From coseismic offsets to -block

George A. Thompsona,1 and Tom Parsonsb,2

aDepartment of Geophysics, Stanford University, Stanford, CA 94305; and bUnited States Geological Survey, Menlo Park, CA 94025

Edited by Mark H. Thiemens, University of California at San Diego, La Jolla, CA, and approved August 8, 2017 (received for review June 29, 2017) In the Basin and Range extensional province of the western United pattern than the topography. Combined GPS and InSAR ob- States, coseismic offsets, under the influence of gravity, display servations (5, 6) show broad areas of uplift and subsidence in the predominantly subsidence of the basin side (fault hanging wall), with vicinity of a chain of magnitude (M) ∼ 7 earthquakes that struck comparatively little or no uplift of the mountainside (fault footwall). A central Nevada between 1915 and 1954 (7, 8). The areal extent of few decades later, geodetic measurements [GPS and interferometric vertical deformation observed over this period spans across both synthetic aperture radar (InSAR)] show broad (∼100 km) aseismic uplift basins and ranges and can be 100–200 km in width (Fig. 2). symmetrically spanning the fault zone. Finally, after millions of years However, a different deformation mode is observed during and hundreds of fault offsets, the blocks display large uplift individual earthquakes (coseismic period). Highly concentrated and tilting over a breadth of only about 10 km. These sparse but subsidence (10–15 km wide) of the basins is observed (9–13) robust observations pose a problem in that the coesismic uplifts of whereas the ranges are stable, or rise very slightly (Fig. 3). the footwall are small and inadequate to raise the mountain blocks. To These observations are based on leveling surveys recorded in address this paradox we develop finite-element models subjected to Nevada before and after the 1954 M = 7.2 Fairview Peak and extensional and gravitational forces to study time-varying deforma- M = 6.5 Dixie Valley earthquakes, and in Idaho before and tion associated with normal faulting. Stretching the model under grav- after the 1983 M = 6.9 Borah Peak earthquake. Clearly, the evo- ity demonstrates that asymmetric slip via collapse of the hanging wall lution from seismic slip on faults to the final topographic signature is a natural consequence of coseismic deformation. Focused flow in the of crustal extension passes through different temporal phases. We upper mantle imposed by deformation of the lower crust localizes develop a physical hypothesis based on isostasy and crustal litho- uplift, which is predicted to take place within one to two decades after spheric rheology, and test it using finite-element models. each large earthquake. Thus, the best-preserved topographic signature of earthquakes is expected to occur early in the postseismic period. Conceptual Model EARTH, ATMOSPHERIC,

