Fluid–earthquake interplay, as evidenced by aftershock distributions or earthquake-induced effects on near-surface aquifers, has suggested that earthquakes dynamically affect permeability of the Earth’s crust. The connection between the mid-crust and the surface was further supported by instances of carbon dioxide (CO2) emissions associated with seismic activity, so far only observed in magmatic context. Here we report spectacular non-volcanic CO2 emissions and hydrothermal disturbances at the front of the Nepal Himalayas following the deadly 25 April 2015 Gorkha earthquake (moment magnitude Mw = 7.8). The data show unambiguously the appearance, after the earthquake, sometimes with a delay of several months, of CO2 emissions at several sites separated by > 10 kilometres, associated with persistent changes in hydrothermal discharges, including a complete cessation. These observations reveal that Himalayan hydrothermal systems are sensitive to co- and post- seismic deformation, leading to non-stationary release of metamorphic CO2 from active orogens. Possible pre-seismic effects need further confirmation.
Understanding fluid–earthquake interplay has long received a lot of attention1. For the last 20 years, there has been a growing number of evidence for fluid-driven earthquakes2. Deep fluids have been shown controlling aftershock distribution in the tectonic contexts of rifting3,4,5, subduction6,7, and reverse8 and strike-slip faulting9. Contemporaneously, numerous observations of earthquake-induced effects on near-surface aquifers have been reported10, including mainly changes in stream and spring discharge11,12,13, groundwater level14,15 and temperature16,17,18,19 in various tectonic contexts. These observations have suggested that earthquakes dynamically affect permeability of the Earth’s crust20,21. Carbon dioxide (CO2) emissions were observed in association with seismicity in the case of the Matsushiro swarm9 in Japan or, recently, at the Lassen volcano22 (Cascades Range, USA) and in the Eger Rift23 (Czech Republic), which suggests connection between the mid-crust and the ground surface through gas transport. However, such examples so far were only observed in the presence of magmatic activity.
The Himalayan orogen results from the India–Eurasia collision24, the Main Himalayan Thrust (MHT) accommodating at 2 cm year−1 half of the shortening between the two continents25. The largest earthquake in Nepal before 2015, the 1934 Bihar-Nepal earthquake (moment magnitude Mw~ 8.2) in Eastern Nepal, which claimed >15,000 lives, ruptured the MHT up to the surface over 150 kilometres along the Main Frontal Thrust26. Inter-seismic deformation is associated with intense background seismicity27, with 4–5 events of local magnitude ML > 5 per year, concentrated between 10 and 25 kilometres depth at the foot of the Himalayan topographic rise28. This region – the Main Central Thrust (MCT) zone29 – also exhibits numerous hydrothermal systems30 (Fig. 1). Following evidence of degassing from chemical and isotopic analysis of hot springs and rivers31,32, large CO2 emissions were discovered near hot springs30,33, with CO2 fluxes at places similar to diffusive fluxes from active volcanoes30,34. The seasonal and yearly stability of the soil–gas radon concentration time-series34,35, the invariant radon–CO2 fluxes relationship30,36, and the results of watering experiments36 at selected sites attest to the remarkable temporal stability of these hydrothermal systems, even during monsoon. These non-volcanic CO2 emissions are characterised by radiogenic helium, high radon content, and carbon isotopic compositions suggesting metamorphic CO2 production at >5 kilometres depth30,31,32,34. In the accepted conceptual model31,32, from the decarbonation source at >5 kilometres depth, CO2 percolates through fracture networks in the MCT zone, where it mixes with meteoric water. Near the water table, a fraction of CO2 may degas, and water discharges eventually at the surface as a hot spring. Degassed CO2 may also be transported faster to the surface through a network of interrelated faults without interaction with hydrothermal circulations37.
The Mw = 7.8 25 April 2015 Gorkha earthquake38 (Fig. 1) caused >9000 deaths, with fatality rates >1% in mountainous areas north of Kathmandu39, reaching 100% at some places40. It partly ruptured the MHT along a 120-kilometre-long segment east from the epicentre41, whose northern limit coincides with the MCT zone (Fig. 1), releasing only partially the stored elastic energy sufficient to produce a possible Mw~ 9 mega-earthquake42. The mainshock was followed by an intense aftershock sequence43, with >90 (ML > 5) events over a year. From 25 April 2015 to 31 December 2017, >10,200 (ML > 3), >1100 (ML > 4) and >135 (ML > 5) earthquakes occurred in Central Nepal (Fig. 1), leading to the most important seismic crisis in Nepal since 1934. The Gorkha earthquake is the first large Himalayan earthquake with seismic43 and Global Positioning System44,45 (GPS) data, as well as prior data on hydrothermal systems and CO2 emissions30.
In this paper, we report spectacular outbursts of CO2 and hydrothermal disturbances at several sites in Central Nepal separated by >10 kilometres that have been triggered by the 2015 Gorkha earthquake. These observations, in particular the first earthquake-induced gaseous changes in the absence of magmatic activity, reveal high sensitivity of Himalayan hydrothermal systems to co-, post- and possibly pre- seismic deformation, and non-stationary release of metamorphic CO2 from active orogens.
