Fig. 1: Establishment of a dust storm proxy in the sediments of Lake Gonghai.

a Measured grain-size distributions of 340 samples from core GH09B. b Frequency distribution of the component size of the sediments of core GH09B using the grain-size distribution function method¹⁹. From this analysis, a distinguishable coarse silt component is identified, indicated by the yellow shading. c Measured grain-size distributions of surrounding loess and modern dust storms deposits⁴⁴. d Distributions of loess (yellow shading)¹⁵, desert (orange shading), and Gobi (gray shading) (http://westdc.westgis.ac.cn/) in China. The modern AM limit is shown by the solid green line. Locations of Lake Gonghai (blue dot), the Mu Us Desert (MU), and the trajectories of dust storms (yellow–brown arrows) and the AM (blue arrows) are indicated. e Reconstructed dust storm variations (yellow curve) based on the CSC_13F and its relationship with monitored dust storm frequency (red curve) and estimated dust emission (blue curve)²⁶. Station numbers of dust storms in China are based on ref. ²⁶ and the China National Meteorological Data Service Center (http://data.cma.cn/). The resolution of the grain-size data were transformed to a 1 year/interval using linear interpolation, for comparison with the meteorological and simulated data. The blue shading corresponds to periods of substantial change in reconstructed dust storm activity, and the gray arrows indicate the main trends. The Pearson’s correlation coefficient between the dust storm proxy and meteorological and simulated data is also shown. f–j Zircon U–Pb dating results of bedrock (f), surrounding loess (g), the isolated CSC in lake sediments (h), surface sediments of the Mu Us Desert²⁰ (i), and bulk lake sediments (j). The black lines and shaded area represent probability density and kernel density estimation plots, respectively. Panel d was created using Arcmap 10.2.