Understanding historical earthquakes by mapping coseismic landslides in the Loess Plateau, northwest China

Locating the epicenters and quantifying the magnitudes of historical earthquakes are important, yet difficult, tasks. These tasks often entail estimating seismic intensities based on written documentation, which suffers from biases and uncertainties that are difficult to evaluate. However, past strong earthquakes may trigger numerous landslides that remain in the landscape to the present day, whose number and distribution are correlated with the magnitude and location of the earthquakes. Thus, mapping historical coseismic landslides may provide a useful tool for re-evaluating the magnitudes of historical earthquakes. We use remote sensing images to map landslides in the southern Loess Plateau, China. We suggested that c. 5000 preserved landslides are related to the 734 ce Tianshui earthquake. These landslides are densely distributed along a c. 70-km-long section of the West Qinling Fault. Based on the assumption that the length of the zone in which substantial landslides occurred is equal to the fault rupture length, we estimated a M-w of 7.2 for the 734 ce Tianshui event, which is similar to previous estimates (M-w = 6.8-7.5). We confirmed that the 1920 ce Haiyuan, 1718 ce Tongwei, and 1654 ce Lixian earthquakes did not contribute to the historical landslides observed in the study area by combination of the mapping and Chinese related literature. We estimated an M-w of 7.0 for the 1654 ce Lixian earthquake, which is lower than previous estimates (M-w = 8.0). We suggest that coseismic landslide of medium and large sizes with areas > 10(4) m(2) can be used as a criterion to locating and quantifying historical earthquakes, thereby reducing uncertainties in the estimated magnitudes of historical earthquakes that lack instrument records.