Assessing whether the 2017 Mw 5.4 Pohang earthquake in South Korea was an induced event

See allHide authors and affiliations

Science  01 Jun 2018:
Vol. 360, Issue 6392, pp. 1007-1009
DOI: 10.1126/science.aat6081

Triggering quakes in a geothermal space

Enhanced geothermal systems (EGSs) provide a potentially clean and abundant energy source. However, two magnitude-5 earthquakes recently occurred in South Korea during EGS site development. Grigoli et al. and Kim et al. present seismic and geophysical evidence that may implicate the second of these earthquakes, which occurred in Pohang, as an induced event. The combination of data from a local seismometer network, well logs, satellite observations, teleseismic waveform analysis, and stress modeling leads to the assessment that the earthquake was probably or almost certainly anthropogenically induced. The possibility remains that the earthquake occurred coincidentally at the EGS site location, but the aftershock distribution and other lines of evidence are concerning for future development of this geothermal resource.

Science, this issue p. 1003, p. 1007


The moment magnitude (Mw) 5.4 Pohang earthquake, the most damaging event in South Korea since instrumental seismic observation began in 1905, occurred beneath the Pohang geothermal power plant in 2017. Geological and geophysical data suggest that the Pohang earthquake was induced by fluid from an enhanced geothermal system (EGS) site, which was injected directly into a near-critically stressed subsurface fault zone. The magnitude of the mainshock makes it the largest known induced earthquake at an EGS site.

The injection of fluid into reservoir rocks, which facilitates oil and gas recovery, enhances geothermal systems, and aids in the disposal of wastewater and CO2 gas, also has a small chance of inducing earthquakes [e.g., (14)]. Empirical and theoretical relationships exist to connect the maximum magnitude of an induced earthquake and the injected fluid volume (2, 5). The magnitude of an induced earthquake may be tectonically controlled—for example, owing to the presence of a fault suitably oriented for slip under a given stress field that is also located adjacent to one or more injection or production wells (6, 7). The earthquake nucleation itself might be controlled by the injection (6). Previously observed magnitudes of induced seismicity at enhanced geothermal system (EGS) sites have been relatively small (8); the largest local magnitude (ML) reported was 3.4, in Basel, Switzerland (9), although a much larger, possibly induced earthquake has been reported in the Cerro Prieto geothermal field, Mexico, a tectonically active area (10).

A moment magnitude (Mw) 5.4 earthquake occurred at the Pohang EGS site in southeastern Korea on 15 November 2017. The earthquake was the most damaging and the second-largest in magnitude in South Korea since the first seismograph was installed in 1905. The earthquake injured 90 people, and the estimated property damage was US$52 million (11). We present evidence that the Pohang earthquake was the largest induced event to have occurred at any EGS site worldwide. Moreover, this event indicates that injected fluid volumes much smaller than predicted by theory can trigger a relatively large earthquake, at least under the right set of conditions.

The Korean Peninsula lies within the Eurasian Plate (Fig. 1), although it was composed of continental magmatic arcs at a plate boundary until the early Tertiary period (~30 million years ago) (12). North-northeast (NNE)–striking strike-slip faults and NNE- to NE-striking normal faults developed predominantly in southeastern Korea and adjacent offshore areas when the East Sea (or Japan Sea) opened as a back-arc basin in the early to middle Tertiary (~30 to 15 million years ago), with the coetaneous formation of smaller-scale basins, including the Pohang (1214). Some of these faults have been reactivated as strike-slip and thrust faults in the current compressional regime (1315). The axes of compression determined by focal mechanism solutions indicate shallow plunges to the ENE throughout the southern Korean Peninsula (15, 16).

Fig. 1 Tectonic map of northeast Asia.

Saw-toothed lines with solid teeth denote subduction zones. The broken line with open teeth represents an incipient subduction zone (30). EU, Eurasian Plate; NA, North American Plate; PS, Philippine Sea Plate; PA, Pacific Plate.

The Pohang basin consists of nonmarine to deep marine sedimentary strata dating to the Miocene (~20 million years ago), with a basement composed of Cretaceous to Eocene sedimentary and volcanic rocks and late Paleozoic to Eocene granitoids (1719) (Fig. 2A). The geology of the Pohang EGS site comprises (from top to bottom) Quaternary alluvia (<10 m thick), Miocene semi-consolidated mudstone (200 to 400 m thick), Cretaceous to Eocene sedimentary and igneous rocks (~1000 m thick), and Permian granodiorite with gabbroic dykes (1921) (Fig. 2B).

Fig. 2 Geologic map and column of the Pohang basin.

(A) Map showing rocks and faults of the Pohang basin and adjacent area. One permanent seismic station operated by the Korea Meteorological Administration (PHA2) and our eight temporary seismic stations are represented by a dark blue square and green triangles, respectively. The green square denotes the site of the Pohang enhanced geothermal system (EGS). The geologic map was compiled from (18, 19). (B) Geologic column of the Pohang EGS site with injection well. The geologic column was compiled from (19, 20).

