The precursors before earthquakes are very useful to earthquake prediction, and fluid anomalies before earthquakes are very important to precursory observations. This paper reviews the characteristics of hydrochemical ions and well-aquifer permeability anomalies of the Dazhai observation Well in Pu’er, which is in the Yunnan-Southwestern region of China, for all M≥5.5 earthquakes since 2004. We find that both the chemical ions and physical parameters before the Mojiang M5.9 earthquake exhibited the largest magnitude of changes since observation, and the abnormal state was much stronger than that of previous historical earthquakes, but the magnitude of the earthquake was below 6. About 1.5-2a before the M5.9 Mojiang earthquake, the composition of hydrogen and oxygen isotopes of the water samples in the Dazhai observation Well showed a significant deviation, accompanied by a continuously increasing concentration of fluoride ions from sources at deeper depths. This might suggest that the deep material in the earthquake source area began to be active. At the same time, starting one year before the earthquake, the phase lag of the water level in the wellhole changed from negative to positive, indicating that the source and pathway of well water recharge have been changed. In addition, around half a year before the earthquake, the continuously observed water chemical ions at shallow depths in the wellhole began to show a dramatic change. Moreover, macroscopic anomalies of hot spring water volume increased sharply before the earthquake, showing a remarkable evolution process from deep to shallow, from background to short-term, and from micro anomalies to macro anomalies before the earthquake. To investigate the causes and mechanisms of this phenomenon, we attempt to discuss the abnormal evolution process before the M5.9 Mojiang earthquake from the aspects of regional deep material activity and regional stress level. The abnormal concentration of the hydrochemical ions and the change of aquifer permeability observed continuously at the Dazhai observation well before the M5.9 Mojiang earthquake were caused by the continuous increase in shear stress in the region, which caused the aquifer to be compressed, resulting in a vertical fluid recharge and ultimately the alternation and mixing of different aquifer water bodies. In addition to being controlled by the continuous increase in regional vertical shear stress, the abnormal formation process was also accompanied by the intense activity of deep-sourced chemical elements such as helium isotope and fluoride ion. The abnormal evolution process showed a remarkably coupled process of migration from deep to shallow, which may be the reason why the shallow ion anomaly before the M5.9 Mojiang earthquake was the most significant among all the observed cases. Therefore, the evolution process of fluid activity starting from the deep and continuously transmitting to the surface with the accumulation of regional stress is essential to the abnormal evolution of the hydrological phenomenon before the M5.9 Mojiang earthquake. The regional stress and the process of deep material activity are the biggest differences between the M5.9 Mojiang earthquake and other historical earthquake cases in the study area, which will be the two main factors to be considered when similar ion changes occur again in the future. Our study provides insight into a comprehensive understanding of the predictive significance of underground fluid anomalies in the Dazhai well and the coupled evolution process of deep-shallow fluid anomalies before the earthquake.

