The cosmogenic radionuclide 10 Be is produced by cosmic-ray spallation in Earth's atmosphere. Its production rate is regulated by the geomagnetic field intensity, so that its accumulation rate in aeolian sediments can, in principle, be used to derive high-resolution records of geomagnetic field changes. However, 10 Be atmospheric fallout rate also varies locally depending on rainfall rate. The accumulation rate of 10 Be in sediments is further complicated by overprinting of the geomagnetic and precipitation signals by 10 Be attached to remobilized dust, which fell from the atmosphere at some time in the past.Here, we demonstrate that these signals can be deconvoluted to derive both geomagnetic field intensity and paleoprecipitation records of Asian Monsoon intensity in an 80,000-yr-long 10 Be record from Chinese loess. The strong similarity between our derived paleomagnetic intensity record and the SINT 200 (Guyodo andValet 1996) and NAPIS 75 (Laj et al. 2002) stackedmarine records suggests that this method might be used to produce multimillion-yr-long records of paleomagnetic intensity from loess. This technique also reveals a new method for extracting quantitative paleoprecipitation records from continental interior regions. Our derived precipitation record is broadly similar to the speleothem δ 18 O-based records of paleo-Asian Monsoon intensity from Dongge (Yuan et al. 2004) and Hulu (Wang et al. 2001) caves, and suggests that the paleo-Asian Monsoon intensity may be responding to a combination of both Northern and Southern Hemisphere insolation forcing.

ABSTRACT. Efforts to extract weak geomagnetic excursion signals from Chinese loess-paleosol 10 Be have generally been unsuccessful due to the complexities of its accumulation, because the geomagnetic and climate (precipitation and dust) signals contained in loess-paleosol sequence are tightly overprinted. Here, we present a reconstruction of geomagnetic relative paleointensities for the past 130 kyr from 10 Be records in 2 Chinese loess-paleosol sections using a correction based on the correlation of 10 Be with magnetic susceptibility (SUS) to remove the climatic contamination. Both these records reveal the Laschamp and Blake events, which lie in the loess and paleosol (L 1 SS 1 and S 1 SS 3 ) horizons corresponding to mid-MIS 3 and 5e, respectively. The good agreement between our results and other geomagnetic intensities reconstructions from Atlantic and Pacific sediments indicates that our method is robust. Our study suggests the potential application of loess-paleosol 10 Be for reconstructing geomagnetic intensity variations spanning the whole Quaternary.