Conceptually, the process begins with elastic rebound, which in AND PLANETARY SCIENCES rifting | finite-element modeling | earthquakes | crustal deformation | normal faulting requires a horizontal withdrawal of mass from Basin and Range the fault zone and thus an unloading of the footwall (14, 15) (Fig. 4). Isostatic forces and inflow of lower crust and upper mantle then ith its repeating series of parallel ranges rising sharply cause a postseismic upward-directed bulging, focused initially near Wabove deep valleys, the is one the fault zone and later, after several decades, becomes broader. of Earth’s most distinctive (Fig. 1). However, this active The basin side is repeatedly dropped with each large earthquake. ∼10–15-Ma-old (1) landscape bears little resemblance to present- Isostasy in the Earth’s crust means that its shallowest elastic day measures of deformation. During earthquakes, the basins drop parts float in the denser, hotter, and more ductile substrate below. (2), but the ranges show little or no rise. Satellite ranging [GPS, These floating masses find their equilibrium heights over time, with interferometric synthetic aperture radar (InSAR)] shows areas of that duration depending on the rheology of the substrate. Basic broad uplift and subsidence that do not directly correspond with isostatic calculations, wherein the sudden mass change after an topography. We use numerical modeling techniques constrained earthquake (1-m slip event) is treated in theoretical floating ver- with observations over three periods: coseismic (deformation tical columns, predict an upward directed force of ∼70 MPa during earthquakes), postseismic (deformation after earthquakes), and multiseismic (topography resulting from repeated earth- Significance quakes). We further constrain our models with the concept of isostasy, the gravitational equilibrium that must ultimately result Observations at different times during extensional faulting ’ from any vertical change to the Earth s surface. The addition of cycles show dramatically different deformation. Available this constraint allows us to produce a single unifying model that coseismic and postseismic observations bear little resemblance explains variations in fault-related deformation over time. to the topography of rifted zones, yet this topography is the end result of repeated earthquakes. During earthquakes, and Observations during periods decades later, there is little evidence of The Basin and Range province topography was built in response to flank, or range-front deformation, yet strong bending and broad crustal extension between the ranges to the uplift of these features ultimately define extended terranes. west, and the to the east. Repeated earthquakes Numerical modeling incorporating gravity and the principle of on dipping (45°–60°) faults offset crystalline bedrock ranges (fault isostatic balance predicts strong vertical forces during the first footwalls) against basins filled with sediment (fault hanging walls) decade after a significant earthquake that are preferentially (Fig. 1), meaning that roughly half of the resulting deformation is focused beneath range fronts. We conclude these forces are vertical. Cumulative slip on these faults (faults that accommodate responsible for characteristic rift topography. This hypothesis is extension are called normal faults) can be up to ∼10 km in mag- testable with intermediate period geodetic observations that nitude, and is the result of many hundreds of earthquakes. We thus are rare for extensional earthquakes. term this deformation as multiseismic. Observations and modeling results show mountain blocks that are tilted or bent upward over a Author contributions: G.A.T. and T.P. designed research, performed research, analyzed width of about 10 km (Fig. 1). Wider mountain blocks (∼20 km) data, and wrote the paper. display more bending, which requires permanent rock deformation, The authors declare no conflict of interest. probably in part expressed by myriad minor faults and joints (3, 4). This article is a PNAS Direct Submission. Deformation measured over the span of decades (postseismic 1Deceased May 12, 2017. period) using remote sensing methods shows a very different 2To whom correspondence should be addressed. Email: [email protected].

www.pnas.org/cgi/doi/10.1073/pnas.1711203114 PNAS Early Edition | 1of6 Downloaded by guest on September 25, 2021 -120˚ -118˚ -116˚ 5 Warner Range Surprise Valley

0 Depth (km) 40˚

-5

0510 15 Distance (km)

e

g e 38˚ n a g R n

a e

l R a 3.4 g e B uneroded surface B’ n v i t a 2.8 h w VE 2.5:1 g i h a 2.2 N S B B’ 1.6

Elevation (km) 1.0

Syncline axis Normal Dip fault Dir. 0102030 Distance (km)

Fig. 1. Characteristic basin and range topography, with narrow chains of ranges interspersed by basins. (Upper, Inset) Cross-section model of the Warner Range and its bent range front (3, 35). (Lower) The range blocks are bent as a result of repeated faulting along their fronts rather than being tilted. In this example, a single block is faced on either side by normal faults, has no significant internal faulting, and both sides are bent upward (36).