Various studies have reported on temporal changes of seismic velocities in the crust before and after earthquake. New time-lapse seismic tomographic scheme based on double-difference tomography can measure the temporal changes of seismic-wave velocities in the Earth and can offer a higher spatial resolution. The result is less affected by different data distribution and quality in different time periods. On October 7, 2014, an M_S6.6 earthquake occurred in Jinggu County, Pu'er City, Yunnan Province, and then on December 6, 2014, two strong aftershocks with magnitude M_S5.8 and M_S5.9 occurred successively. In order to obtain the high spatial resolution P-wave velocity changes in the hypocenter region of the 2014 Jinggu M_S6.6 earthquake, firstly, we used the seismic data in the hypocenter region of the Jinggu earthquake recorded by the Yunnan digital seismic network from January 1, 2008 to December 31, 2017 to invert the high-resolution three-dimensional P-wave velocity structure in the hypocenter region of the Jinggu earthquake by combining the absolute and relative arrival times using the double-difference tomography method. The inversion results show that the aftershock sequence is distributed at the junction between P-wave high-velocity anomaly area and low-velocity anomaly area. This may be the reason why the depth distribution of aftershocks is shallow in NW and deep in SE, and the number of aftershocks decreases fast in NW and slow in SE. The faults that intersect the Lancangjiang Fault are in the low-velocity anomaly zone, so the low velocity anomaly may be related to the fluid in the faults. Then, according to the technical route, this 3D velocity structure was taken as the initial model to invert for the 3D velocity structure of the five periods, and the 3D P-wave velocity structures of the five periods were obtained by using double differential tomography. Finally, the three-dimensional P-wave velocity model of the five periods was taken as the initial model and the new time-lapse tomography was used to obtain the spatial and temporal distribution of the P-wave velocity changes between different periods. In addition, combining the results with the existing geological and geophysical research results, the characteristics and mechanism of P-wave velocity changes are explored, our results indicate that:
(1)The maximum decrease in P-wave velocity at the shallow depth near the epicenter of the main earthquake is 0.2%, which occurred two months after the main earthquake and was caused mainly by rock failure.
(2)There is a P-wave velocity rising zone at a depth of 5km to 15km which is not affected by the rupture of the main earthquake. The existence of this zone caused the P-wave velocity change in the focal area to be discontinuous in depth. It is speculated that the reason for the existence of this zone is that there is a brittle-ductile transition zone with high-strength and high-resistance medium at this depth range. After the occurrence of the M_S5.8 and M_S5.9 aftershocks on December 6, the direction of distribution of aftershocks changed significantly, and also the focal depths showed a deepening trend. The distribution of the two strong aftershocks and their aftershocks were mainly located in the brittle-ductile transition zone, thus affecting the medium within a depth range of 5 to 15km, resulting in decrease of P-wave velocity with a 3.8%decline. It shows that the two strong aftershocks above magnitude 5 have an impact on the brittle-ductile transition zone, and the occurrence characteristics of aftershocks are usually consistent with the characteristics of P-wave velocity change.
(3)About three years after the Jinggu main earthquake, the amplitude of P-wave velocity increase is much larger than that of the previous P-wave velocity drop in the focal area. The P-wave velocity exceeded the pre-earthquake level. This indicates that the area experienced not only a post-earthquake seismic velocity recovery process, but also other physical processes. Combining with the results on strain field change obtained by the GPS data, it is inferred that the significant increase of P-wave velocity in this area is attributed to the superposition between the P-wave velocity increase due to the stress accumulation before the September 8, 2018, Yunnan Mojiang M_S5.9 earthquake and the post-earthquake seismic recovery process. So the P-wave velocity increase in this area is a complex process.

Using the seismic waveform data of Xiaowan seismic network and Yunnan seismic network, we determined the focal mechanisms of 36 earthquakes(M_L ≥ 3.0)from Jun. 2005 to Dec. 2008 and 51 earthquakes(M_L ≥ 2.5)from Jan. 2009 to Dec. 2015 by generalized polarity and amplitude technique.
We inverted tectonic stress field of the Xiaowan reservoir before impounding, using the focal mechanisms of 36 earthquakes(M_L ≥ 3.0)from Jun. 2005 to Dec. 2008 and CAP solutions of 58 earthquakes(M_L ≥ 4.0)collected and the solutions in the Global Centroid Moment Tensor(GCMT)catalog; We inverted local stress field of the reservoir-triggered earthquake clustering area, using 51 earthquakes(M_L ≥ 2.5)from Jan. 2009 to Dec. 2015.
Focal mechanisms statistics show that, the Weixi-Qiaohou Fault is the seismic fault. Focal mechanisms were strike-slip type in initial stage, but normal fault type in later stage. Focal depths statistics of 51 earthquakes(M_L ≥ 2.5)show that, the average value of focal depths in period Ⅰ, period Ⅱ and period Ⅲ are 8.2km, 7.3km and 7.8km respectively and the standard deviations are 4.3km, 3.5km and 6.0km respectively. The average value of focal depths is basically stable in different period, only the standard deviation is slightly different. Therefore, there is not positive connection between focal depth and deviation of focal mechanisms. What's more, there are 2 earthquakes(number 46 and number 47 in Fig.5 and Table 3)with almost the same magnitude, epicenter and focal depth, but they have different faulting types as normal and strike-slip. The focal mechanism of event No.46 is strike:302°, dip:40° and rake:-97° for plane Ⅰ, however, the focal mechanism of event No.47 is strike:292°, dip:82° and rake:140° for plane Ⅰ. Likewise, earthquake of number 3 and number 18 have similar characteristic. Therefore, the obvious focal mechanism difference of similar earthquake pair indicates the complexity of Weixi-Qiaohou Fault.
Considering the quiet-active character of reservoir-triggered earthquakes, we discussed the change of local stress field in different period. The σ₁ of tectonic stress field was in the near-south direction, with a dip angle of 14° before the impoundment, however, the direction of σ₁ of local stress field changed continuously, with the dip angle getting larger after the impoundment. The direction of σ₁ of local stress field of reservoir-triggered earthquake clustering area is close to the strike of Weixi-Qiaohou Fault, and reservoir impoundment increased the shear stress in the fault, so the weakening of fault was beneficial to trigger earthquakes. Comprehensive analysis suggests that fluid permeation and pore pressure diffusion caused by the water impounding, and the weakening of fault caused by local stress field are the key factors to trigger earthquake in the Xiaowan reservoir.

