Abstract:Ningxia and its neighboring area are located in two blocks of the northeastern margin of the Qingzang-Tibet plateau:the Erdos block and the Alashan block.The geological structure in this area of the plateau is known to be unstable.Over recorded time,many large or great earthquakes have occurred in this area.Based on previous research results,we employed two different earthquake location methods,i.e.,double-difference and genetic algorithm,to recompute the location of small earthquakes that were recorded by the Ningxia Regional Earthquake Monitoring Network from 1991 to May 2014.This process consisted of first employing the theoretical travel time method to delete or modify the seismic events whose arrival time error exceeded 10 s.Next,we selected seismic events that contained S and P waves recorded by at least three stations.In addition,Pg,Sg,Pn,and Sn seismic phases are included in each earthquake event.The velocity model was obtained from four seismic sounding profiles that crossed the research area.In addition,a horizontally layered velocity model was used.The double-difference method results showed that the seismic source depth was not so different from that obtained using the genetic algorithm method.But only a few earthquake events (approximately 25% of the total 4 688 seismic events) can meet the relocation requirement in the double-difference method.Therefore,the genetic algorithm method can be used to recalculate and reanalyze the depth characteristics of small earthquakes and their relation to the deep fault in the research area.Relocation results showed that the seismic zone distribution by depth is more apparent.Statistical analysis,taking the relocation calculation into account,showed that more earthquakes happened from 5 to 30 km.The average depth was 18 km.The space distribution of small earthquakes in this area was obtained.To facilitate the analysis of the relationship between the occurrence of earthquakes and faults in this region,we divided the geographic research region into four districts on the basis of their tectonic characteristics and concentration region of small earthquakes.These districts were named A,B,C,and D. In each district,the seismic profile perpendicular to fault was obtained.In district A,the results showed that the seismic depth was mainly distributed from 5 to 25 km.The earthquake distribution width was 50 to 60 km.In district A,earthquakes occurred primarily in the upper and middle crust or in the top of the lower crust.It was postulated that fractures parted only the upper or middle crust while not the Moho.In district B,the small earthquake distribution was primarily in the north-south direction.In this district,the earthquakes were primarily distributed from 5 to 35 km.The width of the epicenter was approximately 50 km.The dip of the seismic zone was from SE or NE,with an angle of approximately 50°.The fault fracture was primarily in the middle or lower crust.In district C,small earthquakes occurred along fault F10 and F11.The epicenter width was approximately 70 km.The seismic source was mainly ≤22 km.The dip of the seismic zone was to the NE,and the dip angle was approximately 40°.The fault fracture was confined to the middle crust.In district D,the depth range of the seismic source was primarily from 0 to 25 km.The dip of the seismic zone was to the NEE,and the dip angle was approximately 50°.But the dip of the Haiyuan fault was to the SW,which is different from the results reported in previous studies.The geologic structure and the fault activity of this district are more complex.These results imply that the fracture depth of fault F9 and F14 is deep to at least the middle crust. The genetic algorithm calculation results revealed that the depth distribution of a large number of reported earthquakes is denser and that the epicenter of the seismic distribution is along the faults that resulted in the great earthquakes.The depths of the deep faults extended to the middle or lower crust.