applied across the narrow zone of postfaulting mass change (Fig. responsible for the signature deformation that characterizes the 4). Additionally, transferred from primary earthquake slip Bain and Range province and other rift zones. (16) onto the continuation of a zone into the deep crust can cause afterslip and additional deformation (17). Numerical Model A distinction regarding continental dip–slip faults that cause ele- A numerical model that is allowed to freely respond to forces vation changes emerges because of isostasy. Faults form and fail as a that represent our best understanding of those acting in the result of differences in the magnitudes and directions of the principal Earth’s crust can yield revealing results provided the solutions stresses in the crust, typically at a frictionally dependent angle in- are not overly guided, or overly sensitive to parameter choices. clined 30°–60° to the least compressive stress direction (18). In Gravity is the driving force acting on the crust, and the primary compressional (thrust) faulting, the least stress is vertically inclined, feature of extensional terranes is that their boundaries can ex- and the greatest stress is oriented horizontally, whereas strike–slip pand laterally, enabling the crust to collapse under its weight. We faulting regimes have horizontally directed greatest and least stresses. develop a simple finite-element model under these conditions In extensional settings, the least stress is oriented horizontally, and (21), with layers based on physical properties of rock subjected to greatest stress is the vertical gravitational load. This means that faults gravity, and with boundaries that expand laterally at observed that change the elevation profile also create postearthquake verti- rates (22). The model has three layers (Fig. 5): a brittle, break- able upper-crustal layer (15 km thick), a 15-km-thick ductile cally directed isostatic forces, which add stress to the fault that just lower crustal layer where strain is accommodated primarily by failed. Upward-directed isostatic forces result from unloading of the flow, and a 170-km-thick upper mantle layer that is deep enough fault zone in extensional settings and downward-directed isostatic to prevent the bottom model boundary from affecting results. A forces result from loading of the fault zone in compressional settings breakable upper-crustal layer is necessary to best replicate the (19). Both actions thus increase the differential stresses that lead to long-term plastic deformation of a fractured and faulted shallow fault failure. Isostatic loading is expected to be more rapid and im- crust (21). We find the best fit to geodetic observations if the portant in extensional settings because higher heat flow, thinner lower-crustal layer is stronger than the upper mantle, as have crust, and increased lower crust and upper mantle mobility (20) many other Basin and Range studies (6, 23–30). We embed a 45° enable a more immediate response than in convergent terranes dipping normal fault through the upper crust, and permit shear where a cooler, more rigid rheology can resist isostatic forces. on a deeper extension of the fault into the lower crust. Isostatic reloading of faults has energy balance implications We can use this model to study effects of faulting over the because dip–slip faults are stressed both by tectonic plate mo- coseismic, postseismic, and multiseismic time frames. The first tions and by the act of faulting itself. The additional isostatic simulation has a model with a locked normal fault, gravitational stress must be shed as aftershocks, postseismic slip, by increasing load, and expanding boundaries. After extensional stresses are magnitudes of subsequent earthquakes, or a progressive decrease allowed to build, we release the fault, which slips, and the in recurrence intervals. Our conceptual model therefore hypothe- footwall collapses as observed following real earthquakes (Fig. sizes that additional isostatic stressing acting on normal faults is 5B). The second simulation is similar to the first except that the

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1711203114 Thompson and Parsons Downloaded by guest on September 25, 2021 1.5 A C 1.0 500 ≤-2.0 0.0 ≥2.0 uz (mm/yr) 0.5 0.0 Dixie Valley Dixie Valley a -0.5 fault trace 250 b -1.0 -1.5

1.5 0 B 1.0 0.5 GPS Vertical velocity (mm/yr) 0.0 Distance (km) from 38˚N -250 -0.5 -1.0 -1.5 -400 -200 0 200 400 -100 -50 0 50 100 150 200 Distance (km) from -118˚E 0.0

D EARTH, ATMOSPHERIC, 350 Valley -0.5 E AND PLANETARY SCIENCES

-1.0 InSAR

Pleasant Valley -1.5 fault trace 300 -2.0 -2.5

-3.0 250 -100 -50 0 50 100 150 200

Line of sight velocity (mm/yr) Distance (km) from -118˚E

Distance (km) from 38˚N 200 ≤-1.5 0.0 ≥1.5

Dixie Valley uz (mm/yr) fault trace

-50 0 50 100 Distance (km) from -118˚E

Fig. 2. Geodetic measures of central Basin and Range Province uplift during the past two decades (5, 6). Uplift tends to be symmetric around relatively recently ruptured faults (1954 in this case), which differs from the sharp topographic signal marked by range-front faults (Fig. 1). A and B show west-to-east uplift profiles vs. distance along the transects marked a and b on the map panel in C. An InSAR profile is also shown in D, and a close-up contour map around the Dixie Valley fault is shown in E.