Differently from the existing studies, about 210 days of the original seismic recordings since the Ludian M_S6.5 earthquake are collected from almost all of the nearby stations, and a velocity model and a non-linear location technique are specially selected, in order to relocate the sources of the earthquake sequences. What is more, the same model as used in determining the absolute locations is adopted as the DD technique is used to determine their relative locations. Then the strikes and dips of the seismogenic faults are estimated by linearly fitting the source locations, and finally a new explanation is proposed for the sequence formation. It is shown that the sequence may be divided into 4 sub-areas spatially, each of which corresponds to a nearly vertical fault with but different dimensions and striking azimuths, and that two of them are relatively larger and linked with each other, being the main faults of the sequence, and two others are relatively smaller and separated away from the main faults. These 4 faults, together with the local existing faults, form a radiating-shaped structure reflecting the complicated tectonics, which is very likely to be related with the density variation in lower crust.

We summarized the fluid anomalies associated with the Ludian M_S6.5 earthquake based on the Sichuan and Yunnan earthquake network observations and field survey. The fluid anomalies were divided into long-term, medium-term, short-term, imminent and macroscopic anomalies according to the basic principles of earthquake forecasting. The long-term and medium-term anomalies distributed mainly in the range 300～500km away from the epicenter. By contrast, the short term, imminent and macroscopic anomalies clustered in an epicentral distance less than 100km. The underground fluid anomalies in the higher station density area reflect the enhancement of fluid movement, which are conducive to determine the seismic risk area and trace the short-term precursor of earthquake. The regional stress variations may cause the fractures in a fault zone open and close, leading to the change of water level and temperature in boreholes or spring and emission of deep-sourced gases. It may also lead to intense water-rock reaction and groundwater intrusion, resulting in the change of ion contents in groundwater, or sometimes, the occurrence of significant macroscopic anomalies. Therefore, it is highly possible to obtain reliable earthquake precursor information for predicting the forthcoming earthquake risk zone in the region with dense observation stations.

Since the prediction method for maximum magnitude of reservoir induced seismicity by using comprehensive effecting parameter E-value was proposed in 1987,it has been applied to many largescale reservoirs with good effects.After that,a group of reservoir induced earthquakes were confirmed in the past 20 years.The applicability of the model is worth studying.Based on collection of new cases of reservoir induced earthquakes,48 reservoir induced earthquake cases are selected in the paper. A new empirical prediction model of the maximum magnitude with E-value is statistically obtained and compared with the old one.At last,the upper limit of the maximum magnitude of reservoir induced earthquake is estimated to be less than M7.

Two large earthquakes occurred continuously in the sea northwest of Sumatra,Indonesia,on 26 Dec.2004 and 29 Mar.2005.The observation of water level in Yunnan recorded abundant information about the two earthquakes.This paper presents the water level response to the two earthquakes in Yunnan and makes a preliminary analysis.The present digital water level recording may possibly lose some information,so large earthquake caused abnormal change of water level recorded by analog recording is clearer than that by digital recording.The well water level rise or decline caused by the two large earthquakes respectively was centralized in a small tectonic unit divided by large faults in Yunnan,and the change of water level rise or decline is evidently correlated with geological configuration.Water level rise or decline may be correlated with the change of regional stress field,but it wasn't caused by the change in direction of original stress field.The recent tectonic stress field in Yunnan is mainly divided into three areas.The first area is located east of Xiaojiang Fault;the second located in Sichuan-Yunnan Rhomboid Block,west of Xiaojiang Fault,north of Honghe Fault and east of Jinshanjiang fault;and the third located west of Honghe Fault.Water level rise or decline areas caused by Indonesia large earthquakes are not consistent with subareas according to direction of original stress field.The direction of stress field west of Lancangjiang Fault is basically same,but the change of water level caused by Indonesia large earthquake is divided into two sections by Nandinghe Fault,the north rises and the south declines.We think the reason for water level rise or decline is possibly that the seismic wave changes the stress state of the tectonic unit.Abnormal change of water level caused by large earthquake is mainly attributed to well water bearing bed responding to seismic wave.The change is mainly elastic deformation,so the records are mostly water wave-like as seismic wave.Because of difference of hydrogeologic conditions and effect of regional stress field and seismic wave additive stress field,the water bearing bed appeared to be in a stress adjustment stage after earthquake in a block or a fault.So in some wells,changes of water level rise or decline appeared,and the changes are also correlated with geological configuration.The two Indonesia large earthquakes occurred along the Sumatra Fault,its focal mechanism is similar to reverse fault.The shape of recordings of water level response to the two earthquakes in a same well is very similar except amplitude.So we inferred the water level response mode in same well is completely same for earthquake occurring on same fault and its fracture mode is similar,the difference is just that the response amplitude increases with the growth of magnitude.