One vertical injection well (4346 m deep; PX2) and another deviated production well (4362 m deep; PX1) were drilled into Permian granodiorite with gabbroic dykes for the EGS, with an expected electricity production of 1.2 MW (21). The PX1 well, only 6 m from PX2 at the surface, is 600 m northwest of PX2 at the bottom (Fig. 3). The drilling began in September 2012 and was completed in November 2015. No earthquakes with ML > 2.0 were recorded within 10 km of the Pohang EGS site between 1978 and 2015 (22); a total of six earthquakes with ML 1.2 to 1.9 were detected in the area between 2006 and 2015. To further examine the seismicity around the EGS site, we improved the earthquake catalog by applying a matched filter to continuous waveforms (23) recorded by a permanent seismic station (PHA2; Fig. 2A), located about 10 km north of the EGS site, during the period between 1 January 2012 and 14 November 2017. Once detected, waveforms were visually inspected for time differences between P- and S-wave arrivals consistent with a source at Pohang (~1.54 s). The matched-filter analysis found no noticeable earthquakes at the EGS site before the completion of drilling. We detected a total of 148 earthquakes by the match filtering that all occurred after the completion of the drilling, including four earthquakes with ML > 2.0.

Fig. 3 Spatial distribution of epicenters and hypocenters of the 2017 Pohang earthquake sequence.

(A) Epicenters of six foreshocks (red circles), mainshock (red star), and 210 aftershocks (black open circles) recorded in the first 3 hours after the mainshock. The location of the Pohang EGS is indicated by a green square. Blue triangles represent our eight temporary seismic stations. The red beach ball represents the source mechanism of the mainshock. Black beach balls show the focal mechanism solutions of representative aftershocks. Numbers above beach balls are local magnitudes. The black beach balls of the ML 2.4 and 3.5 aftershocks, recorded 1 and 3 days after the mainshock, respectively, show strike-slip faulting. X−X′ and Y−Y′ denote the locations of the cross sections shown in (B) and (C), respectively. (B and C) Hypocentral distributions of earthquakes, projected onto vertical planes along the lines X−X′ (B) and Y−Y′ (C) shown in (A). The red beach ball in (B) represents the focal mechanism of the mainshock projected onto a vertical cross section. PX1 and PX2 denote production and injection wells, respectively. Other symbols are the same as for (A).

Hydraulic stimulation began on 29 January 2016 and comprised four phases of injection with a total volume of 12,800 m3 at injection rates of 1.00 to 46.83 liter s−1 (Fig. 4). Fluid was injected into both PX1 (phases 2 and 4) and PX2 (phases 1, 3, and 4). To investigate the relationship between seismicity and fluid injection, we used data on regional earthquakes detected by the matched filter together with earthquake data provided by the Ministry of Trade, Industry, and Energy (MTIE), Republic of Korea (23). We did not detect noticeable microearthquakes before the drilling. The timing of the earthquakes coincides with that of fluid injections. The first reported hydraulic stimulation (phase 1) was carried out between 29 January 2016 and 20 February 2016, followed by three additional phases of fluid injection (Fig. 4). Each injection phase was accompanied by intense seismic activity that started only a few days after injection. Microseismic activity decreased rapidly after the termination of fluid injection. The magnitudes of induced earthquakes tend to increase with the net volume of injected fluid. After a ML 3.1 earthquake on 15 April 2017, which was the largest felt event near the EGS site before the ML 5.4 Pohang earthquake, we deployed eight temporary seismic stations around the EGS site. Each standalone station was equipped with a three-component velocity–type short-period sensor. These sensors record continuous seismic data at a sampling frequency of 200 Hz. Installation was completed on 10 November 2017. All of them are still in active operation.

Fig. 4 History of fluid injection volume.

Volume of injected and bleed-off fluid (left axis) and net fluid volume (right axis) as a function of time. Red circles denote times of seismic events from data provided by the MTIE (63 events; magnitude information is only available for M > 1). Dark blue circles denote seismic events determined by matched-filter analysis (148 events). The right axis also gives the local magnitudes of seismic events.

