<p>Seismic hazard assessment for the NE Tibetan Plateau is of paramount importance because of the growing population density and the accelerated communication and trade activities along the rejuvenated Ancient Silk Road, following the Belt and Road Initiative, and the opening of the high speed railways. Previous-generation seismic hazard assessments were largely based on earthquake catalogues which are shorter than typical earthquake cycles and are temporally and spatially incomplete. This is exacerbated by the fact that magnitudes of many historical Chinese earthquakes are overestimated. In this study, we present new earthquake rate estimates for the NE Tibetan Plateau derived from both an InSAR strain rate map and a re-estimated magnitude of the 1920 Haiyuan Earthquake. First, we obtain a ~100 m resolution strain rate map from five years of Sentinel-1 InSAR data covering an area of 439254 km2 which shows strain concentrated along the Haiyuan and East Kunlun Faults and distributed across the Qilian thrusts and the West Qingling Fault. Second, the magnitude of the Haiyuan Earthquake has been re-estimated to Mw 7.9 &#177; 0.2 using both historical seismograms and offset measurements. Taking the total moment release rate given by the strain rate map and the magnitude of the 1920 Haiyuan Earthquake as the largest magnitude in the Gutenberg-Richter relationship, we generate rate-balancing frequency-magnitude models with different b values and percentages of seismic moment release. Comparing our models against four earthquake catalogues covering different periods and magnitude ranges suggests the following: (1) With a b value of 1 and 75% seismic moment release, the calculated relationship fits well the International Seismological Centre - Global Earthquakes Catalogue (ISC-GEM, 97 years) catalogue in the range Mw>6.5, but overestimates all other catalogues not containing the Haiyuan Earthquake; (2) keeping a b value of 1 and in order to fit the Global Centroid Moment Tensor Catalogue (GCMT, 34 years), the China Earthquake Networks Center Catalogue (CENC, 12 years) and the China Historical Strong Earthquakes Catalogue (CHSEC, 411 years), a low seismic release rate of 30% would be required; the resultant relationship also fits the ISC-GEM catalogue excluding the Haiyuan Earthquake and its aftershocks; (3) to fit all of the catalogues, it is necessary to reduce the b value to 0.7, in which case only 25% aseismic moment release would be required, giving confidence that Mw 7.9 &#177; 0.2 is likely the largest magnitude required to balance the tectonic strain in the NE Tibetan Plateau. This study highlights the dominating strain release by, and the effect on the b value of, the largest earthquake and demonstrates the advantage of combining tectonic strain and earthquake catalogues for seismic hazard assessment.</p>
Magnitude problems in historical earthquake catalogues and their impact on seismic hazard assessment