Hydrogeochemical Characteristic of Geothermal Water and Precursory Anomalies along the Xianshuihe Fault Zone, Southwestern China

: Hydrogeochemical changes in association with earthquakes are considered as a potential means of identifying earthquake precursors. The Xianshuihe fault zone (XSHF) is considered one of the most active seismic fault zones in China; 43 hot springs were sampled and analysed in the laboratory for major elements, silica, stable isotopes ( δ D and δ 18 O) and strontium isotopes were investigated from 2008 to 2021. The meteoric water acted as the primary water source of the hot spring in the XSHF, and recharged elevations ranged from 1.9 to 4.8 km. The geothermometers method was used to estimate the region of thermal storage temperature and its temperature between 8 and 142 ◦ C. And the circulation depth ranged from 0.1 to 6.9 km. Most of the hot spring water was immature water with a weak degree of water-rock reaction. However, the degree of water-rock reaction and the depth of hot spring water circulation were high in part of the Kangding and Daufu segments, which also had the highest reservoir temperature and the most frequent strong earthquakes. Temporal variations of hydrogeochemical showed that Na + , Cl − and SO 42 − decreased obviously following the 12 May 2008 Wenchuan M s8.0 and existed abnormal value ﬂuctuations from the 20 April 2013 Lushan M s7.0 to 22 November 2014 Kangding M s6.3 occurred and after 20 July 2017 returned to the normal levels. And the ion concentrations in hot springs increased by 5% to 35% three months before 22 November 2014 Kangding M s6.3 with the obvious precursor anomaly. Hydrogeochemical anomalies could be useful for predicting an earthquake in the study area.


Introduction
Since the 1960s, there have been extensive reports of changes in groundwater chemistry before and after the earthquake [1,2], including the changes in concentrations of Na, Si and Ca, radon count rates and stable isotope ratios [3][4][5][6].The precursor changes reported from the literature range from 1 day to 6 months and from 5 to 400 km over time and length scales, respectively [7][8][9][10].Tsunogai and Wakita (1995) [11] found obvious changes in groundwater flow and ion concentrations (Cl − and SO 4 2− ) changes were observed at the two monitoring stations around the Kobe earthquake.Skelton et al. (2019) [9] predicted earthquake-related hydrochemical changes before and after earthquakes based on longterm groundwater chemistry (Na + and Ca 2+ ) and isotopic data (δD and δ 18 O).Based on a large-scale stable isotope dataset, Hosono et al. (2020) [12] showed improved permeability after the 2016 Mw7.0 Kumamoto earthquake.
The XSHF is one of the primary left-lateral strike-slip faults originating from Tibet.It crosses the entire lithosphere and cut into the upper mantle [13].As one of the most dynamic faults in the world.It is similar to the San Andreas fault, where at least 10 earthquakes with a magnitude greater than 7 have occurred along 350 km segments of the XSHF [14].Chen et al. (2014) [15] studied the changes of dissolved ion concentrations in hot springs in western Sichuan during the 12 May 2008 Wenchuan Ms8.0 earthquake and 20 April 2013 Lushan Ms7.0 earthquake.Li et al. (2019) [16] showed significant changes in water chemistry in the Erdaoqiao (EDQ), Longtougou (LTG) and Guanding (GD) hot springs in the Kangding geothermal area during the 2008 Ms8.0 Wenchuan earthquake, 2013 Ms7.0 Lushan earthquake and 2014 Ms6.4 Kangding earthquake.Zhang et al. (2021) [17] researched the temporal variations in stable isotope ratios at the EDQ, LTG and GD hot springs in the Kangding geothermal area experienced obvious changes before and after the 2019 Ms6.0 Changning earthquake.Previous studies have mostly focused on the Kangding region, and systematic comparative studies on the distribution of other regions on the XSHF are still lacking.
This paper aims to analyse the elemental composition and isotopic characteristics of thermal springs distributed in the XSHF.We analysed physicochemical parameters of groundwater, such as the temperature, flow rate, dissolved hydrogen, oxygen and ion concentrations which may capture seismic anomalies before and after to explore the relationship between changes in the hydrogeochemical characteristic along the XSHF in southwestern China.It is critical for identifying earthquake precursors and learning about water-rock interactions in fault zones.