Outbursts of CO2 triggered by the Gorkha earthquake
New CO2 emissions were observed after the Gorkha earthquake, sometimes spatially associated with substantial changes from the pre-existing CO2 emissions (Table 1 and Supplementary Table 1). Because a large number of CO2 flux measurements (see Methods) were made before (n = 1718; December 2006 – January 2011) and after the earthquake (n = 1053; November 2015 – January 2018), a quantitative comparison of CO2 emissions before and after the earthquake can be undertaken in the Upper Trisuli valley. Spectacular effects were observed first in Syabru-Bensi, 56 kilometres east from the epicentre (Fig. 1), a site that had been extensively studied from 2006 to 2011. The main gas emission zone, located above the hot springs (Fig. 2; Supplementary Fig. 1), on the west bank of the Trisuli river, changed substantially. Significant CO2 fluxes and discharges appeared near the dwellings above the previously existing gas zone, in a cultivated area where flux level was close to local background levels before 2009. Hydrogen sulphide is now frequently smelled inside the houses. Total CO2 discharge in the area now reaches 1010 ± 110 mmol s−1 (3.8 ± 0.4 ton d−1), corresponding to an increase factor of 2.1 ± 0.3 (Fig. 2). Fluxes at two reference locations (K+6 and K+12; Fig. 2), regularly monitored since November 2015 (Fig. 3), peaked in November 2016 and, in January 2018, continued to remain 2-order of magnitude higher than their pre-seismic values. By contrast, the cavity characterised by the largest pre-earthquake CO2 fluxes ( >105 g m−2 d−1) showed one-order-of-magnitude smaller CO2 emission after the earthquake (Supplementary Fig. 4). The cavity fluxes returned to pre-earthquake values in January 2018, more than 2.7 years after the mainshock.
New CO2 emissions and changes in pre-existing CO2 emissions were also observed at other locations in the same valley (Supplementary Fig. 5). In Timure, nine kilometres to the north, CO2 emissions tripled after the earthquake (in November 2015 and January 2016) along a profile which had been measured precisely at different times before 2011 (Supplementary Fig. 6). Repeated measurements in September 2017 yielded a post-seismic CO2 emission increase of a factor of 8 ± 3 compared with pre-earthquake values. By contrast, at the Chilime hot spring site, seven kilometres to the west of Timure (Fig. 1), while CO2 fluxes remain significant, the total discharge is reduced by a factor of 5.2 ± 1.7 in January 2016, and by a factor of 25 ± 8 in January 2018. The CO2 emissions in Timure and Chilime thus depict opposite and unstabilized post-seismic responses > 2.7 years after the mainshock.
Two kilometres to the west of Chilime, at the Sanjen hydropower construction site (Fig. 1), spectacular bubbling was suddenly noticed, beginning of November 2015, in two 40-metre-deep piezometers whose water level was being monitored. When construction resumed during 2016, CO2 emissions appeared at several locations in a tunnel being excavated nearby (Supplementary Fig. 7 and Movie 1). The CO2 emission in the tunnel was still present in January 2018. The CO2 concentration in the air of the tunnel then ranged from 4 to 5 vol%, thus creating a major health hazard, and the total CO2 discharge was estimated in the tunnel to 580 ± 150 mmol s−1 (2.2 ± 0.6 ton d−1), hence of the same order as the main Syabru-Bensi discharge.
In the Budhi Gandaki valley, 16.5 kilometres to the epicentre (Fig. 1), the highest CO2 emissions ever reported in Central Nepal were observed in January 2017 on the partly flooded riverbank, where no phenomenon had been known to the locals before the earthquake (Supplementary Movie 2). The CO2 emission was still present in January 2018, with a total discharge higher than 1240 ± 330 mmol s−1 (4.7 ± 1.3 ton d−1), similar to the whole CO2 discharge observed in Syabru-Bensi. In addition to the flux from the bank, an innovative method was used to measure the CO2 flux from and through the hot water pond (see Methods and Supplementary Fig. 8). This is the most spectacular new CO2 emission observed so far in the Himalayas and elsewhere in the absence of volcanic activity. Together with the Upper Trisuli valley (Syabru-Bensi, Sanjen and Timure), such a persistent post-seismic CO2 outburst is unusual. For instance, small outbursts following the 2008 Wenchuan earthquake in China lasted only a few months46. In the Marsyandi valley, further west (Fig. 1), by contrast, no comparable phenomenon was observed and peak CO2 fluxes (5300 to 28,700 g m−2 d−1), while significant, are smaller than in the Upper Trisuli valley.
All these CO2 emissions were quantified in the field during the dry season or in absence of rain, as shown by the 2015–2017 rainfall data in Dhunche (seven kilometres southwest to Syabru-Bensi) and Timure (Fig. 3). At a given site, the CO2 flux data were obtained before and after the earthquake at about the same periods under similar meteorological conditions. Besides, the observed changes in the CO2 emission were done contemporary at several sites separated by more than 10 kilometres. All these care and observations preclude to these changes any environmental, meteorological or shallow origin.
Carbon isotopic ratio of the CO2 emissions (δ13C; see Methods) was systematically measured at all occurrences (Table 1 and Supplementary Table 1), giving the most comprehensive data set (n = 77) so far in a seismically active area without magmatic activity. Along the rupture zone, average δ13C values range from –6.9 ‰ to –0.1 ‰, and appear similar at the various sites of a given valley. CO2 emissions in Sanjen and Syabru-Bensi, although separated by eight kilometres, have comparable CO2 concentration (96–98%) and δ13C signature (from –0.9 ‰ to –0.7 ‰), precluding shallow sources, but instead suggesting similar CO2 source and transport from a crustal-scale reservoir. In addition, the δ13C values remain relatively stable after the earthquake (Supplementary Fig. 4), indicating that the earthquake revealed a pre-existing CO2 reservoir.
