Groundwater level and temperature changes following the great Tangshan earthquake of 1976 near the epicenter

Abstract Several meters of rise in groundwater level with significant changes in groundwater temperature were observed near the epicenter of the 1976 Tangshan earthquake—an intraplate earthquake with a magnitude of MS=7.8 that occurred in northern China. The origin of the groundwater level rise, however, remains a mystery. Studying the groundwater level rise can provide insight into the faulting process and has implications for water resources. In this study, we analyze the indications of the groundwater temperature data on the characteristics of the rising groundwater. The results show that the upward flow was maintained for approximately 5 days, and the depth of the rising groundwater was about 1–2 km. We then examine the groundwater level and temperature data with several reported models in order to explain the cause of the groundwater level rise. The results indicate that the groundwater level rise is attributable to the earthquake-induced enhancement of vertical hydraulic connection between the shallow and deep aquifers, leading to the upwelling of deep fluids that causes the groundwater levels to rise.


Introduction
Hydrological responses such as liquefaction of sediments, increased stream discharge, changes in water level and temperature of groundwater have been widely documented following large earthquakes.The hydrological responses in the vicinity of seismogenic faults (near-field) are the key in understanding earthquake processes and are important for analyzing the impact of earthquakes on water resources (Wang and Manga 2021).On 28 July 1976, an earthquake with a magnitude of M S ¼7.8 occurred in the city of Tangshan, northern China.The earthquake struck the densely populated city and caused $242,000 fatalities and huge economic losses (State Seismological Bureau 1982).The earthquake also induced widespread liquefaction of sediments (Liu et al. 1982;Tang 1985) and caused large-scale changes in groundwater level (Wang et al. 1979;Zhang and Li 1981;Wang and Yin 1982).The groundwater level changes are prominent in the vicinity of the epicenter, with multiple wells recording several meters of rise.The groundwater level rise has attracted the attention of many scholars, whereas their explanations for the origin of the groundwater level rise are competing: some suggested that it might be associated with the cessation of drainage in Tangshan Mine caused by the power outage due to the Tangshan earthquake (Chen 1981;Zhang and Li 1981), whereas others argued that it was a direct result of the earthquake (Wang et al. 1979;Wu et al. 1980;Wang and Yin 1982).Thus, the mechanism responsible for the rising of groundwater level still deserves further study.
Many models such as the coseismic static strain, the consolidation and liquefaction, and the permeability enhancement have been developed to explain the near-field hydrological responses observed in various settings.Wakita (1975) invoked a coseismic static strain model on the basis that the locations of coseismic rise and fall in well water level around the 1974 Izu-Hanto-Oki earthquake nearly coincide with the areas of contraction and dilatation caused by the earthquake faulting.In the subsequent studies, the signs of coseismic changes in some cases were nearly consistent with the calculated static strains (Igarashi and Wakita 1991;Muir-Wood and King 1993;Roeloffs 1996;Quilty and Roeloffs 1997;Ge and Stover 2000;J onsson et al. 2003;Shi et al. 2013;Lai et al. 2014), whereas the observations in some other cases did not match the coseismic static strain model (Chia et al. 2001;Wang et al. 2001;Koizumi et al. 2004;Mohr et al. 2017, Hosono et al. 2019).The consolidation and liquefaction model has been demonstrated by laboratory experiments, in which loose sediments consolidate under cyclic shear stress, resulting in an increase in pore pressure and a decrease in effective stress, which eventually induce the liquefaction of sediments (Terzaghi et al. 1996;Wang 2007); in field observation, the model dominates the groundwater level increase in an alluvial fan adjacent to the 1999 Chi-Chi earthquake (Wang et al. 2001), and the model also works in interpreting the responses of streamflow to large earthquakes occurred in the western United States (Manga 2001;Manga et al. 2003).The permeability enhancement model has been suggested to account for the hydrological responses induced by local and distant earthquakes in some previous studies (Briggs 1991;Rojstaczer et al. 1995;Roeloffs 1998).Researches on the mechanisms of permeability enhancement are ongoing.Brodsky et al. (2003) proposed a hypothesis to explain the coseismic water level steps in a well near Oregon that permeability enhancement is caused by earthquake-induced mobilization of colloidal particle; the hypothesis has been validated by a growing number of observations (Elkhoury et al. 2006;Xue et al. 2013;Yan et al. 2014;Shi et al. 2015;He et al. 2017, He andSingh 2019;Zhang et al. 2019Zhang et al. , 2023)).Wang et al. (2004) attributed the increased streamflow near the Chi-Chi earthquake to the coseismic release of water from mountains, which is considered to be a result of created vertical permeability; the mechanism of created vertical permeability has been supported by many subsequent observations (Liao et al. 2015;Wang and Manga 2015;Wang et al. 2016;Hosono et al. 2020).To clarify the origin of hydrological responses in the near-field, additional data are needed to test the reported or new-developed models.Some recent studies have shown that groundwater temperature responses to earthquakes can provide additional insights into the mechanisms of hydrological responses to earthquakes (Wang et al. 2012;Cox et al. 2015;Ma 2016;Shi et al. 2018;Miyakoshi et al. 2020).In this study, we investigate the groundwater level rise and changes in groundwater temperature following the 1976 Tangshan earthquake near the epicenter.We analyze the indications of the groundwater temperature data on the characteristics of the rising groundwater.We revisit the correlation between the groundwater rise and drainage in Tangshan Mine, and we further discuss the possible mechanism of the groundwater level rise based on several reported models.