Geological Setting
Due to the continuing clash between the Eurasian and Indian plates, the Tibetan Plateau squeezed southeastward and was blocked by the Sichuan Basin and the South China Block [18,19].An active fault belt was formed with near-N-S oriented strike-slip features, comprising the XSHF, Anninghe fault zone (ANHF), Zemuhe fault zone (ZMHF) and Xiaojiang fault zone (XJF) from north to south (1) [20].The XSHF is the most active fault in the left-slip fault zone on the eastern edge of the Qinghai-Tibet Plateau [21].The activation of the XSHF began in the Early Permian, and during the Permian-Late Triassic, a large number of mixed rock systems, large-scale basal magmatic overflows, and deep-sea tensional rift deposits represented by radiolarian siliciclastic rocks developed within the fault zone, marking the formation of the initial oceanic crust [22].The XFZ is composed of northwestern, middle and southeastern segments.The northwestern segment includes the Luhuo, Daofu and Bamei faults.The middle segment consists of the Yalahe, Zheduotang and Selaha faults which are called the Kangding section here.The southeastern segment contains the Moxi faults [23].The slip rate of the Zhuwo, Luhuo, Daofu, Bami, Yalahe, Selaha, Zheduotang and Moxi segments were about 4 mm•a-1, 13 mm•a-1, 13 mm•a-1, 12 mm•a-1, 4.0 mm•a-1, 7.0 mm•a-1, 6.5 mm•a-1 and 9.5 mm•a-1, respectively [14].The geographical of the XSHF were high mountains with low valleys.The elevations of the XSHF ranged from 3000 to 7556 m [16].
The regional stratigraphy mainly includes Triassic rocks consisting of deformed metamorphic feldspathic quartz sandstone, tuffaceous sandstone, siltstone and mudstone on both sides of the Chuanxi Depression, and Precambrian and Paleozoic rocks consisting of shallow metamorphic sandstone, slate, limestone lenses, limestone and metamorphic quartz sandstone on the east side of the rift zone.The XSHF is widely distributed with late Mesozoic-Cenozoic granite and diorite intrusions [24].A larger number of geothermal springs has been investigated in the middle and southeastern segments of the XSHF, the sources of which include the deep magma heat, radioactive heat from granitoids and strike-slip frictional heat [25].
Numerous earthquakes, including the 2008 Wenchuan Ms8.0 earthquake, the 2013 Lushang Ms7.0 earthquake and the 2014 M s 6.4 Kangding earthquake, have occurred in the Longmenshan Fault Zone near the XSHF (Figure 1) [14].The Wenchuan Ms8.0 earthquake occurred on 12 May 2008 in Wenchuan, Sichuan Province.It has caused a 285 km surface rupture belt which includes the Yingxiu-Beichuan, Guanxian-Anxian and Qingchuan faults, with a vertical offset of up to 6.2 m from the surface.The Lushan earthquake occurred ~85 km southwest of the Wenchuan earthquake on 20 April 2013, with a rupture length of approximately 66.5 km along the strike of the Longmenshan Fault Zone [26].These two large earthquakes caused significant hydrochemical variation of thermal springs in this study area [16].
Water 2022, 14, 550 3 of 21 Numerous earthquakes, including the 2008 Wenchuan MS 8.0 earthquake, the 2013 Lushang MS 7.0 earthquake and the 2014 MS 6.4 Kangding earthquake, have occurred in the Longmenshan Fault Zone near the XSHF (Figure 1) [14].The Wenchuan MS 8.0 earthquake occurred on 12 May 2008 in Wenchuan, Sichuan Province.It has caused a 285 km surface rupture belt which includes the Yingxiu-Beichuan, Guanxian-Anxian and Qingchuan faults, with a vertical offset of up to 6.2 m from the surface.The Lushan earthquake occurred ~85 km southwest of the Wenchuan earthquake on 20 April 2013, with a rupture length of approximately 66.5 km along the strike of the Longmenshan Fault Zone [26].These two large earthquakes caused significant hydrochemical variation of thermal springs in this study area [16].

Sampling and Methods
The groundwater samples were obtained from 43 1 and 2 and Supplementary Table S1).Geochemical analysis of hot water sampled every 3 days taken from five of these springs during 2019 and 2021, but some of these data from 2008 to 2013 are based on other scholars (Chen and Zhang) [15,17] in the study area.

Sampling and Methods
The groundwater samples were obtained from 43 2 and Supplementary Table S1).Geochemical analysis of hot water sampled every 3 days taken from five of these springs during 2019 and 2021, but some of these data from 2008 to 2013 are based on other scholars (Chen and Zhang) [15,17] in the study area.
At each site, four samples were collected after filtration through a 0.45 µm membrane to analyse the water samples for major element concentrations, the isotopes of hydrogen and oxygen, SiO 2 concentration and strontium isotopes.The water temperatures were measured in the field with a digital thermometer with an accuracy of 0.1 • C. The conductivity and pH, which are unstable parameters, were measured in situ with handheld meters calibrated prior to sampling.The concentrations of cations (K + , Na + , Mg 2+ and Ca 2+ ) and anions (Cl − , Br − , NO 3 − and SO 4 2− ) were determined with a Dionex ICS-900 ion chromatograph and an AS40 automatic with a ±5% or less reproducibility and a 0.01 mg/L detection limit [27] sampler from the Earthquake Forecasting Key Lab of China Earthquake Administration.We determined HCO 3 − and CO 3 2− concentrations using a ZDJ-100 potentiometer titrator procedure with titrator procedures of 0.05 mol/L HCl with 0.1% methyl orange and 1% phenolphthalein (reproducibility within ±2%).The ionic charge equilibrium defined in meq/L is expressed as (cations − anions)/(cations + anions) and is within ±5% of the ionic equilibrium for each sample [28].The hydrogen and oxygen isotopes were analysed using a Finnigan MAT253 mass spectrometer with the TC/EA method.The isotope accuracies of V-SMOW and analysed water samples were ±0.2% and ±1%, respectively [29].The SiO 2 concentration of the samples was analysed using an inductively coupled plasma emission spectrometer Optima-5300 DV [30].Sr elements were measured by Element XR ICP-MS from the Test Center of the Research Institute of Uranium Geology [31].