Hydrothermal unrest triggered by the Gorkha earthquake
The CO2 emissions were associated with changes of hydrothermal activity (Table 2 and Supplementary Table 2). In Syabru-Bensi, new hot springs appeared after the earthquake, with a few persisting in January 2018, > 2.7 years after the mainshock. The temperature of the main Syabru-Bensi hot spring (SBP0), which had been stable at 60.7 ± 0.1 °C for more than 12 years before the earthquake (Fig. 3), increased to 64.1 ± 0.3 °C after the earthquake (November 2015 – January 2016), along with a significant flow rate increase of 16 ± 1%. Water temperature peaked about 1 year after the mainshock. Water warming of SBP0 persisted more than 2 years after the mainshock, before starting to decrease to pre-earthquake values in January 2018, more than 2.7 years after. Other springs in Syabru-Bensi (SBB5) also showed warming, persisting in January 2018 (Supplementary Fig. 9). A slight pre-seismic water-cooling was possibly detected in Syabru-Bensi (Fig. 3). More significantly, three other hot springs of the same valley (Timure, Langtang and Chilime) also showed cooling a few weeks before the earthquake (Supplementary Fig. 9). In Tatopani (Budhi Gandaki valley), locals reported hot spring temperature decrease a few weeks before the earthquake. While co-, post- and sometimes pre- seismic changes in hot spring temperature have been reported in the literature16,19, a two-year-long or longer warming is unusual.
The most spectacular change in hydrothermal activity, however, is the complete cessation of the Chilime hot spring at the end of October 2015, after periods of unusual intermittence between April and June 2015 (Supplementary Fig. 10). Before the earthquake, this spring had a stable flow rate of > 5 L s−1; it was the pillar of local economy, being the most important in Central Nepal after the Kodari hot spring (Fig. 1). The village elders had previously reported spring intermittence at the time of the 1934 earthquake, but no cessation. Particularly impressive is the fact that it happened at about the time of the CO2 outburst in Sanjen.
Dissolved Inorganic Carbon (DIC) concentration and isotopic ratio (CDIC and δ13CDIC; see Methods) were systematically measured for all hot spring waters (Table 2 and Supplementary Table 2). In Syabru-Bensi, where the number of measurements is significant (n = 8), CDIC of the SBP0 hot spring increased by 29 ± 2% after the mainshock and, in September 2017, returned to values measured before the mainshock (Fig. 3), while Ca and Na concentrations remained similar within 5%. The δ13CDIC decreased also significantly (Supplementary Fig. 11). These observations suggest a larger amount of dissolved carbon in the SBP0 hot spring after the earthquake, which is compatible with the aforementioned increase in gaseous CO2 emissions. Available pH values (Supplementary Fig. 9) also show a slow return to pre-earthquake values, with anomalously high values before the earthquake, a fact to be interpreted with caution given the lack of additional geochemical information. In Syabru-Bensi, dissolved radon and radium concentrations in SBP0 and SBB5 hot springs changed after the earthquake (Supplementary Fig. 9). In January 2018, >2.7 years after the mainshock, the warmest spring in Syabru-Bensi (SBP0) returned to pre-earthquake conditions, while those more dependent on superficial effects (e.g., SBB5) did not yet.
First non-volcanic earthquake-induced gaseous changes
Hydrologic responses to seismic stimulation have been observed in numerous instances10, and correlated with Seismic Energy Density (SED) or Peak Ground Velocity (PGV) values. In Lassen (Cascades Range, USA) for example, a 2014 volcano-seismic swarm peaking at Mw = 3.85 at 5.7 kilometres distance, corresponding to SED∼0.2 J m−3 and PGV∼0.2 cm s−1, caused an outburst of geothermal fluids, explained by a two-fold permeability increase22. In the case of the Matsushiro earthquake swarm in Japan, modelling indicates the necessity of a 2-order-of-magnitude permeability increase with a small overpressure of only a few megapascals9. In Syabru-Bensi, the Gorkha earthquake produced SED∼63 J m−3 and PGV∼26 cm s−1 (see Methods; Supplementary Table 3), hence strong enough to affect the hydrothermal system, with several aftershocks maintaining high ground motion during the following 31 months (Supplementary Fig. 12). These estimates are confirmed by GPS time-series of the Chilime station44,45 (Fig. 1), giving PGV∼49 cm s−1 (Supplementary Fig. 13). While the first clearly documented instances in the Himalayas, our observed near-field CO2 outbursts and hydrothermal unrest remain compatible with previously compiled earthquake-induced changes (Supplementary Fig. 14). Our observed changes in the CO2 emissions are however the first earthquake-induced gaseous changes in a non-volcanic region.
The observations following the Gorkha earthquake give some indications that the standard model, where CO2 is degassed from hydrothermal waters, needs some modifications at least in part. Indeed, previous work in Syabru-Bensi34 identified large CO2 emissions without the presence of an important hot spring in the vicinity. Now, the evidence for an extended reservoir of crustal CO2 is overwhelming, which we accommodate with the following conceptual models (Fig. 4). In the case of the Trisuli valley (Syabru-Bensi and Timure) (Fig. 4a, b), hydrothermal circulations are important and increased after the earthquake. The main CO2 source can therefore be the hydrothermal circulations, with near surface degassing and subsurface accumulation in the whole fault zone. Possible additional CO2 can also be released in the footwall directly from the production source. The earthquake caused (Fig. 4b) an increasing water discharge, with increased CO2 degassing, or better communication to the surface of previously accumulated CO2. In the case of the Chilime valley (Chilime, Sanjen and Brapche) (Fig. 4c, d), however, the pervading presence of CO2, revealed by the earthquake, kilometres away from hot springs, is better explained if CO2 is directly accumulated from below, possibly through the water table, without advection by hydrothermal circulations.
Strikingly, CO2 emissions with similar isotopic anomaly were observed over the whole region, whenever faults, boreholes or a tunnel gave the opportunity. These observations also attest the presence of a large, relatively shallow, reservoir of CO2 in the Himalayan crust, suggesting that metamorphic CO2 produced at depth is huge, as independently shown by petrological estimates47,48, and unlikely sequestered. Nevertheless, as hot water could also be found when deep boreholes are available, the overwhelming presence of hot water could match the evidence of an extended CO2 reservoir, and the debate cannot be considered as closed.