The dynamic pattern of moderate-strong earthquakes in Yunnan region has been traced in this paper. It is discovered that a sequential migration of moderate-strong earthquakes occurs 1～4 years before the active period of strong earthquakes in Yunnan region; the sequentially migrated moderate-strong earthquakes indicate the general layout of M ≥6.7 strong earthquakes during the active period, while about 80% of the strong earthquakes occur within a range of 150km from the epicenters of these moderate-strong earthquakes. The first strong earthquake of the active period occurs within 2 days to 11 months after the termination of sequential migration of moderate-strong earthquakes, and about 75% of the first strong earthquakes occur within 3 months. During the active periods Ⅰ and Ⅲ in Yunnan region, 8 of 10 strong earthquakes occur within the range of 190km from the apexes of tetragon formed by linking the sequentially migrated epicenters of moderate-strong earthquakes(such as the epicenters of Huaning, Ludian, Jianchuan, Longling, Qiaojia, and Jinning earthquakes, as well as those of the Jiangcheng, Ruili, Zhongdian, Yongshan, and Jinghong earthquakes), while about 60% of the strong earthquakes occur within a range of 50～90km from the epicenter locations of moderate-strong earthquakes. The M 7.7 Tonghai earthquake of January 5, 1970 occurred on the eastern side of the tetragon mentioned above, showing an equidistant layout of 230km from Yongshan in the north to Dongchuan, Tonghai and Jiangcheng in the south. The relationship between sequential migration of moderate-strong earthquake and the occurrence of strong earthquake for the active periods Ⅱ and Ⅳ has a little difference from that for the periods Ⅰ and Ⅲ. Sequential migrations of moderate-strong earthquakes from Lancang to Lufeng, Shiping and Tengchong, as well as from Shidian to Ninglang, Huidong and Lancang make up tetragons, while the strong earthquakes occur at the starting or ending positions of the migration, i.e. within a range of 140km from both ends of the diagonal lines of the tetragons, such as Lancang and Lufeng, as well as Lancang and Ninglang. About 75% of the strong earthquakes occur within a range of 100km from the epicenters of moderate-strong earthquakes. The strong earthquakes densely occur in the southwest Yunnan region, where is the starting or ending places of sequential migrations of moderate-strong earthquakes. The Tonghai and Longling earthquakes of magnitude 7 did not occur until the active period Ⅲ, located at the other ends of the diagonal line, 50km from Shiping and Yuxi earthquake epicenters, and 150km from Tengchong earthquake epicenter that occurred before the active period Ⅱ. The starting moderate-strong earthquakes of the migration sequence during the active periods Ⅰ and Ⅲ were M 6.0 double shocks, which are rarely seen in Yunnan. The ending moderate-strong earthquake occurred within a range of 50～70km from the epicenter of a certain moderate-strong earthquake from the migration sequence. The starting moderate-strong earthquake before the active period Ⅱ was M 6.0, and both the starting and ending moderate-strong earthquakes before the active period Ⅳ were all M 5.0 double shocks, possessing a certain peculiarity. Some earthquake examples, such as equidistant migration of strong earthquakes, sequential migration of moderate-strong earthquakes during the later stage of the active period, sequential migration before and after a great earthquake and oriented migration of moderate-strong earthquakes are also introduced in this paper. However, further study is needed for understanding the mechanism of sequential migration of seismicity.