Recharge Sources of Hot Springs
The comparison of δ 18 O and δD of water samples with global and regional atmospheric precipitation lines provided insight into the hydrological cycle of water sources and recharge, water-rock interaction and groundwater mixing [32].The oxygen and hydrogen isotopic of water samples from the XSHF varied from −129.80 ‰ to −78.77 ‰, and from −21.70 ‰ to −10.04 ‰.The Global Meteoric Water Line (GMWL) equation was found to be δD = 8δ 18  and δD distribution maps showed that most of the hot springs in the study area were in close proximity to the GMWL and LMWL (Figure 2).This provided important information about the source of geothermal water, which is that they were recharged by infiltrating atmospheric precipitation.
atmospheric precipitation lines provided insight into the hydrological cycle of water sources and recharge, water-rock interaction and groundwater mixing [32].The oxygen and hydrogen isotopic of water samples from the XSHF varied from −129.80 ‰ to −78.77 ‰, and from −21.70 ‰ to −10.04 ‰.The Global Meteoric Water Line (GMWL) equation was found to be δD = 8δ 18 O + 10 [28].The study area was located on the eastern edge of the Tibetan Plateau and the atmospheric precipitation line for the Tibetan Plateau is the Local Meteoric Water Line (LMWL) with the equation δD = 8.41δ 18 O + 16.72 [33].The δ 18 O and δD distribution maps showed that most of the hot springs in the study area were in close proximity to the GMWL and LMWL (Figure 2).This provided important information about the source of geothermal water, which is that they were recharged by infiltrating atmospheric precipitation.The following two interesting characteristics of geothermal waters can be identified in this figure .(1) The hydrogen and oxygen isotopes from 2008 showed a strong δ 18 O right shift effect (Figure 2b), which the strong positive δ 18 O shift was generally owed to strong water-rock interactions influenced by three factors [34]: 1 the high temperature of the reservoirs, 2 the long circulation time and 3 the high ratios of rock to water.Nevertheless, it was unclear what factor controlled the δ 18 O shift occurring; (2) The hydrogen and oxygen isotopes from 2013 showed a strong δ 18 O left shift that may be because of the δ 18 O exchange that occurred during the dissolution of CO 2 from deep source [35].
Previous studies have shown that the elevation effect was an important factor affecting the isotope of atmospheric precipitation, showing a decrease in the isotope value with increasing elevation, so the recharge elevation of hot spring water can be estimated by using the elevation effect of the isotope [36].The formula for calculating the δD value of precipitation and elevation H in western China was δD = −0.026H − 30.2 [37].Based on the equation, we calculated the recharge elevation of the study area to be 1.9-4.8km.

Origin of Major Elements
The water samples collected from the hot springs were analysed, and the results are shown in Supplementary Table S1.The ion balances of all samples were less than 5%, indicating that the analytical results for these samples were plausible.Temperatures of water springs were in the range of 10 • C to 82 • C, with a pH ranging from 6.25 to 9.45.Conductivity ranged from 3.67 to 57600.00 µs/cm, and TDS values ranged from 80.86 to 2754.01 mg/L.The main cations in hot springs were Na + , Ca 2+ and Mg 2+ , while the main anions were HCO 3 − .The concentrations of Na + , K + , Mg 2+ and Ca 2+ ranged from 2.75 to 958.23 mg/L, 0.00 to 128.24 mg/L, 0.00 to 440.35 mg/L and 0.00 to 286.59 mg/L, respectively; the concentrations of Cl − , SO 4 2− , and HCO 3 − ranged from 0.31 to 871.45 mg/L, 0.00 to 688.86 mg/L, and 37.82 to 2588.79 mg/L, respectively.
The results of chemical analyses of hot spring waters from the study area were given in Supplementary Table S1.The main chemical compositions of the waters were plotted on the Piper diagram, and water samples were plotted in 1, 2, 4 and 6 blocks (Figure 3).The type of hydrochemistry of most groundwater from the XSHF were HCO 3 -Na, apart from some samples which were of the HCO 3 •Cl-Na, HCO 3 -Ca-Na, HCO 3 -Na-Ca, HCO 3 -Na•Mg, HCO 3 -Ca and HCO 3 -SO 4 -Mg type, The correlation graph of Na + +K + vs. Ca 2+ +Mg 2+ , in which iso-ionic-salinity lines are drawn for reference (Figure 4) [38,39].The hot spring samples have different total salinity (TIS), ranging from 3.7 to 90.9 meq/kg.Twenty-four hot springs were of the HCO 3 -Na type.Based on previous studies, HCO 3 -Na type geothermal water was typical of high-temperature geothermal systems and was generated by chemical reactions between infiltrated meteoric water, dissolved carbon dioxide, and reservoir rocks containing dolomite and microcline as the primary minerals [40].The total salinity (TIS) ranged from 3.7 to 90.9 meq/kg.The process could be illustrated in the Equations of ( 1) and ( 2).
However, the hot springs of W31, W32, W34, W35, W37 and W41 were HCO 3 •Cl-Na types.The total salinity (TIS) ranged from 29.5 to 90.3 meq/kg.Interestingly, the primary properties of these six hot springs are high temperature and high Cl − concentrations, which suggests that the parent geothermal fluid was formed by the mixing of snow and juvenile water deep in the subsurface under the influence of magma, with the deep parent geothermal fluid having travelled long distances to get to the surface, where water from different sources participated in the mixing during the ascent.It included three mixed endmembers of the cold water, consisting of local precipitation, river water and snowmelt water [16].Furthermore, the high Ca 2+ , Mg 2+ and SO 4 2− concentrations of most geothermal waters (W1-W3, W8, W9, W11, W19, W21, W38 and W42) also indicated that the geothermal waters in the study area might not have been influenced by magma indicating that the magma-influenced hydrochemical signature was instead masked by mixing processes.The total salinity (TIS) ranged from 5.2 to 71.8 meq/kg.The process could be illustrated in the Equations of (3)-(5).