The effects of the earthquake on our hydrothermal systems are not unexpected. Indeed, our observed CO2 emissions appear consistent with hydrothermal outburst effects, as observed in Lassen22, suggesting that apparent two-fold permeability increases and/or changes in hydrothermal pathways for pre-existing CO2 emissions could have been major effects of the Gorkha earthquake. First-order modelling indeed confirms that at least a doubling of permeability could explain CO2 flux increases (Supplementary Fig. 15). In addition, first-order modelling supports that a lasting permeability increase18 of 10–20% could accommodate the Syabru-Bensi hot spring warming and increased flow (Supplementary Figs. 16 and 17). These vertical permeability increases appear compatible with post-seismic afterslip45. The global pre-seismic water-cooling could be evidence for a pre-seismic dilatancy effect49 of the impending rupture zone.
Compared with reported earthquake-induced hydrological effects10,16,19,46, some of our observations are unusual: first, the six-month delay of the Sanjen CO2 outburst and the shut down of the Chilime hot spring, and second, the persistence (and rising at some places) > 2.7 years after in January 2018 of all the increased CO2 emissions and hydrothermal unrest. In Lassen22, the permeability increase was inferred to be maintained for 50–60 days. In the Napa valley13, a vertical permeability change, following diffusion of a pressure pulse, was proposed to explain post-seismic stream discharge, suggesting basin-scale fluid diffusivity of 0.1 m2 s−1. Taking this conservative value, for one-kilometre spatial scale, the timescale for fluid displacement is ~ 120 days, compatible with our six-month delay time. This suggests that the Sanjen CO2 outburst and the Chilime cessation likely result from hydrogeological fluid diffusion.
Other explanations are, however, possible. In Central Nepal, post-seismic deformation (afterslip on the MHT) and associated aftershocks remained active several months after the mainshock. About 6 months after the mainshock, contemporaneous with the Sanjen outburst and the Chilime cessation (Fig. 3a), the number of seismic events producing PGV>1 cm s−1 in Syabru-Bensi increased, and a linear trend in the GPS time-series, continuing to October 2017 about 2.5 years after the mainshock, was initiated. Thus, some post-seismic CO2 emissions and hydrothermal unrest may be related to changes in the state of post-seismic relaxation. Incidentally, the time of the Sanjen outburst and the Chilime cessation coincides with renewal of aftershock activity near Syabru-Bensi in the fall of 2015 (Fig. 3a), including an ML = 5.3 event a few weeks before. In addition, two ML > 4 events occurred at about the same time (end of October – beginning of November), less than 15 kilometres from Syabru-Bensi, producing PGV > 1 cm s−1 (Supplementary Table 3).
The Himalayan hydrothermal systems appear highly sensitive to small deformation rates and are therefore in near-critical condition. This suggests that post-seismic relaxation of co-seismic stress may result from pore pressure changes, and that metamorphic CO2 may in turn play a role in the installation of the next inter-seismic regime. Alternatively, the still evolving CO2 emissions in Syabru-Bensi, Timure and Chilime may also indicate a currently unstabilized system, able to diverge unpredictably. Given the issue of a pending mega-earthquake in the region42, long-term monitoring of CO2 emissions should seriously be considered as a chance to capture possible pre-earthquake signals.
In this paper, we have presented the first assessment of CO2 emissions triggered by a major earthquake, demonstrating the coupling between mechanical deformation and fluid transport properties at the crustal scale, highlighting that crustal deformation dynamically affects permeability during the seismic cycle. The large post-seismic CO2 emissions observed in the Narayani basin suggest non-stationary metamorphic CO2 production, and that its current estimate30,32 (>1.3 × 1010 mol year−1 with >(1.0 ± 0.2) × 108 mol year−1 from direct gaseous CO2 emissions), independently confirmed by petrological studies48, may be enhanced during a significant fraction of the seismic cycle. Metamorphic CO2 and its transport therefore emerge as an essential component of mountain build-up and the associated dynamics of large earthquakes.
Carbon dioxide flux measurement and mapping
The accumulation chamber method50 was used to measure surface CO2 flux and to quantify the total CO2 discharge of a given site and the associated uncertainties, whose assessment is based on numerous systematic tests and our 10-year experience34. The method is robust, even in remote locations51 and during the monsoon36, and allows the measurement of CO2 fluxes over more than five orders of magnitude (Supplementary Fig. 2). The increase in CO2 concentration in the chamber is measured using various portable infrared CO2 sensors (Testo™ 535, Testo AG, Germany; Airwatch™ PM 1500, Geotechnical Instruments Ltd., UK; Vaisala™ CARBOCAP® Hand-Held GM70, Finland), that are regularly inter-calibrated in the laboratory. The CO2 flux is expressed in grams per squared metre per day (g m−2 d−1). The total CO2 discharge, expressed in mol s−1 (or ton d−1), is estimated using the CO2 flux data-set by kriging and interpolation procedures34. CO2 fluxes (n = 1720) and total CO2 discharges obtained before the Gorkha earthquake (from 2006 to 2011) are published elsewhere30,33,34,51. Here we present for the first time CO2 fluxes (n = 1668) and total discharges obtained after the earthquake in the Marsyandi, Budhi Gandaki and Upper Trisuli valleys relying on seven measurement campaigns carried out in November 2015, in January, May and November 2016, in January and September 2017, and in January 2018. Pre-earthquake and post-earthquake campaigns were performed outside the monsoon periods to reduce the meteorological effects on CO2 flux data. Every uncertainty is given around one-sigma standard deviation (68% confidence level) and averages are arithmetic means except otherwise stated. Data are summarised in Table 1 and Supplementary Table 1.