Geological setting
The Tangshan earthquake and the wells studied are located in the city of Tangshan.The city is situated on the border of the Yanshan mountain belt and the North China Plain.The northern part of the city is low hills, and the southern part is the alluvial plain of the Douhe River.The Douhe River flows through the Douhe reservoir and the eastern part of Tangshan, and the river finally flows into the Bohai Sea (Figure 1).Tangshan City is located at the junction of the eastern Beiziyuan anticline and the western Kaiping syncline (Hou 1985).The stratigraphic lithology from top to bottom is the Quaternary clay and sand, the Carboniferous and Permian sandstones, the shale and coal, and the Ordovician limestone (Figure 2).The Quaternary sediments cover most of the area.The thickness of the sediments gradually increases from north to south, from a few meters to hundreds of meters.The Carboniferous and Permian coal is the mining layer of Tangshan Mine.Tangshan Mine, one of the principal mine pitheads of Kailuan Mining Bureau, has a mining history of more than 100 years.Tangshan Mine has a mining depth down to $815 m.Beneath the coal is the Ordovician limestone.According to the exploration data from Tangshan Mine (Yang and Yu 1981;Gu and Zhou 1985), there are four aquifers with relatively high water content.The topmost aquifer is the pebble and gravel aquifer in Quaternary sediments.The depth of this aquifer is below 120-230 m, and the top of the aquifer is covered by a 5-30 m thick clay aquitard.The specific field at some depths of the aquifer is up to 2.8 L/(sÁm).There are two aquifers in the Carboniferous and Permian rocks: the fractured sandstone aquifer above the roof of the 5 th mining gallery, and the fractured aquifer between the 12 th and the 14 th mining galleries.The specific field of the two aquifers is 0.003-0.2L/(sÁm) and 0.01-0.1 L/(sÁm), respectively.The bottommost aquifer is the karst water in the Ordovician limestone, and its vertical distance from the 14 th mining gallery is about 100-120 m.
A northeast-trending fault zone, the Tangshan fault zone, passes through the city of Tangshan (Figure 1).The fault zone is composed of three parallel faults, and from west to east are the Douhe fault, the Changshan-Weishan fault, and the Tangshan-Guye fault (Guo et al. 1977;Liu et al. 1982;2013).The Douhe fault has a total length of $50 km.The section in Tangshan Mine was named Shan I fault by Kailuan Mining Bureau.The fault is a normal fault with a dip angle of 50-90 .The strata near the fault are strongly folded and are reversed in some sections (Figure 2).The Changshan-Weishan fault is $20 km long.It was subdivided into Shan II and Shan Tangshan earthquake, and the coseismic deformation.The WL and WT represent wells monitoring water level and temperature, respectively.The coordinate information of the faults, the surface rupture, the horizontal coseismic displacement, and the coseismic vertical displacement is from Liu et al. (2013), Guo et al. (2011), Wan et al. (2017), andGuo et al. (1977), respectively.III faults.The Shan II and Shan III faults are reverse faults.The dip angles of the faults decrease with increasing depth.The dip angle is 30-60 above 500 m and less than 15 below 500 m.The Tangshan-Guye fault has a length of $30 km and was subdivided into Shan IV and Shan V faults.The Shan IV fault is a normal fault with a dip angle of 60 , and the Shan V fault is a reverse fault with a dip angle of 45 (Chen 1981;Liu et al. 1982;Gu and Zhou 1985).
The great earthquake of 1976 occurred in the south of Tangshan.The earthquake is a large continental intraplate earthquake.The highest intensity affected by the earthquake is XI on the Chinese intensity scale, and the intensity in Tangshan City is XI.The surface rupture zone generated by the earthquake is 8-11 km long and is mainly distributed along the Tangshan-Guye fault, which strikes $ N30 E (Figure 1).The horizontal displacement is up to 2.3 m in a right-lateral strike-slip sense.The northwest side was raised, and the vertical displacement was up to 0.7 m (Guo et al. 1977;Yang 1982;Du et al. 1985).Some further investigations have revealed that the surface rupture zone extended southward with a total length of more than 47 km (Qiu et al. 2005;Jiang 2006;Guo et al. 2011).The largest aftershock with a magnitude of M S ¼7.1 occurred $15 h after the main shock near Luanxian, a county $45 km northeast of Tangshan.Other aftershocks are distributed in an elliptical area 50 km wide and 140 km long, and the direction of the long axis of the ellipse is $ N50 E.