Characteristics of Strontium Isotopes
The 87 Sr/ 86 Sr in a hot spring was frequently similar to the rock minerals it comes in touch with [41].Therefore, this ratio can be an effective tracer of the interaction between various rock minerals.Silicate, carbonate and sulphate minerals are important sources of Sr in groundwater and influence the 87 Sr/ 86 Sr of groundwater [42].The 87 Sr/ 86 Sr values of carbonate and sulphate weathering sources were about 0.708000, and the aluminosilicate weathering sources varied between 0.716000 and 0.720000 [40,43].The ratio of 87 Sr/ 86 Sr in the thermal springs in the study area ranged from 0.707290 to 0.716574 (Figure 5 and Table 2).The W12 and W22 hot springs belong to the silicate mineral weathering, and the W24, W25, W27 and W42 hot springs belong to the carbonate mineral weathering, and the remnant springs were between carbonate and silicate mineral weathering.They were formed by the deep circulation of atmospheric precipitation in the local heat flow system interacting with Sr-bearing rocks in the crust consistent with the results of hot spring chemistry.The Na-K-Mg triangle can be used to assess whether the groundwater was in equilibrium, partially in equilibrium or immature, which indicated the degree of water-rock reaction [44].The Na-K-Mg triangular plot (Figure 6) showed that W4, W6, W7, W12, W16, W25 and W35 water samples were located in the partial equilibration zone, while other hot spring samples were located in the immature water zone.This indicated that most of the hot springs on the XSHF were mainly recharged by atmospheric precipitation and circulated at a relatively rapid rate, with a few having some degree of hydromorphic reaction.

Mineral Saturation States
Regarding the analysis of hot spring water samples, the mineral saturation index (SI) is calculated using the PHREEQC software.The results of this study are presented in Figure 7: it can be noticed that all spring water samples are saturated (SI > 0) with respect to Calcite, Chalcedony, Dolomite, Quartz, Halite and Aragonite are basically in equilibrium (SI ≈ 0).However, SI with respect to Dolomite varies greatly in each hot spring water.Groundwater samples in W5, W9, W10, W13, W14, W15, and W20 are in a supersaturated state (SI values are 3.18, 4.69, 4.14, 3.93,4.49,3.84 and 3.96, respectively).However, they are unsaturated in W24, W25, W33, W38 and W43 (SI values are −1.31,−0.73, −0.71, −0.04 and −2.04, respectively), and the rest is basically in equilibrium (SI ≈ 0).This phenomenon may reflect the difference in the surrounding rock characteristics.The supersaturation indicates a high content of these minerals and a long residence time in the aquifer system [45].However, albite, K-feldspar, anorthite, chrysotile and halite are found in an unsaturated state in the majority of spring waters, indicating that they are relatively soluble or have insufficient reaction time with hot water.Regarding the analysis of hot spring water samples, the mineral saturation index (SI) is calculated using the PHREEQC software.The results of this study are presented in Figure 7: it can be noticed that all spring water samples are saturated (SI > 0) with respect to Calcite, Chalcedony, Dolomite, Quartz, Halite and Aragonite are basically in equilibrium (SI ≈ 0).However, SI with respect to Dolomite varies greatly in each hot spring water.Groundwater samples in W5, W9, W10, W13, W14, W15, and W20 are in a supersaturated state (SI values are 3.18, 4.69, 4.14, 3.93,4.49,3.84 and 3.96, respectively).However, they are unsaturated in W24, W25, W33, W38 and W43 (SI values are −1.31,−0.73, −0.71, −0.04 and −2.04, respectively), and the rest is basically in equilibrium (SI ≈ 0).This phenomenon may reflect the difference in the surrounding rock characteristics.The supersaturation indicates a high content of these minerals and a long residence time in the aquifer system [45].However, albite, K-feldspar, anorthite, chrysotile and halite are found in an unsaturated state in the majority of spring waters, indicating that they are relatively soluble or have insufficient reaction time with hot water.