Carbon dioxide flux measurement through a water layer
The bubbling CO2 flux from or through water was measured at the new degassing site of Machhakhola (Budhi Gandaki valley) in the following manner. A collecting container was installed upside down on the water with a pipe leading to an accumulation chamber installed on the ground nearby (Supplementary Fig. 8). Then, the flux was measured in this accumulation chamber as described above. This method yielded a minimum value to the large CO2 discharge observed in Machhakhola (Supplementary Movie 2).
Carbon isotopic composition of the gas phase
We sampled gas in the field using evacuated glass tubes. CO2 fraction of the gas sample was determined manometrically. The δ13C of CO2 of the gas, expressed in per mil relative to the standard values of Vienna Pee Dee Belemnite (V-PDB), was measured after off-line purification by mass spectrometry on a Finnigan™ MAT-253 mass spectrometer (Thermo Electron Corp., Germany) in CRPG (Nancy, France)32,34. External repeatability of a given sample was ±0.1 ‰. The twenty δ13C values measured before the Gorkha earthquake are published elsewhere30,33,34,52. Fifty-seven measurements were carried out after the earthquake from January 2016 to January 2018. Data are summarised in Table 1 and Supplementary Table 1.
Water temperature, pH and flow rate measurements
To measure water temperature of springs, we used thermometers (Generic TP101 Digital Thermometer, China) regularly inter-calibrated in the laboratory with a reference thermometer not currently used in the field (Digital Thermometer model 4400 Ertco Eutechnics, USA), and compared with high-precision (10−3 °C) and high-sensitivity (10−4 °C) thermometers (Seabird™ 39plus, Sea-Bird Scientific, USA). Due to the time response of the instruments, several minutes are needed to measure temperature. Experimental uncertainty of a given measurement was ±0.1 °C. The pH of the water in thermal springs was measured with various pH metres (H170 Portable pH metre, Hach, Germany; HI98107 and HI98130 pH metres, Hanna Instruments, USA), systematically recalibrated in the field using buffer solutions. Experimental uncertainty of a given measurement was ±0.1. The flow rate of thermal springs was determined using stopwatch and measured cylinders or buckets and was repeated at least three times. Data are summarised in Table 2 and Supplementary Table 2.
Carbon isotopic composition and DIC concentration in water
We sampled every water in the field using two to three glass screw cap vials of 12 millilitres volume each. DIC concentration (CDIC = [H2CO3] + [HCO3–] + [CO32–]) and its isotopic composition (δ13CDIC), expressed in mmol L−1 and in per mil relative to V-PDB, respectively, were determined using a gas chromatograph coupled to an isotope ratio mass spectrometer (GCIRMS, GV 2003, GV Instruments, UK) in IPGP (Paris, France). The whole procedure is described elsewhere53. The relative experimental uncertainty of CDIC was 1–2% and the experimental uncertainty of a given δ13CDIC measurement was ±0.1‰. For a given sample, final values correspond to weighted arithmetic averages of two to three measurements. The CDIC and δ13CDIC values measured before the Gorkha earthquake are published elsewhere30,31,32,33. A total of 71 measurements were carried out after the earthquake from January 2016 to January 2018. Data are summarised in Table 2 and Supplementary Table 2.
Water radon-222 and radium-226 concentration measurements
Dissolved radon concentration in water was measured in the field by emanometry in air54. After air–water equilibrium is reached by manual shaking, radon concentration is inferred from scintillation flask sampling and photomultiplier counting, as described elsewhere51. Radon concentration in water is expressed in Bq L−1. Experimental uncertainty ranged from 5 to 30%. The radium concentration in water was similarly measured in the IPGP laboratory after keeping the bottle closed for at least 50–80 days55. Expressed in mBq L−1, the experimental uncertainty was the same as for radon concentration in water. Here we present only the dissolved radon and radium concentrations in the main Syabru-Bensi hot springs (Supplementary Fig. 9).
Detection of thermal springs and CO2 degassing areas
The CO2 degassing areas were detected based on pervasive hydrogen sulphide odour, measurement of high radon-222 flux (radioactive gas of half-life 3.8 days), surface temperature anomalies, occurrence of water bubbles, presence of cavities, occurrence of inactive or active travertine deposits, and discussion with local people, or a combination of the above30,51. In remote places, the detection of previously unknown thermal springs relied on the use of hand-held thermal infrared cameras34,35,51 (model 880-V3 before 2015 and model 875-1i after, Testo™ AG, Germany).
Determination of SED, PGV and PDS
At the sites which experienced the most significant post-seismic changes in CO2 emissions and hot springs, we calculated SED56 and vertical PGV57 produced by the Gorkha earthquake and its main aftershocks. Empirical equations are used to estimate SED56 and PGV58. Peak Dynamic Stress (PDS) is estimated using: PGV × shear modulus (3 × 1010 Pa) / shear wave velocity (3500 km s−1). For the mainshock and six main aftershocks, values of SED, PGV and PDS at eight hydrothermal sites are gathered in Supplementary Table 3. In addition, SED, PGV and PDS produced at sites of the Upper Trisuli valley by three aftershocks located near Syabru-Bensi are also given.
Determination of Seismic Moment Release Rate
We calculated the Seismic Moment Release Rate (SMRR) produced by the aftershocks of the Gorkha earthquake using the NSC earthquake catalogue from 25 April 2015 to 31 December 2017. We converted local magnitude (ML) given by the catalogue into moment magnitude (Mw), using the following linear calibration: Mw = 1.109 ML – 0.626. The seismic moment was obtained using the classical relationship: log(M0) = 1.5Mw + 9.1. The SMRR, expressed in N.m per 20 days, was calculated for all seismic events around Syabru-Bensi and is shown in Fig. 3.