Groundwater level changes
We collected water level data from 8 wells near the Tangshan earthquake (one is the discharge data of Tangshan Mine).The basic information and the spatial distribution of these wells are provided in Table 1 and Figure 1, respectively.The eight wells are located 3-6 km from the epicenter and are distributed in the west of the surface rupture zone (Figure 1).The water levels of these wells rose by $2-10 m (the discharge increased by $3 times).The water level rise is sustained, and it took $3.5 years for most of the wells to recover to the pre-seismic levels (Figure 3).To facilitate the following analysis and discussion, here we describe the observations of these eight wells separately.
The TSK stands for Tangshan Mine.The discharge in the TSK is the sum of discharges from various mining galleries within Tangshan Mine.Following the Tangshan earthquake, the discharge increased from 23.6 to 80.11 m 3 /minute; after that, the discharge decreased and recovered to 56.5 m 3 /minute on 9 August 1976 (Figure 3a).The discharge of each mining gallery experienced a rising-recovery process, and their rising amplitudes and recovery rates are inconsistent.
The SXS2 well is located in Dazhao Park, Tangshan.The well has a depth of 287 m and is cased from the wellhead to 154 m.The screened section of the well is between 154 and 217 m, and the lithology of the screened section is the Ordovician limestone.The water level has been recorded since January 1974.The recorder between 1974 and 1979 was an SW-40 gauge with an accuracy of 1 cm.On 27 July 1976, one day before the Tangshan earthquake, the water level (elevation of water surface) was À8.17 m.The observation was interrupted after the earthquake.On 23 August 1976, the water level observation was resumed, and the water level was À2.01 m.The water level rise associated with the Tangshan earthquake was 6.16 m (Figure 3b).The actual water level rise was higher than 6.16 m due to the interruption of water level observation.According to Wang et al. (1988), a thermometer was installed $2 m below the water surface ($32 m to the wellhead), and the thermometer was thrown out of the well after the earthquake.
The SW1 and SW2 wells are located in Tangshan Hydrogeological Observation Station.The SW1 well has a depth of 185 m and is cased from the wellhead to  140.56 m.The screened section of the well is between 140.56 and 185 m, and the lithology of the screened section is Sinian siliceous limestone.The SW2 well has a depth of 71 m and is cased from the wellhead to the bottom.The screened section is not clear.The lithology from the wellhead to the bottom is Quaternary sediments.The water level of the SW1 well was monitored by a self-recording gauge between 1975 and 1979.The water levels of the SW1 and SW2 wells had similar annual variations.The water level observations were continuous before and after the Tangshan earthquake.Following the Tangshan earthquake, the depth of water level in the SW1 well increased from 30.3 m to 22.01 m, and the depth of water level in the SW2 well increased from 11 m to 9.17 m.The water level in the SW1 well decreased rapidly in the first 80 days after the earthquake, and it gradually recovered to the pre-seismic level in the following three years (Figure 3c).The recovery of water level in the SW2 well is difficult to distinguish because it was submerged in the background fluctuations (Figure 3d).The JS3 well, as well as the JS2 well, is located in the coal mining area of Tangshan Mine.The JS3 well has a depth of 563 m, and the lithology of the observed aquifer is the Ordovician limestone.The water level of the JS3 well increased by more than 6 m following the Tangshan earthquake (Figure 3e).
The Y42 well is located near the surface rupture zone of the Tangshan earthquake.The well has a depth of 707 m and is cased from the wellhead to 546.24 m.The screened section is between 546.24 and 707 m, and the lithology of the observed aquifer is the Ordovician limestone.The water level has been recorded since 1975.The water level after 1977 was monitored by a Hongqi-1 gauge with an accuracy of 1 cm.The groundwater level after 1977 was $7 m higher than that before the Tangshan earthquake (Figure 3f).It was reported that the water in the well began to overflow following the earthquake (Chen 1981;Wang and Yin 1982); in this case, the water level increased by more than 10 m after the earthquake.The inset of Figure 3f, in which the water level data were collected from Chen (1981), shows that the water level was 0.5 m seven days after the earthquake.
The S049 well penetrates the Shan I fault.The well has an original depth of 460 m.The lithology at 110-150 m and 150-445 m is the Quaternary pebble gravel and the Ordovician limestone, respectively.The strata below 445 m are reversed, and the age of the strata is Carboniferous (Chen 1981).The existing depth of the well was 250 m in Chen (1981) or 200 m in Zhang and Li (1981) and Wang and Yin (1982).The water level in the S049 well increased by more than 7 m following the Tangshan earthquake (Figure 3g).
The DC well is the No. 9 well of Tangshan Power Plant.The well is submersible.The well has a depth of 25 m.The observed section of the well is located at 12-20 m, and the lithology is the Quaternary sediments.The depth of water level was 21.56 m one day before the Tangshan earthquake.The observation was interrupted after the earthquake.On 9 September 1976, the water level observation was resumed, and the depth of water level was 12 m.The water level rise associated with the Tangshan earthquake was 9.56 m (Figure 3h).
In addition, for those wells whose water level recorders are not described above, the water levels were probably observed manually (Data Collection Group of Tangshan Earthquake 1981; Wang and Yin 1982).