Reservoir Temperature and Circulation Depth
Geothermometers can be used to estimate the thermal storage temperature of hot springs.Frequently used geothermal temperature scales include cationic temperature scales as well as SiO2 temperature scales [46].A comprehensive analysis of the Na-K-Mg triangle diagram and the hydrochemical characteristics of the hot springs, The Na-K and K-Mg geothermometers were used to calculate the thermal storage temperature in study areas with W4, W6, W7, W12, W16, W25, W35 and SiO2 was more stable than other minerals and can therefore indicate the thermal storage temperature of immature water.The Na-K geothermometers indicated the more complete equilibrium of geothermal

Reservoir Temperature and Circulation Depth
Geothermometers can be used to estimate the thermal storage temperature of hot springs.Frequently used geothermal temperature scales include cationic temperature scales as well as SiO 2 temperature scales [46].A comprehensive analysis of the Na-K-Mg triangle diagram and the hydrochemical characteristics of the hot springs, The Na-K and K-Mg geothermometers were used to calculate the thermal storage temperature in study areas with W4, W6, W7, W12, W16, W25, W35 and SiO 2 was more stable than other minerals and can therefore indicate the thermal storage temperature of immature water.The Na-K geothermometers indicated the more complete equilibrium of geothermal water, with the result corresponding to the highest temperature at the equilibrium of geothermal waterrock action [47], while the K-Mg geothermometers indicated the shallower equilibrium of geothermal water, with the result corresponding to the lowest temperature at the equilibrium of geothermal water-rock action [44,48].Therefore, we used multiple geothermal temperatures scaling methods to calculate the thermal storage temperature in the study area (Table 3).Therefore, the reservoir temperature of spring samples in the XSHF was mainly 8-142 • C. The depth of circulation of the hot spring samples was assessed according to formulation (6): Z is the circulation depth (km); Z 0 is the depth of constant temperature zone (km); T is the reservoir temperature ( • C); T 0 is the temperature of constant temperature zone ( • C), namely the average local temperature; T grad is the geothermal gradient ( • C/km), and reflects the geothermal changes for each kilometre below the constant temperature zone [49].Based on previous studies in the area, the geothermal gradient T grad was assumed as 20.4 • C/km, the annual mean temperature T 0 assumed as 7 • C, and the depth Z 0 of the constant temperature zone assumed as 30 m [17].The circulation depth of the XSHF was about 0.1-6.9km, as calculated.

Spatial Distribution of Hydrogeochemical Characteristics in XSHF
The 43 hot springs were distributed along the five segments of the XSHF.As shown in Figure 8.The hot springs in the study area were consistent in terms of the magnitude and depth of the earthquakes that occur, the depth of circulation and the slip rate in the different sections.The rate of slip in the Zhuwo (W1, W2) and the Luhuo faults (W3, W4, W5) was 4 mm•a −1 , The water temperature of the hot springs in these two segments was low, ranging from 10 to 26 • C. The depth of circulation was shallow, ranging from 0.1 to 2.7 km, and the magnitude of the earthquakes in these two areas was small, with most earthquakes less than magnitude 3, with shallow source depths.Next were the Bamei (W16-W20) and the Moxi (W36-W43) segment hot springs, finally, The rate of slip, reservoir temperature and circulation depth of the Kangding (W21-W35) and the Daofu (W9-W15) segments were ranged from 13 mm•a −1 to 6.5 mm•a-1, 16 • C to 142 • C and 0.5 km to 6.7 km, which had the highest reservoir temperature and the most frequent strong earthquakes in the study area (Figure 8).Some of the hot springs had high temperatures, and the highest temperature reached 82 • C (W35).
The XSHF was a fracture at a depth that reached the Moho phase; multiple phases of magmatism have occurred in the Kangding area [25].Therefore, there was a high probability that the Kangding hydrothermal system would have been influenced by the hydrochemical species of the hydrothermal fluids by the magma.The value of 3 He/ 4 He indicated that helium from the springs was partially derived from the mantle of the Kangding segment [13,50].Thus, the hydrochemical characteristics of spring water in the Kangding geothermal field were controlled by the Kangding hydrothermal system.In contrast, the Kangding hydrothermal system has developed several fractures, which cut deep into the crust, established channels for hydrothermal activity, and provided a tectonic basis for the hydrothermal system [25].