Modelling of earthquake-induced changes
To study the effect of permeability changes on CO2 flux before and after the earthquake, we used a 2-D model, described elsewhere37, in which a vertical fault (f) surrounded by two media (a and b) transports the gaseous CO2 to the surface by advection. Each medium is characterised by connected porosity (ε) and permeability (k). A pressure source is fixed at depth. Darcy’s law is defined according to a pressure distribution (p) that follows Δp2 = 0 (ref.59). An example calculation is shown in Supplementary Fig. 15.
To study the effect of permeability changes on spring water flow rate at the surface, we adapted the 2-D fault model described above37, to the calculation of water flow rate. We consider a vertical fault (f) driving the water flow to the surface, surrounded by two media (a and b). Each medium is characterised by permeability (k). Pressure source is fixed at depth. Solutions of the pressure distribution are expressed as a sum of exponential terms on horizontal axis, modulated by a sinusoidal signal along vertical axis37. An example of calculation is shown in Supplementary Fig. 16.
To study the effect of water flow rate changes on hot spring exit temperature, we relied on an analytical first-order model60 (Supplementary Fig. 17a). We consider a vertical conduit, characterised by perimeter p and thermal conductivity KT, that drives hot water from depth (z = h) to the surface (z = 0), with flow rate Q, density ρ and specific heat CP. Temperature at depth h is defined by T(h) = βh + T(0), where β is the thermal gradient, which is poorly constrained in this hydrothermal region61 and taken equal to 55–75 °C km−1. We consider a quasi-static state having characteristic length λT and we define θ = T–T(0) and α = ρCPQλT/(pKTh). Using the initial conditions, the differential equation αh(dθ/dz) + θ = –βz has the following solution: θ = αhβ(1–e–1/α). Examples of calculation of the water temperature as a function of water flow rate, for two values of the thermal gradient, are shown in Supplementary Fig. 17b.
The data that support the findings of this study are available in the article, in Supplementary Information, and from the corresponding author upon request (email@example.com).
Irwin, W. P. & Barnes, I. Tectonic relations of carbon dioxide discharges and earthquakes. J. Geophys. Res. 85, 3115–3121 (1980).
Miller, S. A. Chapter 1 - The role of fluids in tectonic and earthquake processes. Adv. Geophys. 54, 1–46 (2013).
Noir, J. et al. Fluid flow triggered migration of events in the 1989 Dobi earthquake sequence of Central Afar. Geophys. Res. Lett. 24, 2335–2338 (1997).
Miller, S. A. et al. Aftershocks driven by a high-pressure CO2 source at depth. Nature 427, 724–727 (2004).
Terakawa, T., Zoporowski, A., Galvan, B. & Miller, S. A. High-pressure fluid at hypocentral depths in the L’Aquila region inferred from earthquake focal mechanisms. Geology 38, 995–998 (2010).
Waldhauser, F., Schaff, D. P., Diehl, T. & Engdahl, E. R. Splay faults imaged by fluid-driven aftershocks of the 2004 Mw 9.2 Sumatra-Andaman earthquake. Geology 40, 243–246 (2012).
Wang, Z., Huang, W., Zhao, D. & Pei, S. Mapping the Tohoku forearc: Implications for the mechanism of the 2011 East Japan earthquake (Mw 9.0). Tectonophysics 524-525, 147–154 (2012).
Sibson, R. H. An episode of fault-valve behaviour during compressional inversion? — The 2004 MJ6.8 Mid-Niigata Prefecture, Japan, earthquake sequence. Earth Planet. Sci. Lett. 257, 188–199 (2007).
Cappa, F., Rutqvist, J. & Yamamoto, K. Modeling crustal deformation and rupture processes related to upwelling of deep CO2-rich fluids during the 1965–1967 Matsushiro earthquake swarm in Japan. J. Geophys. Res. 114, B10304 (2009).
Wang, C.-Y. & Manga, M. Earthquakes and water. Lect. Notes Earth Sci. 114, 235 (2010).
Muir-Wood, R. & King, G. C. P. Hydrological signatures of earthquake strain. J. Geophys. Res. 98, 22035–22068 (1993).
Sato, T., Sakai, R., Furuya, K. & Kodama, T. Coseismic spring flow changes associated with the 1995 Kobe earthquake. Geophys. Res. Lett. 27, 1219–1222 (2000).
Wang, C.-Y. & Manga, M. New streams and springs after the 2014 Mw6.0 South Napa earthquake. Nat. Commun. 6, 7597 (2015).
Roeloffs, E. A. Persistent water level changes in a well near Parkfield, California, due to local and distant earthquakes. J. Geophys. Res. 103, 869–889 (1998).
Brodsky, E. E., Roeloffs, E., Woodcock, D., Gall, I. & Manga, M. A mechanism for sustained groundwater pressure changes induced by distant earthquakes. J. Geophys. Res. 108, 2390 (2003).
Mogi, K., Mochizuki, H. & Kurokawa, Y. Temperature changes in an artesian spring at Usami in the Izu Peninsula (Japan) and their relation to earthquakes. Tectonophysics 159, 95–108 (1989).
Ma, Z. et al. Earthquake Prediction: Nine Major Earthquakes in China (1966–1976) (Springer-Verlag, Berlin Heidelberg, 1990).
Wang, C.-Y., Manga, M., Wang, C.-H. & Chen, C.-H. Transient change in groundwater temperature after earthquakes. Geology 40, 119–122 (2012).