Groundwater temperature changes
We collected water temperature data from five wells.Four wells have water level observations, and one has only water temperature observations.The basic information about these five wells is listed in Table 1.The temperature measured between 1973 and 1975 was primarily for studying the ground temperature of Tangshan Mine, and the temperature measured in August 1976 was for studying the Tangshan earthquake.The temperature was measured with a semiconductor thermistor thermometer, which has an accuracy of $0.1 C (Chen 1981(Chen , 1988)).
The temperature-depth profiles of the 5 wells are displayed in Figures 4 and 5a.Compared with the temperature before the Tangshan earthquake, the temperature changes after the earthquake varied at different depths.The post-seismic water temperature was higher than the pre-seismic values at the shallow parts of the SXS2, JS2, and S049 wells, whereas the temperature difference at the shallow parts of the JS3 and Y42 wells was not obvious.At the deep part of the SXS2 well, it is difficult to distinguish the water temperature change after the earthquake due to the slight difference in water temperature observed at different times.The post-seismic water temperature at some depths of the JS3, JS2, and S049 wells recovered to the pre-seismic values, while the temperature was changed at some other depths (residual changes).The JS3 well and JS2 well had a higher post-seismic water temperature than the preseismic values at 185-315 m and 350-425 m, respectively.The S049 well had a lower post-seismic water temperature than the pre-seismic values at 110-185 m (Figure 4).For the Y42 well (Figure 5a), a well with a large temperature difference, the water temperature at 0-430 m still increased by $1.5 C on the ninth day after the earthquake.The water temperature recovered to the pre-seismic level within $15 days.The water temperature at 430-585 m was disturbed slightly following the earthquake, and the temperature did not recover within $15 days (residual change).
The temperature-depth profiles in Figures 4 and 5 also show that the temperature gradient is relatively large (1.4-3.2C/100 m) in the Quaternary and Carboniferous strata, and the temperature gradient in the Ordovician strata is relatively small (around $0.1 C/100 m).