Correlation between Hydrogeochemical Changes and Earthquakes Precursory and Post-Seismic Anomalies
Great earthquakes are usually associated with the physicochemical variations of groundwater and hot springs [51,52].The short-term hydrogeochemical precursors for earthquakes, the co-seismic response of hydrochemistry and post-seismic geochemical and isotopic changes of hot springs have been reported [37,51,53,54].We studied the changes in water chemistry in nine thermal springs (W25, W27, W30, W31, W34, W36, W37, W39  4 and Figure 7).The ion concentrations (Na + , Cl − and SO 4 2− ) in the hot springs in the XSHF were plotted as a line graph of the main ion concentrations over time, and the mean + 2 times the standard variance of the collected samples was used to determine the threshold of anomalies in the study area.The results showed that the temporal variation of ions (such as Na + , Cl − and SO 4 2− ) in the hot spring waters decreased obviously after the 12 May 2008 Wenchuan Ms8.0, with unusual value fluctuations between the 20 April 2013 Lushan Ms7.0 and 22 November 2014 Kangding Ms6.3 earthquakes occurred.In addition, the concentrations of these ions returned to the normal levels after 20 July 2017 (Table 4 and Figure 9).
Many mechanisms have been proposed to explain the post-seismic anomalies.For example, Hydrostatic strain caused by earthquakes leads to mixing of deep and shallow water, increased reactive surface area following rock surface rupture leads to accelerated waterrock interaction, and the hydrological barrier between two chemically distinct aquifers may rupture, re-establishing water-rock chemical equilibrium [1,55,56].Stable isotope changes can be traced to hydrogeological processes such as changes in permeability, groundwater mixing degree and water-rock interactions [1,47,51], which are closely related to seismically induced stress changes.Hence, we also obtained some hydrogen and oxygen isotope data for some hot springs from 2008 to 2019.Although D and 18 O isotope data were not complete, it is important to discuss the isotope evolution after the 2008 Wenchuan Ms8.0 earthquake and the 2013 Lushan Ms7.0 earthquake.As shown in Figure 2b, the δD and δ 18 O values in the thermal springs deviated from the LMWL after the 2008 Wenchuan earthquake but shifted toward the LMWL in 2009 and 2010.These δ 18 O was more negative after the 2008 Wenchuan Ms8.0 earthquake was due to hydrostatic interaction with δ 18 O-rich enclosing rocks during deep circulation, accompanied by oxygen isotope exchange, increasing in groundwater δ 18 O [34].The hydrogen and oxygen isotopes of the hot spring water were collected in May 2013 deviated to the upper left of the atmospheric precipitation line.This is probably because of the exchange of δ 18 O isotopes during the dissolution of hot spring water with CO 2 from deeper sources [36].Therefore, it was assumed that the change in ion concentration in each hot spring during the 2013 Lushan Ms7.0 earthquake was due to increased permeability between aquifers and groundwater mixing.
The obvious seismic precursors were captured in the study area, the concentration of dissolved ions (Na + , Cl − and SO 4  2− ) in these hot springs increased by 5% to 36%, with the obvious precursor anomaly three months before the 22 November 2014 Kangding Ms6.3 earthquakes.These phenomena were found in one hot spring and shown in W25, W27, W30, W31, W34, W36, W37, W39 and W43 hot springs.In addition, changes in hydrochemical compositions were more obvious and increased rate about 20% to 30%, which was located 40 km away from the epicentre of the Kangding Ms6.3 earthquake, and the ion concentrations of W36, W37, W38 and W43 hot springs increased slightly only about 10%, which was located 80 km away from the epicentre of the Kangding Ms6.3 earthquake.Moreover, the hydrogeochemical changes of these hot springs showed less pronounced to other three large earthquakes (2010 Yushu Ms7.0, 2017 Jiuzhaigou Ms7.0, and 2019 Changning Ms6.0 earthquakes) occurred around the XSHF may be caused by the distance of the epicentre.The ion concentration of these hot springs (W25, W27, W30, W31, W34, W36, W37, W39 and W43) were sensitive to seismicity located in a huge leftlateral strike-slip active fault (XSHF) [57], which set the eastern boundary of the large-scale clockwise rotation of crustal materials of the Himalaya syntaxis [54,58].When the stress increased up to the sub-instability stress state of the XSHF, the Kangding Ms6.3 earthquakes occurred [59].The concentrations of Na + , Cl − and SO 4 2− were sensitive to the increase of stress in the XSHF, which could enhance the opening of microfractures under these hot springs [60].Based on the various characteristics of stress and strain, the ion concentration also showed different anomalous characteristics with the change of permeability between aquifers, groundwater mixing and water-rock interactions, showing a positive increase in ion concentration of hot springs [61,62].Many mechanisms have been proposed to explain the post-seismic anomalies.Fo example, Hydrostatic strain caused by earthquakes leads to mixing of deep and shallow water, increased reactive surface area following rock surface rupture leads to accelerate water-rock interaction, and the hydrological barrier between two chemically distin aquifers may rupture, re-establishing water-rock chemical equilibrium [1,55,56].Stab isotope changes can be traced to hydrogeological processes such as changes i permeability, groundwater mixing degree and water-rock interactions [1,47,51], whic are closely related to seismically induced stress changes.Hence, we also obtained som hydrogen and oxygen isotope data for some hot springs from 2008 to 2019.Although and 18 O isotope data were not complete, it is important to discuss the isotope evolutio