Cox, S. C. et al. Changes in hot spring temperature and hydrogeology of the Alpine Fault hanging wall, New Zealand, induced by distal South Island earthquakes. Geofluids 15, 216–239 (2015).
Manning, C. E. & Ingebritsen, S. E. Permeability of the continental crust: Implications of geothermal data and metamorphic systems. Rev. Geophys. 37, 127–150 (1999).
Manga, M. et al. Changes in permeability caused by transient stresses: Field observations, experiments, and mechanisms. Rev. Geophys. 50, RG2004 (2012).
Ingebritsen, S. E. et al. Hydrothermal response to a volcano-tectonic earthquake swarm, Lassen, California. Geophys. Res. Lett. 42, 9223–9230 (2015).
Fischer, T., Matyska, C. & Heinicke, J. Earthquake-enhanced permeability – evidence from carbon dioxide release following the ML 3.5 earthquake in West Bohemia. Earth Planet. Sci. Lett. 460, 60–67 (2017).
Avouac, J.-P. Mountain building, erosion, and the seismic cycle in the Nepal Himalaya. Adv. Geophys. 46, 1–80 (2003).
Ader, T. et al. Convergence rate across the Nepal Himalaya and interseismic coupling on the Main Himalayan Thrust: Implications for seismic hazard. J. Geophys. Res. 117, B04403 (2012).
Sapkota, S. N. et al. Primary surface ruptures of the great Himalayan earthquakes in 1934 and 1255. Nat. Geosci. 6, 71–76 (2013).
Pandey, M. R., Tandukar, R. P., Avouac, J.-P., Lavfi, J. & Massot, J. P. Interseismic strain accumulation on the Himalayan crustal ramp (Nepal). Geophys. Res. Lett. 22, 751–754 (1995).
Bollinger, L., Avouac, J.-P., Cattin, R. & Pandey, M. R. Stress buildup in the Himalaya. J. Geophys. Res. 109, B11405 (2004).
Upreti, B. N. An overview of the stratigraphy and tectonics of the Nepal Himalaya. J. Asian Earth Sci. 17, 577–606 (1999).
Girault, F. et al. Large-scale organization of carbon dioxide discharge in the Nepal Himalayas. Geophys. Res. Lett. 41, 6358–6366 (2014).
Becker, J. A., Bickle, M. J., Galy, A. & Holland, T. J. B. Himalayan metamorphic CO2 fluxes: Quantitative constraints from hydrothermal springs. Earth Planet. Sci. Lett. 265, 616–629 (2008).
Evans, M. J., Derry, L. A. & France-Lanord, C. Degassing of metamorphic carbon dioxide from the Nepal Himalaya. Geochem. Geophys. Geosyst. 9, Q04021 (2008).
Perrier, F. et al. A direct evidence for high carbon dioxide and radon-222 discharge in Central Nepal. Earth Planet. Sci. Lett. 278, 198–207 (2009).
Girault, F. et al. The Syabru-Bensi hydrothermal system in central Nepal: 1. Characterization of carbon dioxide and radon fluxes. J. Geophys. Res. Solid Earth 119, 4017–4055 (2014).
Richon, P. et al. Temporal signatures of advective versus diffusive radon transport at a geothermal zone in Central Nepal. J. Environ. Radioact. 102, 88–102 (2011).
Girault, F., Koirala, B. P., Perrier, F., Richon, P. & Rajaure, S. Persistence of radon-222 flux during monsoon at a geothermal zone in Nepal. J. Environ. Radioact. 100, 955–964 (2009).
Girault, F. & Perrier, F. The Syabru-Bensi hydrothermal system in central Nepal: 2. Modeling and significance of the radon signature. J. Geophys. Res. Solid Earth 119, 4056–4089 (2014).
Avouac, J.-P., Meng, L., Wei, S., Wang, T. & Ampuero, J.-P. Lower edge of locked Main Himalayan Thrust unzipped by the 2015 Gorkha earthquake. Nat. Geosci. 8, 708–711 (2015).
Sapkota, S. N., Bollinger, L. & Perrier, F. Fatality rates of the Mw~8.2, 1934, Bihar–Nepal earthquake and comparison with the April 2015 Gorkha earthquake. Earth Planets Space 68, 40 (2016).
Martin, S. S., Hough, S. E. & Hung, C. Ground motions from the 2015 Mw 7.8 Gorkha, Nepal, earthquake constrained by a detailed assessment of macroseismic data. Seismol. Res. Lett. 86, 1524–1532 (2015).
Grandin, R. et al. Rupture process of the Mw = 7.9 2015 Gorkha earthquake (Nepal): Insights into Himalayan megathrust segmentation. Geophys. Res. Lett. 42, 8373–8382 (2016).
Stevens, V. L. & Avouac, J.-P. Millenary Mw > 9.0 earthquakes required by geodetic strain in the Himalaya. Geophys. Res. Lett. 43, 1118–1123 (2016).
Adhikari, L. B. et al. The aftershock sequence of the April 25th 2015 Gorkha-Nepal earthquake. Geophys. J. Int. 203, 2119–2124 (2015).
Galetzka, J. et al. Slip pulse and resonance of Kathmandu basin during the 2015 Mw 7.8 Gorkha earthquake, Nepal imaged with geodesy. Science 349, 1091–1095 (2015).
Gualandi, A. et al. Pre- and post-seismic deformation related to the 2015, Mw7.8 Gorkha earthquake, Nepal. Tectonophysics 714-715, 90–106 (2017).
Zheng, G., Xu, S., Liang, S., Shi, P. & Zhao, J. Gas emission from the Qingzhu River after the 2008 Wenchuan Earthquake, Southwest China. Chem. Geol. 339, 187–193 (2013).