fault is prescribed to slip 1 m, simulating an M ∼ 7 earthquake. deformation observed on coseismic, postseismic, and multiseismic The fault is then locked and the postseismic response of the time scales, but a mystery still persists. lithosphere to fault slip and isostasy is tracked over time (Fig. 5 C–E). Predictions from the model include an early phase of Summary and a Challenge asymmetric uplift focused primarily under the footwall (Fig. 5C) When in the seismic cycle do isostatic forces express themselves? that diminishes with time. By 60 y after faulting, which is equiv- Our numerical model predicts that the strongest uplift should alent to the present-day observation period for GPS and InSAR occur beneath the ranges almost immediately after faulting, with (5, 6) after the 1954 central Nevada earthquakes, a broad upwarp a short (∼10-y) delay required for ductile rocks to begin flowing. is predicted (Fig. 5D) that matches observations (Fig. 5E). In the The early postseismic uplift is predicted, but to the authors’ final simulation, we allow the fault to slip continuously, mimicking knowledge has not yet been observed. Geodetic uplift observa- repeated earthquakes over many millions of years. This test rep- tions do not cover this period in the Basin and Range province, licates the distinctive rifting signature of uplifted and bent ranges but instead verify the upwarping that affects a wider region than (Fig. 5F). A simple finite-element model can thus reproduce an individual faulted range front.

Thompson and Parsons PNAS Early Edition | 3of6 Downloaded by guest on September 25, 2021 40.0˚N . 1954 y F le -200 al V ie Di x

-150 Elevation change (cm) 16 Dec. M=6.5

.

F

. -100 Mt

w

o -50 39.5˚

ainb -40

R t l u -20 24 Aug. M=6.5 a

F

A’ -10 6 July M=6.2 A

w

e i 0 v r i a +10 F 16 Dec. M=7.2 +20 39.0˚ -119.0˚E -118.5˚ -118.0˚ -117.5˚ -117.0˚ 500 1954 Fairview Peak EQ 0 A A’ -500

-1000 Elevation change: 1954-1955 (mm) -1500 -118.25˚ -118.15˚ -118.05˚ -117.95˚ Longitude 500 1983 Borah Peak EQ 0 44.5˚ 1983 -500 Fig. 3. Observations of coseismic deformation. In these -1000 44.0˚ three examples where there exist pre- and post-

Vertical change: 1933/48-1984 (mm) earthquake leveling lines (10, 12), it is evident that the -1500 28 Oct. M=6.9 coseismic vertical deformation consists almost entirely of -35 -30 -25 -20 -15 -10 -5 0 5 subsidence of the hanging walls (basins). Virtually no Distance from Lost River fault (km) 43.5˚ range uplift is observed, which is counter to the bent and -114.5˚ -114.0˚ -113.5˚ elevated ranges observed in the topography (Fig. 1).

Our long-term simulations show that, to get the narrow basin– by slip on faults. However, these simulations cannot pinpoint when range topography and bent ranges, the isostatically derived to- in the seismic cycle isostatic forces are expressed as fault slip. The pography of narrow basins and uplifted ranges must occur entirely reason for this is that we cannot know the frictional locking

Fault locked: stress increases A B Earthquake relieves & causes distributed stress & concentrates elastic extension extension strain