Maintaining time of the upward flow
We infer the maintaining time of the upward flow from modeling the recovery of the post-seismic water temperature changes in the Y42 well.The Y42 well was chosen because the well overflowed after the Tangshan earthquake; and because there is a record of changes in water temperature after pumping from the Y42 well, which is useful in the simulation (Figure 5b).
We used a two-dimensional numerical simulation to describe the recovery of the temperature changes quantitatively.We made two tentative hypotheses in the simulation.The first hypothesis was that the temperature of the upward flow was constant during the overflow.The second hypothesis was that the water temperature recovery in the wellbore was controlled by heat conduction between the well water and the surrounding rock after the well water stopped overflowing.We also made two assumptions for simplicity: one was to ignore the difference between the casing and the surrounding rock; the other was to ignore the heat transfer in the vertical direction due to geothermal gradients.We chose the depth of 150 m below the wellhead as a probe for the simulation.The 150 m was chosen because the temperature around this depth was measured three times after the earthquake; and because the impact of solar radiation at this depth was supposed to be little (Figure 5a).
By employing the program ANSYS, the simulation was implemented with an axial symmetry finite element method.The governing equation for heat conduction is (Yang and Tao 2006) where q is the density, c is the specific heat, T is the temperature, and k is the thermal conductivity.In cylindrical coordinates, the equation can be written as In the equation, the densities of the water and the surrounding rock are taken as 998.2 kgÁm À3 and 2563 kgÁm À3 , respectively.The specific heats of the water and the rock are taken as 4186 JÁkg À1 ÁK À1 and 860 JÁkg À1 ÁK À1 , respectively.The thermal conductivity of the water is taken as 0.6 WÁM À1 ÁK À1 .The observed rock thermal conductivity in Tangshan is between 1.65 WÁM À1 ÁK À1 and 5.77 WÁM À1 ÁK À1 (Wang andHuang 1988, 1990).
In the simulation, the adiabatic boundary is another parameter to be determined.We made two attempts to determine the adiabatic boundary based on the cases covered in the Y42 well.The first attempt was to consider that the well water temperature is constant in the case that the well water overflows.The second attempt was to consider that the well water temperature decays with time, which represents the process that the well water stops overflowing.We set the initial temperature of the upward flow and the initial temperature of the surrounding rock at 150 m to be 29 C and 16.7 C, respectively.The thermal conductivity of the surrounding rock was set to a maximum value of 5.77 WÁM À1 ÁK À1 and the initial adiabatic boundary was set to a possible maximum of 10 m.The results of the first and the second attempts show that the well water has a little thermal effect on the surrounding rock at 6 m and 3 m away from the casing, respectively.Thus, the boundary at 6 m from the casing was used as the adiabatic boundary in the subsequent simulations.
We tested the suitable thermal conductivity of the surrounding rock by modeling the temperature variation after pumping from the Y42 well.Since the pumping lasted for 2 days and the temperature was observed when the pumping was stopped (Chen 1981), we performed the modeling in two steps.We first modeled the temperature distribution of the surrounding rock after 2 days of heat conduction between the well water and the surrounding rock in the case of constant temperature of the water in wellbore.Taking the obtained rock temperature as the initial rock temperature and considering the decay of water temperature with time, we then modeled the variation in well water temperature.We set the thermal conductivity of the surrounding rock to be 1.65 WÁM À1 ÁK À1 and 5.77 WÁM À1 ÁK À1 , respectively.The corresponding variation of water temperature in the center of the well is shown in Figure 6.The figure shows that the results of 1.65 WÁM À1 ÁK À1 and 5.77 WÁM À1 ÁK À1 both partially fit the observation data, and we cannot determine the appropriate thermal conductivity that matches the observation data.In modeling the recovery of the post-seismic water temperature changes, we still set the thermal conductivity of the rock to be 1.65 WÁM À1 ÁK À1 and 5.77 WÁM À1 ÁK À1 , respectively.
According to the water level data of the Y42 well, the water level was 0.5 m seven days after the earthquake (see the inset of Figure 3f).As evidenced by the recovery of the water level, we assumed the duration of the overflow as 6.5 days, 6 days, 5.5 days, and 5 days, respectively.Using the assumed duration of the overflow and the parameters mentioned above, we modeled the recovery of post-seismic water temperature changes in two steps.We first modeled the temperature distribution of the surrounding rock caused by heat conduction between the water and the surrounding rock.The time of heat conduction here was set to be the same as the duration of the overflow.Taking the obtained rock temperature as the initial rock temperature and setting the water temperature to decay with time, we then modeled the variation of water temperature in the center of the well.
The results are shown in Figure 7.The figure shows that among the four assumed durations of the overflow, the result of 5 days is closer to the observation data (Figure 7d).Compared with the observation data, the simulated results after the Tangshan earthquake are larger than the observation with the increased time (Figure 7).The simulated result after the pumping is also larger than the observation data, whereas the difference between the simulated result and the observation data is small (Figure 6).This slight difference may be related to ignoring the vertical heat transfer process, which is consistent with the previous study (Shi et al. 2007).The vertical heat transfer after the pumping should be greater than that after the Tangshan earthquake because the vertical temperature difference after the pumping (which lasted for two days) might be larger than that after the earthquake (the well overflowed for $5 days).We consider that the difference between the simulated result and the observation data after the Tangshan earthquake may be related not only to ignoring the vertical heat transfer but also likely to be related to the assumption that the temperature of the upward flow was constant.The water temperature in the aquifer was relatively stable after the pumping (Figure 5b), whereas the temperature of the upward flow may decrease slightly over time after the earthquake (Figure 5a), resulting in a lower temperature.
Based on the modeling results of the Y42 well, we infer that the upward flow was maintained for $5 days.We have no clue about the maintaining time of other wells.If the upward flow was also maintained for 5 days in other wells, the upward release of groundwater caused by the earthquake would significant.In the case of the TSK, if the discharge in the first 5 days following the earthquake maintained the same as the discharge on the first day after the earthquake, an additional $400,000 m 3 of groundwater would be released in the first 5 days after the Tangshan earthquake.