Conceptual Model of Hot Springs in the XSHF
Geothermal water was recharged by the infiltrated precipitation and heated by fractures that seep deep into the earth and through a deep cycle of groundwater [40].The XSHF is a typical deep-developed fracture system that accompanies high-temperature geothermal systems.The deep faults serve not only as a conduit for further groundwater infiltration but also as a conduit for the upward flow of deep geothermal fluids.The heat source of the high-temperature hydrothermal system is mostly from the deep magmatic heat of the fracture zone, radioactive heat from granitoids and frictional heat (Figure 10).Ai et al. (2021) [63] studied the effect of shear frictional heat on high-temperature hydrothermal activity in the XSHF through numerical simulations.They concluded that shear frictional heat was the controlling heat source for the geothermal field as it may lead to crustal melting at 20-25 km depth.Therefore, the shear frictional heat of the XSHF may have provided energy for a range of high-temperature geothermal systems.Meteoric water penetrates into the aquifer at around 5 km in the Gunga snow mountain and the Great Snow Mountain through the hydraulic conductivity fracture zone along the fractures between and around the mountain ranges and river terraces.As the groundwater circulates downwards to about 6.9 km, it is heated by deep magmatic heat, granite-like radiogenic heat, or sliding frictional heat sources from the XSHF, and the water is heated to a maximum of 124 • C.During the heating process, water-rock reactions occur with the surrounding rocks at different depths, influenced by temperature and pressure.As it rises to the surface, it mixes with surface water or shallow groundwater and is finally exposed to the surface to form the hot spring.If the crustal stresses in the area change, this can affect the equilibrium state of the hot spring water and lead to changes in its hydrogeochemical characteristics [64].Therefore, continuous monitoring could be conducted on a proper hot spring spot of the fault zone to further study the pre-seismic hydrochemical precursors.
Water 2022, 14, 550 18 of 21 area change, this can affect the equilibrium state of the hot spring water and lead to changes in its hydrogeochemical characteristics [64].Therefore, continuous monitoring could be conducted on a proper hot spring spot of the fault zone to further study the preseismic hydrochemical precursors.

Conclusions
Using extensive hydrochemical data from 43 hot spring sites, the mechanisms and processes of geochemical changes in the XSHF were described from the perspective of the regional groundwater flow system.The results of the study showed that,

Conclusions
Using extensive hydrochemical data from 43 hot spring sites, the mechanisms and processes of geochemical changes in the XSHF were described from the perspective of the regional groundwater flow system.The results of the study showed that, (3) The spatial distribution of hydrogeochemical characteristics of hot springs in the XSHF indicates part of the hot springs in the Kangding (W21-W35) and the Daofu (W9-W15) segments which had the highest reservoir temperature and the most frequent strong earthquakes in the two segments.(4) The temporal variations of dissolved chemical concentrations (Na + , Cl − and SO 4 2− ) decreased obviously after the Wenchuan Ms8.0 earthquake, with abnormal value fluctuations between the Lushan Ms7.0 and Kangding Ms6.3 earthquake occurred and after the 30 July 2017 returned to normal levels.And the ion concentrations (Na + , Cl − and SO 4 2− ) in hot springs increased by 5% to 35% three months before 22 November 2014 Kangding Ms6.3 with the obvious precursor anomaly.These results indicate that synchronous isotopic changes, as well as changes in water chemistry between multiple hot springs, were important for earthquake forecasting and the study of the response of earthquake precursors.

Figure 1 .
Figure 1.The plot of sampling site distribution.(a) Localization of the area of this study; (b) Topographic map in the XSHF.XSHF: Xianshuihe fault; LMSF: Longmenshan fault; (c) Geological map in the XSHF.

Figure 1 .
Figure 1.The plot of sampling site distribution.(a) Localization of the area of this study; (b) Topographic map in the XSHF.XSHF: Xianshuihe fault; LMSF: Longmenshan fault; (c) Geological map in the XSHF.
springs along the XSHF fault in June 2008, June 2009, June 2010, April 2013, August 2014, June 2015, January 2016, February 2017, March 2018 and April 2019, respectively (Figure 1, O + 10 [28].The study area was located on the eastern edge of the Tibetan Plateau and the atmospheric precipitation line for the Tibetan Plateau is the Local Meteoric Water Line (LMWL) with the equation δD = 8.41δ 18 O + 16.72 [33].The δ 18 O

Figure 2 .
Figure 2. Stable oxygen and hydrogen isotope of the 43 hot springs and their correlations with GMWL and LMWL.LMWL: δD = 8.41δ 18 O + 16.72 [33].(a) Hydrogen and oxygen isotope distribution of all hot spring sites in the study area.(b) The distribution of hydrogen and oxygen isotopes at the hot spring sites that were measured after the 12 May 2008 Wenchaun MS8.0 and 20 April 2013 Lushan MS7.0, respectively.The rosy symbol represented samples collected after the 20 April 2013 WenchaunMS8.0 and the sky-blue symbol stands for samples collected after the 20 April 2013 Lushan MS7.0.

Figure 2 .
Figure 2. Stable oxygen and hydrogen isotope of the 43 hot springs and their correlations with GMWL and LMWL.LMWL: δD = 8.41δ 18 O + 16.72 [33].(a) Hydrogen and oxygen isotope distribution of all hot spring sites in the study area.(b) The distribution of hydrogen and oxygen isotopes at the hot spring sites that were measured after the 12 May 2008 Wenchaun Ms8.0 and 20 April 2013 Lushan Ms7.0, respectively.The rosy symbol represented samples collected after the 20 April 2013 Wenchaun Ms8.0 and the sky-blue symbol stands for samples collected after the 20 April 2013 Lushan Ms7.0.

Figure 3 .
Figure 3. Piper diagram showing major ion chemistry of the sampled points.Figure 3. Piper diagram showing major ion chemistry of the sampled points.

Figure 3 .
Figure 3. Piper diagram showing major ion chemistry of the sampled points.Figure 3. Piper diagram showing major ion chemistry of the sampled points.

Figure 3 .
Figure 3. Piper diagram showing major ion chemistry of the sampled points.