normal faultdensity of upper crust upper crust decreases in fixed

Previously distributed lower crust density decrease is now narrowly concentrated

Fig. 4. Conceptual model that describes faulting of an Bending elastic crust under extension floating in a denser, ductile C D Uplift substrate. (A) A downward-directed gravitational load Hanging wall falls due combined with expanding lateral boundaries causes to gravity elastic stretching of the brittle upper crust in an extending tectonic setting. The net density is decreased in any fixed reference crustal section as a result. (B) Frictional resistance on a normal fault is overcome Sesimic slip adds and an earthquake occurs. That rupture concentrates the stress to lower previously distributed density decrease into the volume -70.1 MPa crustal shear Flow that contains the fault. (C) The hanging wall (basin side) Unweighting stress stress (=> Afterslip collapses during the earthquake, and the volume con- reduction shearing) taining the fault tries to float upward because it is sud- denly less dense. (D) Ductile rock flows in beneath the faulted region as a new isostatic balance is achieved, and the faulted region bulges upward.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1711203114 Thompson and Parsons Downloaded by guest on September 25, 2021 UpperUpper crust: Elastic- plastic fracturing solid A: Model construction solid Lower crust:

Creeping solid Applied extensional displacement Upper crust: Elastic-plastic fracturing solid B: Coseismic deformation 50

0

Upper Mantle: Creeping solid -50

-100

Elevation change (cm) -150 0 5 10 15 20 25 30 Distance (km) from -118.25˚E

C: Crust and mantle displacement rate t=10 yrs D: Mantle displacement rate t=60 yrs (mm/yr) 0 -20 1 mm/yr 1 mm/yr -40 -60 -80 -100 -120 EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES

Depth (km) -140 -160 -180 -200 -200 -150 -100 -50 0 50 100 150 200 -200 -150 -100 -50 0 50 100 150 200 Distance from fault (km) F: Footwall bending 3 2 0 0.75 1.5 1 Surface Uplift (km) 0 Maximum fault displacement = 4km -1 -2 E: Fit to GPS t=60 yrs 9.1 km -3 Elevation Change (mm/yr) -100 -50 0 50 100 150 200 Distance from fault (km)

Fig. 5. Numerical model of normal fault behavior (21). (A) The model consists of three layers: a 15-km-thick elastic but breakable upper crust, a 15-km-thick ductile lower-crustal layer, and a ductile upper mantle thick enough (170 km) that its boundaries do not affect the solution. A 45° dipping normal faultis embedded in the upper crust in the middle of the model that continues as a into the lower crust. The model can collapse under gravity when its edges are allowed to expand. (B) The model matches observed coseismic subsidence of the fault hanging wall, and stability of the fault foot wall. (C)Ina subsequent test, a 1-m slip event on the fault is imposed, which is equivalent to an M ∼ 7 earthquake. The fault is immediately locked, and the postseismic isostatic effects are calculated for a period 10 y after the earthquake. Strong uplift forces beneath the fault footwall are predicted. Localized variability in vector directions in the upper crust is caused by element . (D) By 60 y after the earthquake (∼present-day observation period after the 1954 Nevada earthquake series), the model predicts a broader uplift effect that spans ∼100 km either side of the fault, and that (E) agrees (red dots show modeled uplift rates) with GPS observations (blue dots with measurement uncertainties plotted as error bars). Outlier points are the result of plastic element fracture in the upper crust. (F) A model of long-term extensional faulting is shown, where the fault is allowed to slip continuously, simulating many earthquakes. This results in the uplifted bent range fronts that characterize the Basin and Range province.

behavior of the ruptured faults over time. If the fault locks up single fault is likely an oversimplification given the many strain immediately after the earthquake and stays locked until the next modes in the Earth. one, then both the hanging wall and footwall will be uplifted as a Deformation from isostatic stress may occur as post- result of isostasy. An energy-balance implication of this scenario is earthquake aseismic creep, or as aftershocks. Postearthquake that each subsequent earthquake would either have to be slightly creep has been observed in extensional settings (31), but more larger than the prior, or come sooner. This is because tectonic observations are needed to know if this is a general occurrence. stressing is essentially monotonic, while each isostatic response A global study finds that the earthquake magnitude–frequency would compound stress accumulation. However, this scenario of distribution on normal faults differs from other modes in a statis- converting gravitational potential energy only into earthquakes on a tically significant way, with an increased rate of smaller earthquakes