Depth of the rising groundwater
We estimated the depth of the rising groundwater based on the residual temperature changes following the Tangshan earthquake.We performed the estimation using a simplified model in which we assumed that the residual changes were caused by the mixing of water at different temperatures in the deep aquifer and the shallow aquifer (the observed aquifer).The specific method is as follows: the mixing temperature between the shallow and the deep aquifers can be described by the following equation (Kitagawa and Koizumi 2000;Zhang et al. 2023): where T mix is the post-seismic temperature of the shallow aquifer, T s is the pre-seismic temperature of the shallow aquifer, D s is the discharge of the shallow aquifer, D d is the discharge of the deep aquifer, T d is the temperature of the deep aquifer.T d can be derived from equation (3) as: If T d is obtained, the depth (H) of the deep aquifer can be obtained by the following equation: where DT=Dh is the temperature gradient between the shallow aquifer and the deep aquifer, H 0 and T 0 is the depth and the temperature at the bottom of the shallow aquifer, respectively.Of the four wells with residual temperature changes (Figures 4b-d and 5a), we estimated the depth of the rising groundwater for the JS3, JS2 and Y42 wells, whereas not for the S049 well because the strata at the bottom of the well are reversed and we had no knowledge of the temperature gradient of the reversed strata.In the estimation, we took T s and T mix as the temperature at the depth where temperature changed the most; we took T 0 and H 0 as the values at the bottom of the residual temperature change.The values of the D s (23.6 m 3 /minute) and D d (56:51 m 3 /minute) are known for the TSK whereas are unknown for the JS3, JS2 and Y42 wells.We assigned D d to be 2.4 times D s in these three wells, similar to those of the TSK.The strata in the deep of the study area are mainly the Ordovician limestone (Figure 2), whereas the temperature gradient of the Ordovician limestone in the deep is less certain.The strata below the depth at which the residual temperature changes were recorded in the JS3 and the Y42 wells are the Ordovician limestone, and therefore we took the observed temperature gradient of the Ordovician limestone in the two wells as the value of DT=Dh: The temperature gradient below the depth at which the residual temperature changes were recorded in JS2 well is unavailable.We set the value of DT/Dh in the JS2 well to be consistent with that of the JS3 well, since the two wells are close.The parameters involved and the estimated depth of the rising groundwater are listed in Table 2.
The results show that the depth of the rising groundwater is between 1 and 2 km.This depth corresponds to the bottommost aquifer of the study area, i.e. the karst water in the Ordovician limestone (Figure 2).This aquifer has high water content and is located below the mining galleries, which has the potential to cause a significant rise in water levels.

Cessation of drainage?
After the Tangshan earthquake, the electric power system failed in Tangshan Mine, causing the cessation of drainage, which may be responsible for the groundwater level rise (Chen 1981;Zhang and Li 1981).We collected data on the discharge amount, the groundwater level, and the pumping amount in Tangshan Mine before and after the Tangshan earthquake, as shown in Figure 8.
The discharge amount increased threefold following the earthquake.After the increase in discharge, the bottom of the mine began to be submerged in groundwater.The groundwater level rose by an amplitude of 229 m in the first ten days.Then the rate of groundwater level rise was 1.2-2.53m/day.By 19 September 1976, the groundwater level (elevation of water surface) was À517.42 m, the highest groundwater level after the mine was submerged, and the total amplitude of groundwater level rise was 298 m (Yang and Yu 1981).The pumping in the mine was restored on 20 August 1976.The pumping was small at first, then gradually increased (Figure 8).
Based on the collected data, the topmost post-seismic groundwater level in Tangshan Mine was À517.42 m, which is lower than the depths of the SXS2, SW1, SW2, JS2, S049, and DC wells (Table 1).The hydraulic characteristics of the four wells and Tangshan Mine are uncertain, and if the comparison is made only according to the depth, then the water level rise in the four wells may not be caused by the cessation of drainage.The topmost groundwater level in Tangshan Mine is higher than the depth of the JS3 well and the screened section of the Y42 well, and we cannot determine whether the water level rise in the two wells was associated with the cessation of drainage.The discharge rise in TSK began to recover before the pumping was restored (Figure 8), suggesting that the discharge rise may not be caused by the cessation of drainage.

