Seismic Characterization of an Active Metamorphic Massif, Nanga Parbat, Pakistan Himalaya

A. Meltzer*, G. Sarker, L. Seeber, J. Armbruster, B. Beaudoin

A. Meltzer, G. Sarker, B. Beaudoin, Earth and Environmental Sciences, Lehigh University, Bethlehem, PA, 18015, USA. Sarker now at Texas A&M, 1000 Discovery Rd., College Station, TX 77840, USA. Beaudoin now at IRIS PASSCAL, New Mexico Tech, Socorro, NM, 87801 USA

L. Seeber, J. Armbruster, Lamont-Doherty Earth Observatory, Palisades, NY, 10964, USA

*To whom correspondence should be addressed.

Earthquakes recorded by a dense array of seismometers deployed around the Nanga Parbat massif in the Himalaya of Pakistan provide new insight into synorogenic metamorphism and mass flow during mountain building. Microseismicity beneath the massif drops-off sharply with depth defining a shallow transition (<5 km depth) between brittle failure and ductile flow. The base of seismicity bows upward beneath the massif mapping a thermal boundary with 3 km of structural relief over a lateral distance of 12 km. Low P and S wave velocity anomalies observed at the core of the massif extend to depth through the entire crust. The main locus of seismicity and low seismic velocities correlate with a region of high topography, rapid exhumation, high geothermal gradients, young metamorphic and igneous ages, and crustal fluid flow suggesting a genetic link between these phenomena in which hot rocks, rapidly advected from depth, are pervasively modified at relatively shallow levels in the crust.

Nanga Parbat lies at the western end of the Himalaya Mountain chain in the core of the western syntaxis. Its summit, sculpted from Pre-Cambrian Indian crust, at 8126 meters stands far above the surrounding Kohistan-Ladak Island arc terrane, which averages just 5 km in elevation. The world's largest relief, 7 km in 21 km horizontal distance, occurs between the peak and the Indus River running at its base. The rocks of Nanga Parbat, part of the Nanga Parbat-Haramosh massif, exhibit a polymetamorphic history acquired during Tertiary continental collision between India and Asia (1). Most recently, pervasive modification of the crust is documented by young (3 Ma) migmatites, the result of in situ melting of host rock under low-pressure (4-6 kbars) high-temperature (600-700°C) metamorphic conditions (2). Young (<1 Ma) intrusive granitic plutons and dikes record recent anatexis under vapor-absent isothermal decompression melting conditions (3). U-Pb and Ar-Ar, cooling ages, and fission tracks document rapid cooling and along with petrologic evidence and fluid inclusion data are consistent with rapid exhumation as high as 5-10mm/yr, over the last 3 million years (4). Modification of pre-existing terrane boundaries occurs locally where the Raikhot fault, a young active thrust structure, cuts the original collisional suture, the Main Mantle thrust (MMT) (5). On the western side of the massif, the MMT is absent, faulted away, and removed by erosion. The Raikhot fault juxtaposes Indian Pre-Cambrian gneisses against mafic rocks of the Kohistan Island arc captured during the collision of India with Asia and in places thrusts crystalline basement over Quaternary gravels (6). Boiling hot springs occur at the surface and there is ample evidence of recent hydrothermal alteration in numerous outcrops. Recent faulting, rapid exhumation, the presence of hot springs, young intrusive rocks, and young metamorphism all suggest a level of vigorous tectonic activity distinct from surrounding terranes and a potentially anomalous thermal structure beneath the massif. As part of a multidisciplinary study to understand the active tectonic processes responsible for crustal reworking at Nanga Parbat we deployed a dense seismic array to characterize seismicity at the massif and to determine crustal structure beneath the mountain.

While a growing number of studies have improved our understanding of the subsurface structure of mountain belts at regional scales (7), detailed characterizations of crustal structure and igneous and metamorphic processes in active orogens remain relatively uncommon. The high and rugged topography associated with these settings makes the dense deployments of seismic arrays required for detailed studies logistically challenging. Data recorded by our 60 station array provides a new opportunity to assess active processes associated with mountain building. Our array consisted of 6 broadband and 50 short period three component seismometers deployed in an approximately 75 x 75 km area in and around the massif (Figure 1). Stations were deployed on the west and east sides of the massif along the Indus and Astor river valleys while access to the interior of the massif was gained along glacial valleys. Station spacing within the massif varied from 1 to 6 km, and station elevations ranged from 1.1 to 4.3 km. Four additional broadband stations were deployed in a more regional context enlarging the total aperture of the network to ~300 km to provide better constraints for locating regional events. In a four-month time window we recorded over 1500 associated events (Figure 2), comprising teleseismic, regional, and local earthquakes. Primary source regions are the Pamir-Hindu Kush, the Karakoram, the Himalayan and Hazara arcs, and local seismicity beneath the massif itself. Abundant seismicity associated with the Pamir-Hindu Kush region originates 200-300 km northwest of Nanga Parbat at 100-300 km depth and serves as a beam source to illuminate the structure beneath the massif.

