降尘^(10)Be浓度对黄土^(10)Be示踪地磁场变化研究的影响

Influence of variation in dust ^(10)Be concentration amplitude on the reliability of geomagnetic tracing study from Chinese loess

运用我国黄土地层中的^(10)Be记录来示踪地磁场漂移事件和重建相对强度变化历史在最近几年取得了显著进展。开展上述研究的关键点之一在于如何对黄土^(10)Be记录中包含的源区降尘^(10)Be浓度等信号的相对贡献进行合理估算。本文参考最新的现代粉尘^(10)Be浓度观测数据,设计了3种改变降尘^(10)Be浓度的情景分别建立了130 ka以来黄土^(10)Be记录的地磁场漂移事件时间序列。综合对比表明,3种情景下的重建结果并无差异,说明源区降尘^(10)Be浓度幅值变化不会对^(10)Be示踪地磁场变化的结果产生显著影响。同时分离的130 ka以来受地磁场变化影响的^(10)Be记录可与PISO-1500等全球典型地磁场强度曲线良好对比,并完整地记录了蒙诺湖(Mono Lake)、拉尚(Laschamp)、挪威海-格陵兰(Norweigian-Greenland sea event)和布莱克(Blake)等主要地磁场漂移事件,揭示了黄土^(10)Be重建的地磁场变化曲线的全球尺度意义并印证了中国黄土在千年尺度的连续性问题。

Background, aim, and scope Studies on geomagnetic excursion tracing and relative paleo-intensity reconstruction using Chinese loess ^10Be proxy have greatly progressed in recent years. In these studies, one of the key points lies in how to reliably estimate the relative contribution from the dust source ^10Be concentration as well as other ^10Be concentration fractions contained in loess. To evaluate the possible influence of variation in dust ^10Be concentration amplitude on the reliability of related geomagnetic tracing study, 3 kinds of separation models are designed by gradually changing the dust ^10Be concentration amplitude as well the dust source magnetic susceptibility amount, to respectively reconstruct the geomagnetic excursion events series using the Chinese loess ^10Be record since the 130 ka. Materials and methods The ^10Be concentration and other climatic proxies such as the magnetic susceptibility are measured from a 1490 cm core of the Luochuan section （35°45′N, 109°25′E） in central Chinese Loess Plateau, which was described in Zhou et al （2010）. To present the 3 different separation models, the observation studies on the modern falling dust ^10Be in Northern China by Xian et al （2012） is also cited to gradually change the dust ^10Be concentration for calculations. The basic step for the test calculation is to firstly separate the dust effects on the ^10Be and magnetic susceptibility in loess, by subtracting a constant value from the total measured ^10Be and magnetic susceptibility respectively. Then a linear regression between the dust free ^10Be concentration and pedogenic susceptibility is developed to separate the Monsoon rainfall effects on loess ^10Be records, and to derive the residual ^10Be signals which is dominantly modulated by geomagnetic field intensity. As for the separation model 1 and model 2, the same amount of dust source magnetic susceptibility （SUS（D）） is assumed, and the dust ^10Be concentration （Be（D）） is the variable to be allowed to change with time. While this feature is totally different in model 3, which the amplitude of SUS（D） and Be（D） are allowed to simultaneously change. All of the aforementioned calculations are based on a ＂Quasi-constant＂ distribution assumption of dust source ^10Be and magnetic susceptibility records. Results The derived geomagnetic excursion events series from the 3 different models show high similarity both in trend and details, with the correlation coefficient r is about 0.99. In addition, the final derived 130 ka geomagnetic excursion sequence using loess ^10Be is well comparable with that of the globally stacked PISO-1500 relative paleo-intensity curve and the normalized relative paleo-intensity record from Pacific sediments, which clearly reveals ahnost all the well dated geomagnetic excursion events known as the Mono Lake, Laschamp, Norweigian Greenland sea, and the Blake events over the last 130 ka. In addition, the derived 130 ka geomagnetic excursion sequence also records the 2 abnormal changes in l-Be concentration at 12 ka and 60 ka, which can be correlate to the Gothenburg flip and the 60 ka peak derived from the ^10Be measurements on sediments in North Atlantic respectively. The loess ^10Be record clearly shows that the 60 ka peak is well comparable with the Norweigian Greenland sea event, although it is not globally named as a typical excursion change in the last Glacial-Interglacial cycle. Discussion The well comparison of the loess ^10Be resulting geomagnetic excursion events series with that of the paleo-intensity records from marine sediments also confirms the significance of global criterion of the geomagnetic tracing studies using Chinese loess ^10Be records. Considering the short-lived feature （〈104 years） of geomagnetic excursion changes, our present study further strongly supports the idea of the continuity of Chinese loess deposition on the millennial scale at least. Conclusions The 130 ka geomagnetic excursion events records from loess ^10Be using 3 different models are roughly the same, which suggests that the reliability of geomagnetic tracing studies from Chinese loess ^10Be is roughly independent on the variation in the amplitude of dust ^10Be concentration. If the ^10Be resulting record of geomagnetic excursion events in loess is robust, it will provide key time markers to loess chronology and synthesize the terrestrial and marine climatic records. Recommendations and perspectives The present work is the preliminary test based on the most ideal situations of the ＂Quasi-constant＂ distribution assumption of dust source --Be and magnetic susceptibility records in Chinese loess. If the possible alternations of loess dust source with different time are considered, this ＂Quasi-constant＂ distribution assumption may need to be clarified by the investigations on surface soil along a transect section from the source region to the Loess Plateau. On the other hand, the atmospheric ^10Be deposition has been proved to be closely associated with local rainfall amount, therefore the rainfall changes are very likely to show potential influence on the reliability of environmental tracing using ^10Be. Especially, when the ^10Be tracing studies are extended to the arid region in western China, where the magnetic susceptibility is not a robust precipitation tracer, future study should be firstly focused on how to select the alternative sensitive proxy, to establish the robust separation model for geomagnetic field changes tracing using loess ^10Be records.