Coseismic static strain?
We estimated coseismic static strain using the COULOMB software (Lin and Stein 2004;Toda et al. 2005).We calculated the static strain induced by the main shock and the largest aftershock (Luanxian earthquake).The static strain was obtained based on the earthquake location, the focal mechanism, and the fault geometry (Table 3).The calculated coseismic static strain is shown in Figure 9a.The result indicates that the water level wells are located near the boundary of contraction and dilatation.The boundary of contraction and dilatation varies by the earthquake location, the focal mechanism, and the fault geometry.There are deviations in the locations and focal mechanisms of the main shock and the Luanxian earthquake given by various agencies.In addition, the fault geometries of the main shock and the Luxian earthquake were smaller than the theoretical predictions.Thus, based on the calculated result, we cannot clarify the relationship between the water level changes and the coseismic static strain.
To evaluate the relation between the water level changes and coseismic static strain, we further collected the coseismic deformation caused by the Tangshan   The earthquake location is from the earthquake catalog of China Earthquake Networks Center; the focal mechanism is from the Global CMT Catalog (https://www.globalcmt.org/CMTsearch.html);the fault geometry is based on the surface rupture zone (Yang 1982). of the seismogenic fault.The data indicate that the plate where the water level wells are located moved toward the northeast after the Tangshan earthquake.The surface rupture zone shows that the seismogenic fault is a right-lateral strike-slip, and the west side of the fault was raised after the Tangshan earthquake (Figure 1).Based on the coseismic deformation data, it is inferred that the plate where the water level wells are located was dilated after the Tangshan earthquake, and the dilatation does not match the water level rising (Figure 10), and therefore we suggest that the rising of groundwater level may not be caused by coseismic static strain.We compared the calculated coseismic static strain with the water level changes in wells outside the epicenter area of the Tangshan earthquake.We collected the water level changes from Wang and Yin (1982), which reported the spatial distribution of water level change on a large scale, including wells in Hebei, Tianjin, and Beijing.The collected water level changes and the calculated coseismic static strain are shown together in Figure 9b.It can be seen that a wide range of water level rise and fall were recorded outside the epicenter area of the Tangshan earthquake, whereas the locations of the water level rise and fall do not coincide well with the areas of contraction and expansion, respectively.

Liquefaction of sediments?
Following the Tangshan earthquake, large-scale liquefaction occurred around the earthquake, especially in the coastal plain adjacent to Bohai bay (Liu et al. 1982;Tang 1985).It is reasonable to believe that the water level rise may be related to the liquefaction of sediments.According to the observations described in Section 2, the observed sections of two wells (SW2 and DC) are the Quaternary sediments, the observed sections of two wells (JS3 and S049) are not specified, and the screened section of four wells (TSK, SXS2, SW1, and Y42) are not the Quaternary sediments.Thus, the model of liquefaction of sediments may work for the water level rise in the SW2 and DC wells, whereas it is not applicable in the TSK, SXS2, SW1, and Y42 wells.

Enhanced permeability?
We discuss the relationship between the groundwater level rising and the model of enhanced permeability by analyzing the water temperature data.Based on the water temperature data, distinct residual changes in groundwater temperature were recorded in the S049, JS3, JS2, and Y42 wells at certain depths more than ten days after the Tangshan earthquake.We consider that the residual changes were caused by the mixing of other sources of different temperature water.The estimation of the depth of the rising groundwater further indicates that the residual temperature changes may be caused by the mixing of water at different temperatures in the shallow and deep aquifers.Thus, we suggest that the hydraulic connection between the shallow and deep aquifers created by the Tangshan earthquake may be a plausible mechanism for the post-seismic rise of groundwater level.We further speculate that the hydraulic connection that linking the aquifers at different depths results from the created vertical permeability induced by dynamic strain.According to the study area's hydrogeological setting, the water levels of the shallow aquifers had dropped for many years before the Tangshan earthquake due to the coal mining in Tangshan Mine (Figure 3), resulting in higher pore pressures in deep aquifers than in shallow aquifers.In such a hydrogeological setting, when an earthquake causes the hydraulic connection between the shallow and deep aquifers, the groundwater in the deep aquifers will flow upward.The deep aquifers in the study area are karst aquifers in limestone, which have high water content and good connectivity.When the water in these deep aquifers is released, a significant rise in water level will be observed.It took $3.5 years for the groundwater levels to recover to the pre-seismic levels (Figure 3), suggesting that the recovery of the enhanced vertical permeability may be slow.
In summary of the discussion on the possible mechanism of the groundwater level rise, we cannot explain the post-seismic water level rise in most of the wells with the models of cessation of drainage and liquefaction of sediments.Nonetheless, we cannot rule out that these two models may play a role in some of the wells.We argue that the rising groundwater may not be caused by coseismic static strain, because the signs of coseismic deformation are inconsistent with the signs of the groundwater level changes.Based on the water temperature data, we suggest that the enhancement of hydraulic connection in the vertical direction between the shallow and deep aquifers created by the Tangshan earthquake may be a plausible mechanism for the groundwater level rise.