Groppo, C., Rolfo, F., Castelli, D. & Mosca, P. Metamorphic CO2 production in collisional orogens: Petrological constraints from phase diagram modeling of Himalayan, scapolite-bearing, calc-silicate rocks in the NKC(F)MAS(T)-HC system. J. Petrol. 58, 53–83 (2017).
Rapa, G. et al. Titanite-bearing calc-silicate rocks constrain timing, duration and magnitude of metamorphic CO2 degassing in the Himalayan belt. Lithos 292–293, 364–378 (2017).
Scholz, C. H., Sykes, L. R. & Aggarwal, Y. P. Earthquake prediction – Physical basis. Science 181, 803–810 (1973).
Chiodini, G., Cioni, R., Guidi, M., Raco, B. & Marini, L. Soil CO2 flux measurements in volcanic and geothermal areas. Appl. Geochem. 13, 543–552 (1998).
Girault, F., Koirala, B. P., Bhattarai, M. & Perrier, F. Radon and carbon dioxide around remote Himalayan thermal springs. Geol. Soc. Lond. S. P. 451, 155–181 (2016).
Becker, J. A. Quantification of Himalayan Metamorphic CO 2 F luxes: Impact on Global Carbon Budgets. PhD Thesis, University of Cambridge (2005).
Assayag, N., Rivé, K., Ader, M., Jézéquel, D. & Agrinier, P. Improved method for isotopic and quantitative analysis of dissolved inorganic carbon in natural water samples. Rapid Commun. Mass Spectrom. 20, 2243–2251 (2006).
Girault, F., Perrier, F. & Przylibski, T. A. Radon-222 and radium-226 occurrence in water: A review. Geol. Soc. Lond. Spec. Publ. 451, 131–154 (2016).
Perrier, F., Aupiais, J., Girault, F., Przylibski, T. A. & Bouquerel, H. Optimized measurement of radium-226 concentration in liquid samples with radon-222 emanation. J. Environ. Radioact. 157, 52–59 (2016).
Wang, C.-Y. & Manga, M. Hydrologic responses to earthquakes and a general metric. Geofluids 10, 206–216 (2010).
Elkhoury, J. E., Brodsky, E. E. & Agnew, D. C. Seismic waves increase permeability. Nature 441, 1135–1138 (2006).
Akkar, S. & Bommer, J. J. Empirical equations for the prediction of PGA, PGV, and spectral accelerations in Europe, the Mediterranean Region, and the Middle East. Seismol. Res. Lett. 81, 195–206 (2010).
Mosley, R. B., Menetrez, M. Y., Snoddy, R. & Brubaker Jr., S. A. The influences of diffusion and advective flow on the distribution of radon activity within USEPA’s soil chamber. Environ. Int. 22, S521–S533 (1996).
Vasseur, G., Michard, G. & Fouillac, C. Contraintes sur la structure profonde et le fonctionnement du système hydrothermal de Chaudes-Aigues (France). Bull. Hydrogéol. 4, 3–17 (1997).
Derry, L. A., Evans, M. J., Darling, R. & France-Lanord, C. Hydrothermal heat flow near the Main Central Thrust, central Nepal Himalaya. Earth Planet. Sci. Lett. 286, 101–109 (2009).
Byrdina, S. et al. Dipolar self-potential anomaly associated with carbon dioxide and radon flux at Syabru-Bensi hot springs in central Nepal. J. Geophys. Res. 114, B10101 (2009).
Le Fort, P. Himalayas: Collided range, present knowledge of continental arc. Am. J. Sci. A275, 1–44 (1975).
Kotarba, M. Hydrogeological investigations in Seti Khola and Trisuli thermal springs areas (Nepal Himalayas). Geologia 12, 37–51 (1986).
We warmly thank Soma Nath Sapkota and the Department of Mines and Geology for enthusiastic support over the years. We are indebted to Buddha Lama, Niraj Jairu and Ohm Tharum for assistance in the field, and to Nabin Tamang and Dawa Tamang for giving access to their property in Syabru-Bensi. Svetlana Byrdina is thanked for preparing the CO2 flux data obtained in 2006–2007. We are thankful to Jean-Philippe Avouac, Steve Ingebritsen, Yves Gaudemer, Damien Jougnot, Christian Fouillac, Pascal Bernard, Hélène Lyon-Caen and Marc Chaussidon for fruitful discussions. Thomas Rigaudier is thanked for the measurement of carbon isotopic ratios in CRPG. Carine Chaduteau is thanked for the measurement of carbon isotopic ratios in IPGP. Narayan P. Malla is thanked for giving access to the Rasuwagadhi Hydroelectric Project. Hélène Bouquerel, François Prevot, Marie Bouih and Roxane Ferry are thanked for their participation in the fieldwork. Financial support for this work was essentially provided by public and private funding of the IPGP team “Physics of Natural Sites”, partially complemented in 2015 by a TelluS-SYSTER INSU CNRS funding to F.G. and in 2016 by a CNRS attribution to C.F.-L. This is IPGP contribution number 3942.
The authors declare no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Girault, F., Adhikari, L.B., France-Lanord, C. et al. Persistent CO2 emissions and hydrothermal unrest following the 2015 earthquake in Nepal. Nat Commun 9, 2956 (2018). https://doi.org/10.1038/s41467-018-05138-z
This article is cited by
Communications Earth & Environment (2022)
Communications Earth & Environment (2022)
SAR and optical images correlation illuminates post-seismic landslide motion after the Mw 7.8 Gorkha earthquake (Nepal)
Scientific Reports (2022)
Radon signature of CO2 flux constrains the depth of degassing: Furnas volcano (Azores, Portugal) versus Syabru-Bensi (Nepal Himalayas)
Scientific Reports (2022)
Nature Communications (2021)