Our array recorded a high level of microseismicity at Nanga Parbat, generally between one and eight small magnitude (<1-2 m_b) events per day. During the four month deployment we recorded a total of 379 associated local events (events recorded by 3 or more stations). Seismicity is somewhat distributed along strike beneath the massif but exhibits a sharp drop-off west of the massif and a more gentle drop-off east of the main summit ridge crest (Figure 2). The adjacent Kohistan terrane is virtually aseismic. The sharp cut-off in seismicity to the west corresponds to the mapped trace of the Raikhot fault (3) where it is coincident with the Indus River. At Bunar Das, the main locus of seismicity cuts back into the interior of the massif following the Diamir shear zone. Brittle deformation is largely restricted to depths < 2 km below sea level (~5-6 km below the average topographic surface). Sixty-five percent of the local events occurred above sea level. Another 27% occurred between 0-2 km depth below sea level (bsl). The base of seismicity forms a prominent antiformal shape beneath the massif and exhibits considerable structural relief, approximately 3 km in a lateral distance of 12 km (Figure 2). The apex of this antiform occurs at 5 km depth bsl and is offset approximately 10 km northwest of the topographic ridge crest. The base of seismicity deepens to 8 km bsl to the NW and SE mapping a thermal boundary and a transition between brittle and plastic deformation that takes place over an ~3km thick zone. There is no seismicity deeper than 8 km depth (bsl) beneath the massif.

The transition from brittle to plastic deformation in the Earth occurs over a range of temperatures depending on lithologic composition, fluid content (%water), pressure, and strain rate (8). The dominant lithology of the Nanga Parbat-Haramosh massif is high-grade granitic gneiss and quartzofeldspathic biotite gneiss (9). For dry quartzofeldspathic rocks the brittle to ductile transition occurs between 300 to 450°C. Evidence from fluid inclusions in veins suggests the transition from brittle to plastic deformation occurs at the upper end of this temperature range at Nanga Parbat. Primary fluid inclusions from veins associated with brittle fractures cutting young structures in the core of the massif include a vapor rich phase that homogenized at temperatures up to 415°C, while fluid inclusions from ductilly deformed veins equilibrated at 450°C (10). High thermal gradients, ~60°C/km, in the shallow crust are also supported by petrologic and thermochronologic data (11). At these high thermal gradients, rocks pass through the 400-450° isotherm at shallow depths entering the ductile regime for quartzofeldspathic rocks. The high temperatures associated with the transition between brittle and plastic deformation imply there is little fluid interaction with the rock matrix within the brittle-ductile transition zone; as little as 0.2% water produces pronounced hydrolytic weakening lowering the temperature at which plastic deformation takes place (8). This result seems counter intuitive given an active hydrothermal system, including boiling hot springs, is observed at the surface. Yet, this conclusion is consistent with the high resistivities (1000 ohm-m) determined by magnetotelluric measurements at Nanga Parbat implying the crust beneath the massif is "dry" (12). Modeling of magnetotelluric data indicate that the maximum interconnected fluid fraction is 0.004% to 50 km depth beneath the massif. The apparent paradox is resolved by a relatively unique hydrothermal system at Nanga Parbat where boiling fluids (vapor plus liquid phase) dominate to ~2 km depth bsl and an unusual dry steam system (vapor phase only) dominates below these depths down to the brittle-ductile transition (10).

Most of the local seismic events recorded by our array have clean impulsive signals allowing high-quality focal mechanisms to be determined. While we recorded some thrust and some right-lateral strike-slip focal mechanisms, much of the observed seismicity exhibits extensional mechanisms and is associated with a set of shallow normal faults striking roughly parallel to the main massif and dipping south to southeast, back toward the summit (Figure 2). These events fall along planes that are relatively high angle, limited in lateral extent, and restricted to the shallow portions of the crust. These normal faults are consistent with extension above the doubly vergent thrust system bounding the massif (13). There is no evidence in either the structural mapping or in the seismicity to indicate the existence of low-angle crustal-scale normal faults typically associated with tectonic denudation in orogenic belts.