Conclusions
We revisit the groundwater level and temperature changes after the Tangshan earthquake of 1976 particularly near the epicenter.The water level data from eight wells (one is discharge data) were investigated.The water levels exhibited several meters of rising (the discharge increased by $3 times), which was followed by a subsequent gradual recovery.The water temperature data from 5 wells were investigated.The collected temperature-depth profiles of 4 wells exhibited residual changes at certain depths more than ten days after the Tangshan earthquake.We analyze the indications of the groundwater temperature data on the characteristics of the rising groundwater.
The modeling of the recovery of the water temperature changes in one of the wells shows that the upward flow was maintained for approximately 5 days.The analysis of the residual temperature changes indicates that the depth of the rising water was about 1-2 km.We analyze the origin of the groundwater level rise with several reported models, such as the cessation of drainage, the coseismic static strain, the liquefaction of sediments, and the permeability enhancement.The results indicate that the groundwater level rise is attributable to the earthquake-induced enhancement of the vertical hydraulic connection between the shallow and deep aquifers, leading to the upwelling of deep fluids that causes the groundwater levels to rise.

Figure 1 .
Figure1.Geological setting, the groundwater monitoring wells, the main shock of the 1976 M S 7.8 Tangshan earthquake, and the coseismic deformation.The WL and WT represent wells monitoring water level and temperature, respectively.The coordinate information of the faults, the surface rupture, the horizontal coseismic displacement, and the coseismic vertical displacement is fromLiu et al. (2013),Guo et al. (2011), Wan et al. (2017), and Guo et al. (1977), respectively.

Figure 2 .
Figure 2. Geological profile of the line A-B in Figure 1.The figure is modified from Chen 1981 and Gu and Zhou 1985.

Figure 3 .
Figure3.Groundwater level (discharge) changes before and after the Tangshan earthquake.The symbol ᭝ denotes the amplitude of the water level change.There are two kinds of water level units in the figure.One is the elevation of the water level (water surface).The other is the depth of water level, the distance between the well mouth and the water surface; the vertical coordinates under this unit are reversed, with a decreasing value denoting an increasing water level.

Figure 4 .
Figure 4. Temperature-depth profiles in four of the wells before and after the Tangshan earthquake (EQ).The Q, C, and O represent the Quaternary clay sand, the Carboniferous sandstone, and the Ordovician limestone, respectively.

Figure 5 .
Figure 5. Temperature-depth profiles in the Y42 well before and after the Tangshan earthquake (a), as well as before and after a pumping (b).The EQ denotes the major shock of the Tangshan earthquake.The Q, C, and O represent the Quaternary clay sand, the Carboniferous sandstone, and the Ordovician limestone, respectively.

Figure 6 .
Figure 6.Comparison between observed changes in temperature after pumping in the Y42 well (black dotted line) and those simulations under different thermal conductivities (blue and red lines).

Figure 7 .
Figure 7. Modeling temperature changes in the Y42 well after the Tangshan earthquake.The dark solid line is the temperature data observed at 150 m after the Tangshan earthquake.The letters 6.5 d, 6 d, 5.5 d, and 5 d attached to the dashed line represent the inferred duration of the postseismic overflow.

Figure 8 .
Figure 8. Variation of the groundwater level, the discharge amount, and the pumping amount in Tangshan Mine before and after the Tangshan earthquake.The data in this figure are from Yang and Yu (1981).
earthquake.The coseismic horizontal displacements obtained from triangulation observations are shown in Figure1.The direction of the coseismic horizontal displacements is northeast in the west of the seismogenic fault and southwest in the east

Figure 9 .
Figure9.Coseismic static strain induced by the main shock (Tangshan) and the most significant aftershock (Luanxian).The positive value stands for dilatation and the negative for contraction.(a) The coseismic static strain near the epicenter of the Tangshan earthquake, and the location of the groundwater level wells studied in this paper (triangles).(b) The coseismic static strain around the Tangshan earthquake on a large scale, the location of the groundwater level wells studied in this paper (triangles), and the location of the groundwater level wells fromWang and Yin (1982) (circles).

Figure 10 .
Figure 10.Conceptual model of the coseismic release of strain for compressional faulting.The figure is modified from Muir-Wood and King 1993.

Table 1 .
Basic information regarding the 9 wells near the epicenter of the 1976 M S 7.8 Tangshan earthquake.
‹ Yang and Yu 1981; › Gu and Zhou 1985; fi Data Collection Group of Tangshan Earthquake 1981, the data in which are daily values; fl Chen 1981; °Zhang and Li 1981; -Wang and Yin 1982;there is no water level data for the JS2 well, whereas there is water temperature data for this well.

Table 2 .
Parameters used for estimating the depth of the rising groundwater in the JS3, JS2 and Y42 wells.

Table 3 .
Parameters used for calculating the coseismic static strain.