Thompson and Parsons PNAS Early Edition | 5of6 Downloaded by guest on September 25, 2021 relative to larger (32). One interpretation of this feature is that fault system during a 15-d period in 1703 (33). A similar normal fault earthquake rupture surfaces are smaller than rerupturing occurred within the same fault system during the other earthquake modes (15). Another interpretation is that 2016 Central Italy earthquake sequence (34). In the authors’ there are more aftershocks (which tend to be smaller events) opinion, these questions can be resolved by a rapid geodetic associated with normal fault mainshocks. Increased aftershock deployment following the next large normal fault earthquake, rates could be a response to isostatic forces, and may help to hence the challenge. shape Basin and Range topography. Evidence from paleo- seismology in the central Italy extensional regime suggests that ACKNOWLEDGMENTS. Bill Hammond and one anonymous reviewer pro- large and moderate earthquakes reruptured the same normal vided constructive and helpful comments.

1. Colgan JP, Dumitru TA, Miller EL (2004) Diachroneity of Basin and Range extension 21. Thompson GA, Parsons T (2016) Vertical deformation associated with normal fault and Yellowstone hotspot volcanism in northwestern Nevada. Geology 32:121–124. systems evolved over coseismic, postseismic, and multiseismic periods. J Geophys Res 2. Martel SJ, Stock GM, Ito G (2014) Mechanics of relative and absolute displacements 121:2153–2173. across normal faults, and implications for uplift and subsidence along the eastern 22. Hammond WC, Blewitt G, Kreemer C (2014) Steady contemporary deformation of the of the Sierra Nevada, California. Geosphere 10:243–263. central Basin and Range Province, western United States. J Geophys Res 119: 3. Colgan JP, Shuster DL, Reiners PW (2008) Two-phase Neogene extension in the 5235–5253. northwestern Basin and Range recorded in a single thermochronology sample. 23. Pollitz FF, Peltzer G, Bürgmann R (2000) Mobility of continental mantle: Evidence Geology 36:631–634. from postseismic geodetic observations following the 1992 Landers earthquake. J 4. Compton RR (1966) Analysis of Pliocene-Pleistocene deformation and stresses in Geophys Res 105:8035–8054. northern Santa Lucia Range, California. Bull Geol Soc Am 77:1361–1380. 24. Wernicke BP, Friedrich AM, Niemi NA, Bennett RA, Davis JL (2000) Dynamics of plate 5. Hammond WC, Blewitt G, Li Z, Plag H-P, Kreemer C (2012) Contemporary uplift of the boundary fault systems from Basin and Range Geodetic Network (BARGEN) and Sierra Nevada, western United States, from GPS and InSAR measurements. Geology geologic data. GSA Today 10:1–7. – 40:667 670. 25. Hetland EA, Hager BH (2003) Postseismic relaxation across the Central Nevada Seismic 6. Gourmelen N, Amelung F (2005) Postseismic mantle relaxation in the Central Nevada Belt. J Geophys Res 108:2394. – Seismic Belt. Science 310:1473 1476. 26. Hammond WC, Thatcher W (2004) Contemporary tectonic deformation of the Basin 7. Doser DI (1986) Earthquake processes in the Rainbow Mountain-Fairview Peak-Dixie and Range province, western United States: 10 years of observation with the Global Valley, Nevada region 1954-1959. J Geophys Res 91:12572–12586. Positioning System. J Geophys Res 109:B08403. 8. Doser DI (1988) Source parameters of earthquakes in the Nevada Seismic Zone, 1915- 27. Freed AM, Bürgmann R, Herring T (2007) Far-reaching transient motions after Mojave 1943. J Geophys Res 93:15001–15015. earthquakes require broad mantle flow beneath a strong crust. Geophys Res Lett 34: 9. Whitten CA (1957) Geodetic measurements in the Dixie Valley area. Bull Seismol Soc L19302. Am 47:321–325. 