A cluster of seismicity is associated with the Tato hot springs including a swarm of 26 events in a 24 hour period. Much of this seismicity, when relocated using cross-correlation techniques, locates within ± 50m of each other defining a fault plane that may serve as a pathway for circulation of meteoric water to and from the surface. Stable-isotope results from rocks in the core of the massif and in fault zones adjacent to the massif indicate that a two component hydrothermal system is active, a shallow system dominated by fracture flow of meteoric waters driven by steep topographic gradients, and a deeper system involving magmatic - metamorphic fluids (14). While we see many seismic events with clean impulsive arrivals and easily identifiable P and S phases, others arrivals appear more harmonic, not unlike source signatures associated with geothermal systems (15) (figure 3).

While many of the igneous, metamorphic, and structural observations at Nanga Parbat could be explained by a large, young, intrusive body at depth, prominent but complicated S wave arrivals at stations throughout the array rule out the possibility of a substantial magma body beneath Nanga Parbat. This observation holds for both Hindu Kush events with a relatively vertical ray path beneath the massif sampling the entire crust and local events travelling more obliquely through the shallow crust.

A 3-D joint inversion for hypocenter location, Vp, and Vs shows that the seismic velocity structure beneath the Nanga Parbat massif is anomalously low, up to 10%, over lateral distances of 10-20 km (16) (Figure 4). Both P and S wave low velocity anomalies are observed within the core of the massif and extend to depth through the entire crust. In the shallow crust (<7 km depth) Vp ranges from 5.5-5.8 km/s and Vs from 3.2-3.5 km/s within the core of the massif compared to velocities of Vp=5.9-6.2 km/s and Vs=3.5-3.7 km/s in adjacent regions. Analyzing arrivals at stations deployed in glacial valleys as small linear arrays yields similar velocity values. At mid-crustal depths (7-20 km) Vp ranges from 5.6-5.75 km/s and Vs from 3.4 to 3.5 km/s within the massif compared to velocities of 6.3-6.5 km/s and Vs= 3.6-3.8 km/s in the surrounding area. At lower crustal depths (20-40 km), Vp at the core of the massif ranges between 5.6 and 6.0 compared to 6.4 to 6.6 km/s, while Vs within the massif ranges between 3.4 and 3.7 compared to Vs=3.7-3.9 km/s in the surrounding area. The uppermost mantle velocity, constrained from teleseismic P arrivals, is also about 10% slower within the core of the massif (Vp=7.5 km/s) compared to the surrounding mantle velocity (Vp=8.2 km/s). The depth to the Moho, as determined from best fit 1-D velocity model inversions and preliminary receiver function analysis, is 40 km. A crustal thickness of 40 km while thin by Himalayan standards in consistent with the few regional observations of crustal thickness in the western syntaxis (17).

Low velocities imply hot rocks at depth. We rule out significant fracture porosity or elevated pore pressure as an explanation for these low velocities. Confining pressure, along with magnetotelluric and petrologic data, rule out the possibility of an extensive connected fracture system at depth. In addition, significant attenuation of both P and S waves is observed in waves travelling through the massif compared to those that do not (18). The distribution of low velocities found at Nanga Parbat is consistent with the prediction of numerical models examining the evolution of the conductive geotherm during rapid advection of crustal rocks which indicate that Nanga Parbat should resemble a column of hot rock rising toward the surface (19). Our seismic results also suggest that the primary flow path of material into the massif is from depth rather than along a shallow detachment. Taken in total, our seismic results indicated that the dramatically high topography of Nanga Parbat sits above thin, hot, weak, crust.