The ML 5.4 mainshock occurred at the Pohang EGS site on 15 November 2017 and was preceded and followed by foreshocks and aftershocks, respectively, all of which were well recorded by our local seismic array at distances of 0.6 to 2.5 km from the mainshock epicenter (Figs. 2A and 3). We precisely relocated this earthquake sequence with the hypoDD software package (23, 24) (fig. S2). We plotted the spatial distribution of six foreshocks (ML ≤ 2.6), the mainshock, and 210 aftershocks that occurred within 3 hours of the mainshock (Fig. 3). The first two foreshocks occurred about 9 hours before the mainshock; the remaining four preceded it by 6 to 7 min. Most hypocentral depths fell in the range of 4 to 6 km, and the mainshock depth was about 4.5 km. Of note, the hypocenters of the foreshocks and mainshock were located immediately adjacent to the bottom of PX1. The hypocentral depths of the Pohang earthquake sequence are shallower than those of most earthquakes in the Korean Peninsula, which tend to occur at depths of 10 to 20 km (25). For comparison, the largest instrumentally recorded earthquake in South Korea, the 2016 Gyeongju event (ML 5.8), had a hypocentral depth of about 14 km (26). The spatiotemporal distribution of hypocenters indicates that the rupture plane consisted of two segments, a main southwestern segment and a subsidiary northeastern segment (Fig. 3 and fig. S3). The aftershocks tended to occur earlier on the main segment than on the subsidiary segment (fig. S4). This observation, together with the locations of the foreshocks and mainshock on the main segment, suggests that the main segment ruptured earlier than the subsidiary one. By statistical plane-fitting using MATLAB software (Mathworks), we determined the best-fit orientations of the main and subsidiary rupture planes to be N36°E (strike)/65°NW (dip) and N19°E/60°NW, respectively (fig. S3). These orientations are consistent with the nodal planes determined by focal mechanism solutions (Fig. 3). The main segment shows thrust faulting with a minor strike-slip component, whereas the subsidiary segment is dominated by strike-slip faulting with a minor dip-slip component. The compression axis trends E–W or ENE–WSW with a shallow plunge, similar to that of other earthquakes in South Korea (15, 16). The locations of the foreshocks and mainshock, at the bottom of the injection well, suggest that fluid was injected directly into the fault zone.

The temporal relationship between seismicity and fluid injection, the spatial relationship between the hypocenters and the EGS site, and the lack of seismicity in the area before the EGS was established all suggest that the Pohang earthquake was induced. Furthermore, the immediate response of seismicity to fluid injection and the locations of the foreshocks and mainshock at the bottom of the injection well suggest that fluid was injected directly into a fault zone. The fault plane inferred from the spatial distribution of hypocenters and focal mechanism solutions strikes NE and dips to the NW (fig. S4A), similar to Quaternary thrust faults in southeastern Korea. In fact, a magnetotelluric survey of the EGS site detected a low-resistivity feature that could be the fault zone, striking NE and dipping to the NW (23, 27) (fig. S5). Reverse slip along a subsurface fault is consistent with the current stress field. All of these lines of evidence indicate that the Pohang earthquake was “almost certainly induced” in Frohlich et al.’s system of assessment (28). If we use McGarr’s (2) equation for the relationship between the maximum magnitude and the total volume of injected fluid, about 4.7 × 106 m3 of injected fluid would be required to induce an Mw 5.4 earthquake, which is more than 810 times the fluid volume injected at the Pohang EGS site. The permeability structure of fault zones is highly heterogeneous, and patches or layers of clay-rich gouge within the fault core act as barriers to fluid flow (29). The pore pressure thus can locally reach a critical value for earthquake nucleation after a relatively small volume of fluid is injected, depending on fault zone structure. Our results imply that if fluid is injected directly into a near-critically stressed fault, it can induce a larger earthquake than current theory predicts. Detailed investigation of the geological, geochemical, and geophysical properties of the Pohang EGS site will improve our understanding of earthquake-inducing processes.

Supplementary Materials

Materials and Methods

Figs. S1 to S5

References (3138)

References And Notes

  1. The Korea Meteorological Administration catalog is available at
  2. Supplementary materials.
Acknowledgments: We thank Representative S. S. Kim of the National Assembly and the Ministry of Trade, Industry and Energy, Republic of Korea, for providing fluid injection data. We are grateful to the Korea Meteorological Administration for providing continuous waveforms used in the study. We also thank two anonymous reviewers for their constructive comments. Funding: This work was supported by the Nuclear Safety Research Program through the Korea Foundation of Nuclear Safety (KoFONS) using financial resources granted by the Nuclear Safety and Security Commission (NSSC) of the Republic of Korea (no. 1705010). Author contributions: K.-H.K.: conceptualization, formal analysis, funding acquisition, methodology, original draft, and review and editing. J.-H.R.: conceptualization, formal analysis, funding acquisition, methodology, original draft, and review and editing. Y.K.: investigation, methodology, validation, and review and editing. S.K.: data curation, formal analysis, investigation, and software. S.Y.K.: data curation, formal analysis, investigation, and software. W.S.: data curation, formal analysis, investigation, and software. Competing interests: The authors declare no conflicts of interest. Data and materials availability: Our earthquake catalog, including the earthquake source parameters (locations and times) and waveforms used in this paper, is available at​. Earthquake waveform data recorded at PHA2 can be acquired from the National Earthquake Comprehensive Information System, Korea Meteorological Administration (; last accessed April 2018). Continuous data may be obtained from the website on request.

Correction (31 May 2018): Two data points, respectively from 31 March 2016 and 6 April 2016, were removed from Fig. 4 as they were borderline detections based on the matched filtering. This was done out of caution and reduces the number of matching events from 150 to 148. In addition, the labels for the wells in Fig. 4 were corrected. The figure legend was changed to indicate the source of the seismicity data, which was provided by MTIE and not directly by the EGS developer.

View Abstract


Navigate This Article