28. Bürgmann R, Dresen G (2008) Rheology of the lower crust and upper mantle: Evi- 10. Meister LJ, Burford RO, Thompson GA, Kovach RL (1968) Surface strain changes and dence from , geodesy, and field observations. Annu Rev Earth Planet strain energy release in the Dixie Valley-Fairview Peak area, Nevada. J Geophys Res Sci 36:531–567. 73:5981–5994. 29. Hammond WC, Kreemer C, Blewitt G (2009) Geodetic constraints on contemporary 11. Koseluk RA, Bischke RE (1981) An elastic rebound model for normal fault earth- deformation in the northern Walker Lane: 3. Central Nevada seismic belt postseismic quakes. J Geophys Res 86:1081–1090. relaxation. Spec Pap Geol Soc Am 447:33–54. 12. Stein RS, Barrientos SE (1985) Planar high-angle faulting in the Basin and Range: 30. Chang W-L, Smith RB, Puskas CM (2013) Effects of lithospheric viscoelastic relaxation Geodetic analysis of the 1983 Borah Peak, Idaho, earthquake. J Geophys Res 90: on the contemporary deformation following the 1959 Mw 7.3 Hebgen Lake, Mon- 11355–11366. 13. Hodgkinson KM, Stein RS, Marshall G (1996) Geometry of the 1954 Fairview Peak- tana, earthquake and other areas of the intermountain seismic belt. Geochem – Dixie Valley earthquake sequence from a of leveling and triangulation Geophys Geosyst 14:1 17. data. J Geophys Res 101:25437–25457. 31. Riva REM, et al. (2007) Viscoelastic relaxation and long-lasting after-slip following the – 14. Vening Meinesz FA (1950) Les africains, resultat de compression ou de 1997 Umbria-Marche (Central Italy) earthquakes. Geophys J Int 169:534 546. dans la croute terrestre? Inst R Colon Belge, Bull Seances 21:539–552. French. 32. Schorlemmer D, Wiemer S, Wyss M (2005) Variations in earthquake-size distribution – 15. Doglioni C, Carminati E, Petricca P, Riguzzi F (2015) Normal fault earthquakes or across different stress regimes. Nature 437:539 542. graviquakes. Sci Rep 5:12110. 33. Galli P, Galadini F, Calzoni F (2005) Surface faulting in Norcia (central Italy): A pale- – 16. Stein RS (1999) The role of stress transfer in earthquake occurrence. Nature 402: oseismological perspective. Tectonophysics 403:117 130. 605–609. 34. Chiaraluce L, et al. (2017) The 2016 Central Italy seismic sequence: A first look 17. Freed AM (2005) Earthquake triggering by static, dynamic, and postseismic stress at the mainshocks, aftershocks, and source models. Seismol Res Lett 88: transfer. Annu Rev Earth Planet Sci 33:335–367. 757–771. 18. Anderson EM (1951) The Dynamics of Faulting and Dyke Formation (Oliver and Boyd, 35. Duffield WA, Weldin RD (1976) Mineral Resources of the South Warner Wilderness, Edinburgh). Modoc County, California, US Geological Survey Bulletin (United States Government 19. Vogfjord S, Langston CA (1987) The Meckering earthquake of 14 October 1968: A Printing Office, Washington, DC), Vol B1385, p 31. possible downward propagating rupture. Bull Seismol Soc Am 77:1558–1578. 36. Van Buer N (2012) Preliminary Geologic Map of the Sahwave and Nightingale Ranges, 20. Kusznir NJ, Park RG (1986) Continental lithosphere strength: The critical role of lower Churchill, Pershing, and Washoe Counties, Nevada, Open File Report 12-2 (Nevada crustal deformation. Geol Soc Lond Spec Publ 24:79–93. Bureau of Mines and Geology, Reno, NV), p 12.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1711203114 Thompson and Parsons Downloaded by guest on September 25, 2021