The main locus of seismicity and regions of low seismic velocity correlate with the area of most rapid exhumation, cooling ages less than 2 Ma, young granitic dikes and plutons, young migmatites formed under low-P, high-T conditions, and high crustal resistivity suggesting a genetic link between these phenomena (22). All of these observations are restricted to the core of massif. The central part of the massif is bounded by two primary shear zones, the Raikhot-Diamir and Rupal, across which major discontinuities in cooling ages, metamorphic facies, and the age of igneous and metamorphic activity are observed (13, 20, 22). Within the core of the massif, biotite cooling ages (cooling through 340°C) are all younger than 2 Ma, with dates as young as 700-900Ka, peak metamorphism occurred at 4-6 kbars and 600-800°C, and igneous and metamorphic activity is all younger than 3 Ma, with much of the igneous activity younger than 1 Ma. West and southeast of the shear zones biotite cooling ages jump rapidly to 10-20 Ma (13). Peak metamorphism occurred at higher pressures (7-11 kbars) and lower temperatures (550-700°C) grading to conditions of ~11-14 kbars, 675-800°C typical of Himalayan metamorphism (20). Granitic intrusions immediately south of the Rupal shear zone are 19 Ma (23), again, more typical of Himalayan age intrusions.

Zeitler et al (22) and Koons (24) have proposed a thermal/mechanical/erosional model coupling exhumation, accentuated by river incision, to metamorphism and melting in the crust. The distribution of seismicity and low seismic velocities is consistent with this model in which rapid exhumation leads to pervasive modification of continental crust as hot rocks flow rapidly from depth to the surface. As this material is brought to the surface advection of isotherms results in low P and S wave velocities throughout the crust and elevates the position of the brittle-ductile transition causing it to bow, convex upward beneath the massif. Brittle fracturing predominates above 2 km depth (bsl). In this region meteoric water circulates and an active hydrothermal system exists. The transition from brittle to plastic deformation takes place over a 3 km thick zone beneath this region. Equating the base of seismicity to the 450°C isotherm yields a geothermal gradient of 56°C/km. At depths of 14-20 km beneath the massif, temperatures reach 600-800°C generating decompression melts. As continental crust passes through this region it is pervasively modified, all previous history is obliterated. The result is a sharp focused metamorphic anomaly separated in time and space from the main collisional event. The base of seismicity maps out an antiform beneath the massif, its shape reflecting the flow of material from depth. The apex of the antiform is offset from the topographic ridge crest and is consistent with particle paths developed in a two-sided orogen (25). Ultimately, this leads to the final conclusion that the remarkably high topography of the massif sits above hot, thin, weak crust, dynamically supported by the continuous flow of material from depth.

References and Notes

1. Chamberlain, Jan, and Zeitler in Tectonics of the Western Himalaya, L.L. Malinconico and R.J. Lillie, Eds. Geol. Soc. Am. Spec Pap. 232, (1989) pp. 23-32; Treloar et al., Tectonics, 8, 881 (1989); Smith, Chamberlain, and Zeitler, Journal of Geology, 102, 493 (1994); Wheeler, Treloar, and Potts, Geological Journal, 30, 349 (1995); Poage, Chamberlain, and Craw, American Journal of Science, in press; R.A.K. Tahirkheli et al., in Geodynamics of Pakistan, A. FARAH and K.A.,DeJong, Eds. (Geological Survey of Pakistan, Quetta, 1979), pp. 130-135; M.P.Coward et al., J. Geol. Soc. London 188, 385 (1987); H.A. Smith, C.P. Chamberlain, and P.K. Zeitler, Earth Planet. Sci. Lett. 113, 93 (1992).

2. Smith; Poage; Whittington.

3. P.K. Zeitler, C.P. Chamberlain, and H.A.Smith, Geology 21, 347 (1993); C. P. Chamberlain et al., J. Geophys. Res. 100, 439 (1995); D.M. Winslow et al., Geology 22, 1075 (1994); D.S. Schneider et al., Geology, 27,999 (1999).

4.

5. D.S. Schneider et al., Geology, 27,999 (1999); R.W.H. Butler and D.J. Prior, Nature 333, 247 (1988); I.P. Madin, R.D. Lawrence, and S. Ur-Rehman in Tectonics of the Western Himalaya, L.L. Malinconico and R.J. Lillie, Eds. Geol. Soc. Am. Spec Pap. 232, (1989) pp. 169-182.

6. R.W.H. Butler and D.J. Prior, Nature 333, 247 (1988); J.F. Shroder et al., in The Himalaya of northern Pakistan, in Malinconico, L.L. and Lillie, R.J., Eds. Geol. Soc. Am. Spec Pap. 232, (1989) pp. 275-294.

7.

