No abstract

No abstract

Waxing and waning ice sheets and mountain glaciers have been a defining feature of the climate oscillations during the Pleistocene ice ages. Ice masses have shaped extensive tracts of Earth's surface, modified oceanic and atmospheric circulation (Toucanne et al., 2015), and stored large volumes of water periodically, driving variations in eustatic sea level (Lambeck et al., 2014;Simms et al., 2019). Significant research efforts have been directed to constraining the Pleistocene extents and chronology of the major existing and ephemeral ice sheets (Batchelor et al., 2019) together with some of the smaller ice sheets and mountain ice masses (e.g., Ivy-Ochs et al. (2008); Licciardi and Pierce (2018)). However, substantial gaps in our knowledge persist regarding the cryosphere and its history. Most notably, we are still unable to match eustatic sea level with reconstructed ice volumes during the Last Glacial Maximum (LGM, Simms et al. ( 2019)), let alone for older glacial cycles. One region in need of attention is Northeast Siberia, east of the Lena River (Figure 1), where the prevailing view is that the western Eurasian Ice Sheet effectively blocked Atlantic moisture sources from reaching Siberia, promoting an extreme continental climate inimical to the growth of major ice sheets (Krinner et al., 2011;Siegert & Marsiat, 2001). And yet, the extensive mountain landscapes of this region display over one million km 2 of formerly glaciated terrain (Barr & Clark, 2012). Glaciation of this extent could account for several meters of global sea-level equivalent, but the timing of any such ice mass is completely unresolved. Here, we aim to provide a chronological record of past glaciation in a key area of this vast, understudied region.

The Ebro Basin constitutes the central part of the southern foreland of the Pyrenees. It was endorheic during the Cenozoic and accumulated sediments. By the end of the Miocene, erosion and river incision reconnected the basin to the Mediterranean Sea, establishing a post-opening drainage network. Those rivers left terraces that we study in this work. We first synthesize previous works on river terraces that are widely dispersed in the basin. We provide new age constraints, up to 3 Ma, obtained thanks to cosmogenic nuclides using both profile and burial methods. We derive a unified fluvial terrace chronology and a homogenized map of the highest terraces over the entire Ebro Basin. The dated terraces labeled A, B, C, D, and E are dated to 2.8 ± 0.7 Ma, 1.15 ± 0.15 Ma, 850 ± 70 ka, 650 ± 130 ka, and 400 ± 120 ka, respectively. The chronology proposed here is similar to other sequences of river terraces dated in the Iberian Peninsula, around the Pyrenees, and elsewhere in Europe. The oldest terraces (A, B, C) are extensive, indicating they form a mobile fluvial network while from D to present, the network was stable and entrenched in 100 to 200 m-deep valleys. The transition from mobile to fixed fluvial network is likely to have occurred during the Middle Pleistocene Transition (MPT, between 0.7 and 1.3 Ma), when long-period/high-intensity climate fluctuations were established in Europe. We estimate that between 2.8–1.15 Ma and present, the incision rates have tripled.

<p>Two published cosmogenic radionuclide (CRN) <sup>26</sup>Al/<sup>10</sup>Be burial age calculation methods developed to correct for post-depositional production of nuclides in settings with low sediment overburden are compared. The advantages and limitations of simple (ISO; [1], [2]) and inverse modelling (INV, [3]) isochrons are investigated.</p> <p>The studied dataset originates from the gravel of a Danube terrace in the Central Vienna Basin (Austria) [4]., where two horizons (5.5 m and at 11.8 m subsurface depth) were sampled. Each sample set contained 6 quartz or quartzite cobbles.</p> <p>The advantage of ISO is that it is uninfluenced by changes in sample depth over time. However, the initial <sup>26</sup>Al/<sup>10</sup>Be ratio is fixed and no pre- and post-burial denudation rates can be calculated. In addition to age, INV models source and sink denudation rates, but assumes constant depth over burial time.</p> <p>For correct application of ISO and INV outliers, must be excluded. The robustness of both methods is tested by systematically including or excluding data points (bootstrapping) to estimate the dependence of numerical ages on sample selection either in the field, or during outlier identification.</p> <p>For outlier identification the traditional method of data exclusion of points above or below the isochron line is used. In addition, a new way is introduced here: the post-burial production is calculated using the modelled burial age and denudation rate and compared to the measured inventories of <sup>10</sup>Be and <sup>26</sup>Al. If the fraction of post-burial production is equal or higher compared to the measured inventory and its ratio is considerably different for the two isotopes from the same sample, the datapoint is invalid.</p> <p>In addition, the influence of each sample on the modelled burial age, tested by bootsrapping, is used to exclude samples with a large effect on the age.</p> <p>The resulting ages at both levels using ISO and INV agree within errors with ISO being systematically slightly younger. The importance of outlier removal is stressed by the fact that inclusion of all samples results in a considerably older age of the stratigraphically higher level compared to the underlying one.   When outliers are excluded, burial ages of the two sampled horizons overlap within uncertainty, suggesting one single deposition event for the whole sediment package.</p> <p>Interestingly, when the entire dataset is merged, both methods provide similar ages regardless of the outliers being excluded or kept in. This demonstrates that a larger sample number increases the robustness of a dataset considerably and decreases the sensitivity of either method to potential outliers.</p> <p>In summary, both ISO and INV are robust ways of CRN burial age determination, provided that model presumptions are not violated and outliers are excluded or the sample number large enough to overprint the influence of outliers.</p> <p>Funding: NKFIH FK124807; OMAA 90ou17; OMAA 98ou17.</p> <p> </p> <p>References</p> <p>[1] Balco, G., Rovey, C.W., 2008. American Journal of Science 308(10), 1083-1114.</p> <p>[2] Erlanger, E.D., et al., 2012. Geology 40(11), 1019-1022.</p> <p>[3] Pappu, S. et al., 2011. Science, 331(6024), 1596-1599.</p> <p>[4] Ruszkiczay-Rüdiger, Zs. et al., 2021. Journal of Radioanalytical and Nuclear Chemistry, 329(3), 1523-1536.</p>