding Lake Van are likely a major source of K-, Th- and U-rich particles into the sediments by surface runoff, which could have masked a (slight) climate signal in the lacustrine sediments. Therefore, we suggest that background sedimentation of volcanic material overprinted the signal from Th to K. However, the spectral analysis results indicate that climate signals were recorded by the U data. The strong mobility of U makes post-depositional relocation likely. Periods of higher production and preservation of organic matter might therefore be indicated by higher concentrations of U. Further investigation of the processes of U enrichment during interglacial/interstadial periods and the apparent lack of climate signals in the Th and K contents could not be performed with our methods.
A reasonable approach to verify the observed cycles would be to directly compare the sedimentary record and the spectra. A repeated sediment succession, which is expressed as a cyclic signal in the spectral analysis, should be evident in the sediments (e.g., cyclic development of clay or carbonate content). However, the sedimentation rates calculated from core data and cyclostratigraphic analysis of downhole data are difficult to compare in the Lake Van sediments. The insufficient core recovery and the use of core material for destructible methods (e.g., pore water analysis) prevent directly linking the cores from hole D at AR to the downhole logs. In addition, the cycles are difficult to detect visually because of the frequent abundance of tephra layers that intersect the lacustrine facies.
The good agreement between the results from the spectral analysis and the interpretation of the cores is promising. Thus, we assume that possible discontinuities (hiatuses, event stratification) in the Lake Van record are of minor significance. As the data show, the major signals can be detected and used for cyclostratigraphic studies.
The minor vertical shifts on the timescale, which were identified by visually matching the NGRIP and the U data, indicate possible fluctuations in the sedimentation rate. Short-term increases or decreases would affect the applied timescale, which is based on an average sedimentation rate of 33 cm/ka. However, the vertical and thus temporal resolution of the downhole data itself and the estimated sedimentation rates, which were calculated using a 66-m-long window (selected by testing), prevent a consideration at greater levels of detail.
Furthermore, the tephra deposits that are up to tens of meters thick (Baumgarten et al. 2014) must have compacted the underlying sediments more than pelagic sedimentation alone. The detected shifts between the U data and δ18O records might also be due to nonlinear compaction, which could have affected the sediment thickness and therefore produced shifts in the timescale.
However, even if the compaction was nonlinear due to irregular compaction caused by the settling of tephra on top of the sediments or strong lake-level changes, it appears to have had little impact on the sediment characteristics because the U record generally correlates well with the NGRIP.