8. R.H. Sibson, J. Geol. Soc. London 133, 191 (1977); R.H. Sibson, Bull. Seismol. Soc. A. 72, 151 (1982); C.H. Scholz, Geol. Rundsch 77, 319 (1988); R.O. Fournier, Geophys. Res. Lett. 18, 959 (1991); R.H. Sibson, Tectonophysics 211, 283 (1992); D. Hill, in Volcanic Seismology, P. Gasparini, R. Scarpa, and K. Aki Eds. IAVCEI Proceedings in Volcanology 3, (1992), pp.432-461.

9.

10. D.P. Craw et al., Earth Planet. Sci. Lett. 128, 169 (1994); D.P. Craw et al., Chemical Geology 142, 11 (1997).

11. P. K. Zeitler, C.P. Chamberlain, and H.A.Smith, Geology 21, 347 (1993); D.M. Winslow et al., Geology 22, 1075 (1994); D.M. Winslow, C.P. Chamberlain, P.K. Zeitler, J. Geol. 103, 395 (1995); A. Whittington, N. Harris, J. Baker, in Geol. Soc. London Spec Pub., P.J. Treloar and P. O'Brien, Eds. In press. A. J. Whittington, N. B. W. Harris, R. W. H. Butler, in A. McFarlane, R. Sorkhabi, J. Quade, Eds., Geol. Soc. Amer. Spec. Paper, in press; Poage, Chamberlain, and Craw, American Journal of Science, in press.

12. Park, S., and Mackie, R., GRL, 24, 2415 (1997); Park, S., and Mackie, R., Tectonophysics, xx, yyyy (1999).

13. D.S. Schneider et al., Geology, 27,999 (1999); Edwards et al., J.Geol. Soc. Lon, xx, yyyy, (1999).

14. C. P. Chamberlain et al., J. Geophys. Res. 100, 439 (1995); Poage, Chamberlain, and Craw, American Journal of Science, in press.

15. C.H. Cramer and S.R. McNutt, Bull. Seismol. Soc. A. 87, 1454 (1997); **** and K. Aki, Nature XX, xx (1995); B. Chouet in Volcanic Seismology, P. Gasparini, R. Scarpa, and K. Aki Eds., IAVCEI Proceedings in Volcanology 3, (1992), pp.133-156; V. Ferrazzini and K. Aki ibid., pp.168-189.

16. . Initial inversions using methods of Roeker et al., (J. Geophys. Res 98, 15,779 (1993)), included 239 high quality regional and local events and 3534 P and 3603 S arrivals. To take advantage of crossing rays from abundant local events initial inversions used a two layer structure (an upper layer from -4.2-10 km and a lower layer 10-50 km) over a half space. In the inversions, the earth beneath the target region is parameterized as a set of constant velocity blocks of uniform size (20 km X 20 km). The parameterization allows both vertical and lateral variations of the model. For blocks with resolution diagonals >0.9, the uncertainties associated with velocity are generally low, varying between 0.05-0.11 km/s. Synthetic tests for sensitivity using a 5% positive perturbation over a homogeneous background of Vp=6 km/s and Vs=3.5 km/s shows recovery of the initial model within 2-3% on average.

17.

18. Sarker, and Meltzer, JGR, in review.

19. D.P. Craw et al., Earth Planet. Sci. Lett. 128, 169 (1994).

20. Poage, Chamberlain, and Craw, American Journal of Science, in press.

21.

22. P.K. Zeitler, et al., Tectonics, (in review).

23. Schneider, D.A., Edwards, M.A., Kidd, W.S.F., Zeitler, P.K., and Coath, C., Earth Planet. Sci. Lett. 167, 121 (1999); Schneider, D.A., Edwards, M.A., Zeitler, P.K., and Coath, C., Contrib. Mineral. Petrol. 136, 273 (1999).

24. P.O. Koons, Tectonics, (in review).

25. P.O. Koons, Geology , 18, 679 (1990); P.O. Koons, Annu. Rev. Earth Planet. Sci. 23, 375 (1995); P.O. Koons J. Geophys. Res. 99, 12301 (1997).

26.This research is supported by the Continental Dynamics Program, Earth Sciences Division of the National Science Foundation, EAR-9418849. Technical support was provided by the IRIS PASSCAL Instrument Centers at Lamont-Doherty Earth Observatory and Stanford University. We also thank Carl Ebeling, Paul Friberg, Lindsay Hinck, Gulbar Hiyatt, Jeff Johnson, Aslam Mohammed (deceased), Michael Schoemann, Alexander Stine, and Chris Taylor for field assistance and logistical support.