Figure 4 .
Figure 4. Correlation plot of Na + +K + vs. Ca 2+ +Mg 2+ for the spring water samples along XSHF, also showing iso-ionic-salinity lines for reference.Figure 4. Correlation plot of Na + + K + vs. Ca 2+ + Mg 2+ for the spring water samples along XSHF, also showing iso-ionic-salinity lines for reference.

Figure 4 .
Figure 4. Correlation plot of Na + +K + vs. Ca 2+ +Mg 2+ for the spring water samples along XSHF, also showing iso-ionic-salinity lines for reference.Figure 4. Correlation plot of Na + + K + vs. Ca 2+ + Mg 2+ for the spring water samples along XSHF, also showing iso-ionic-salinity lines for reference.

Figure 7 .
Figure 7. Saturation indices values of groundwater samples with respect to minerals.

Figure 7 .
Figure 7. Saturation indices values of groundwater samples with respect to minerals.

Figure 8 .
Figure 8.The spatial distribution of temperature, the depth of circulation of 43 hot springs and slip rates, magnitudes and focal depths in the XSHF.The different colours represent different areas of the Xianshui River Fault Zone, with blue representing the Zhuwo and Luhuo segment, Green representing the Daofu segment, Rose representing the Bamei segment, Light red representing the Kangding segment, Orange representing the Moxi segment.

Figure 8 .
Figure 8.The spatial distribution of temperature, the depth of circulation of 43 hot springs and slip rates, magnitudes and focal depths in the XSHF.The different colours represent different areas of the Xianshui River Fault Zone, with blue representing the Zhuwo and Luhuo segment, Green representing the Daofu segment, Rose representing the Bamei segment, Light red representing the Kangding segment, Orange representing the Moxi segment.

Figure 9 .
Figure 9. Temporal variations of concentration of Na + , Cl − , SO 4 2− , δD, δ 18 O and earthquake: (a) is the Erdaoqiao spring (W30); (b) is the Guanding spring (W34); (c) is the Longtougou spring(W31); (d) is the Mugecuo spring (W27) and (e) is the Xinxingxiang spring (W37), Blue bars show the near-field earthquakes within 50 km and magnitude range from M L 1.0 to M L 3.0.Red bars show the far-field earthquakes within 300 km and magnitude range over M L 3.0.Hydrogen and oxygen isotope data for 2019-2020 in the figure are from Zhang et al. (2021) [17].

Figure 10 .
Figure 10.Conceptual model of the origin of groundwater and the hydrogeochemical cycling process in the XSHF.

( 1 )
The hot spring water at XSHF is recharged by infiltrating precipitation at recharge elevations of 1.8 to 4.5 km (2) The hydrochemical types of most hot springs were mainly controlled by aquifer lithology.The heat storage temperature range was inferred from an equation based on SiO2 and chemical geothermometers method as 8-124 °C.The circulation depths for springs were estimated to range from 0.1 to 6.9 km.(3) The spatial distribution of hydrogeochemical characteristics of hot springs in the XSHF indicates part of the hot springs in the Kangding (W21-W35) and the Daofu (W9-W15) segments which had the highest reservoir temperature and the most frequent strong earthquakes in the two segments.(4) The temporal variations of dissolved chemical concentrations (Na + , Cl-and SO4 2-) decreased obviously after the Wenchuan Ms8.0 earthquake, with abnormal value

Figure 10 .
Figure 10.Conceptual model of the origin of groundwater and the hydrogeochemical cycling process in the XSHF.

( 1 )
The hot spring water at XSHF is recharged by infiltrating precipitation at recharge elevations of 1.8 to 4.5 km.(2)The hydrochemical types of most hot springs were mainly controlled by aquifer lithology.The heat storage temperature range was inferred from an equation based on SiO 2 and chemical geothermometers method as 8-124 • C. The circulation depths for springs were estimated to range from 0.1 to 6.9 km.
Contributions: Conceptualization, Y.Y., X.Z. and L.L.; methodology, Y.Y. and X.Z.; software, Y.Y., J.T. and F.L.; validation, S.O., Y.L. and F.L.; formal analysis, J.T. and Y.L.; investigation, S.O., Z.S. and F.L.; data curation, Z.S. and J.T.; writing-original draft preparation, Y.Y.; writing-review and editing, Y.Y., X.Z. and L.L.; visualization, Y.Y.All authors have read and agreed to the published version of the manuscript.Funding: The work was funded by National Key Research and Development Project (2017YFC1500501-05, 2019YFC1509203) and the National Natural Science Foundation of China (41673106, 42073063, 4193000170) and The Special Fund of the Institute of Earthquake Forecasting, China Earthquake Administration (2021IEF0101, 2021IEF1201).Spark Program for Earthquake Science and Technology (XH20066).Institutional Review Board Statement: Not applicable.Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Table 1 .
Location of the surveyed area of hot spring in the XSHF.

Table 2 .
The data of silicon dioxide, strontium and strontium isotope from hot springs.

Table 3 .
Apparent equilibrium temperatures were computed through multi-geothermometers for the samples collected from the thermal springs of XSHF.

Table 4 .
The occurrence time of precursory and post-seismic anomalies in the continuous monitoring sites before three earthquakes.