2002 Stress in Earth Lithosphere

You might also like

Download as pdf
Download as pdf
You are on page 1of 12
Stress in the Earth’s Lithosphere Mark D. Zoback ‘Stanford University |. Basic Definitions IL. Indicators of Contemporary Stross Ill, Distribution of Crustal Stresses IV First-Order Global Stress Patterns V. Sources of Crustal Stress VI. The Critically Stressed Crust VIL. Summary GLOSSARY Critically stressed erust Stress magnitudes inthe brittle crust are equal to the crust’s frictional strength, In sita stress Forces in the lithosphere, Sourees of stress ‘The mechanisms responsible for stress in the lithosphero. Stress map Map showing the orientation and relative ‘magoitude of horizontal principal stress orientations. ‘THE STATE OF STRESS in the lithosphere isthe re- sultof the forees acting upon and within it, Knowledge of ‘he magnitude and distribution ofthese forces can he com- bined with mechanical, taermal, and rheological consta- ints to examine a broad range of lithospheric deformat- {onal processes. For example, such knowledge contributes to a botor understanding of the processes that both drive and inhibit lithospheric plate motions, as wells the forces responsible for the occurrence of crustal earthquakes — both along plate boundaries and in intrapate regions rp tl Saar Tay Pi Wo 16 Sippy hear rt ii gsct pany res Mary Lou Zoback US. Geological Survey ‘While ovr topic is the state of stress in the earth's litho sphere, the comments below come primarily from the per- spective ofthe state of stress in the brittle upper crust. As defined by the depth of shallow earthquakes, the brittle crust extends to a depth of ~15~20 lam at most continen- tal locations around the world. We adopt this perspective because nearly all the data available on lithospheric stress comes from the upper crust of continents. Furthermore, in the sections that follow, we argue that, to first order, the state of stress in the britle crust results from relatively large-scale lithospheric processes, so that knowledge of ‘crustal stress can be used to constrain the forces involved in these processes. |. BASIC DEFINITIONS ‘Stress is a tensor which describes the density of forces act- ing onall surfaces passing through a point. In termsof con- tinuum mechanics, the sresses acting on a homogeneous, isotropic body at depth are describable as a second rank tensor, with nine components (Fig. 1, lef). va 144 lsu Siz Su S=|su Sa So) a S52 Ss ‘The subscripts of the individual stress components refer tothe direction that a given force is acting and the face of the unit cube upon which the stress component acts, Thus, in simplest terms, any given stress component represents force acting ina specific direction on aunit area of given orientation. As illustrated inthe left side of Fig. 1, a stress tensor can be defined in terms of any arbitrary reference frame. Because of equilibrium conditions, @) Sx = Sa so that the order of the subscripts is unimportant. In gen= ‘eral, to fully describe the stao of stress at depth, one must Stress Description of Stresses in 3-D ‘Tensor Transformation (rotation of axes) Sresin tbe Ean’ Libspbere estimate sx tess magnitudes or thee sess magnitades land the three angles that define the orientation ofthe stress coordinate system with respect to a reference coordinate ‘system (such as geographic coordinates, for example). ‘Weuailize the convention that compressive stress is pos- itive because insite stesss a depts greater than a few ‘tens of meters in the earth are always compressive. ‘Ten- sile stresses do not exist at depth in the earth fortwo fun- damental reasons. First, because the tensile strength of rockis generally quite low, significant tensile stress can- not be supported in the earth, Second, because there is always afid phase saturating the pore space in rock at depth (except at depths shellower than the water table), the pore pressure resulting from this uid phase would cause the rock to hydraulically fracture should the least ‘compressive stress reach a value even as low as the pore presse. Once a stress tensors known, itis possible to evaluate stresses in any coordinate system via tensor transforma- tion, To accomplish his transformation, we need to spoc~ ity the direction cosines (qj asillustrated in Fig, 1 center) Principal Stress Tensor (new) 8 Ea Sa Se an an a s=]sn 2 8 cane oo [2 Su Su $9 a an a, Directions Cosines FIGURE 1. Definon of siess tensor in an aibitary Cartesian coordinate sysiem (lef) and the rotation of stress coordinate systems through tensor transformation (middle) and principal stresses (fight) as defined in a coordinate systom in which shear sresces vanish, Sues the Banh Lihsphere that describe the rotation of the coordinate axes between the old and new coordinate systems. Mathematically, the equation which accomplishes this is F=ASA, ° where jar a2 an| Aslan an an| lan a ay| The ability to transform coordinate systems is of inter- est here because we can choose to generally describe the state of stress in terms of the principal coordinate sys- tem, The principal coordinate system isthe one in which shear stresses vanish and only three principal stresses, 5; = 52 = Sy fully describe the stress field (as illustrated in right side of Fig. 1). Thus, we have diagonalized the stress tensor such thatthe principal stresses correspond to the eigenvalues of the stress tensor and the principal stress directions correspond to its eigenvectors. is) 0 0 =|0 % 0 eo 0 0 si ‘The reason this concept is 60 important is that as the carts surface is in contact with a uid (ither air or wae ter) and cannot support shear traction, it is principal stress plane. Thus, one principal stresé is generally ex- pected to be normal to the earth's surface withthe other {wo principal stesses acting in an approximately hori- zontal plane. While itis clea that this must be true very close tothe earth's surface, compilation of earthquake fo- cal mechanism data and other stress indicators (described below) suggest that itis sso generally true to the depth of the bitle-ducile transition inthe upper cnust (Zoback and Zoback, 1989; Zoback, 1992). Assuming this is the case, ‘we must define only four parameters to describe the state of stress at depth; one stress orientation (usually taken to be the azimuth of the maximum horizontal compression, Shins) and thee principal stress magnitudes: S,, the ver tical stress, comesponding the weight ofthe overburden: Stnaas the maxima principal horizontal stress; and Spins the minimum principal horizontal stess, This obviously Jaelps make stress determination in the crust a tractable problem. Tnapplying these concepts to the earth's crust itis help- ful to consider the magnitudes of the greatest, intermedi- ae, and maximum principal stres at depth (Si, Sand 53) in terms of S,, Stina; and Supia in the manner originally proposed by E. M. Anderson. (1951). Ths is illustrated in Fig. 2, There aro a number of simple but fundamental points ebout these seemingly straightforward relations. 145 - Normal oi Sy>Stinae*Shnin Se Strike-Slip Soin | 7 Friw?S>8 hn 5, Reverse = rea Stain | Stim Stnin>Sy FIGURE 2, M. Anderson's classification scheme for relative sires magnitides in normal, strike-slip, and reverse faulting re- ions. Corresponding focal pane mechanisms are shown tothe Fight Fit, the two horizontal principal stresses in the earth, ‘Stimax aNd Stysins Can be described relative to the vertical principal stress, S, whose magnitude coresponds tothe overburden, Mathematically, 5, is equivalent to integra- tion of density from the surface tothe depth of intrest, z. nother words, Ss f play dz © Dge, 6 where (2) is the density as a function of depth, gis the gravitational acceleration, and is the mean overburden density. tis, of course, necessary to.add atmospheric pres- sure and the pressure resulting from the weight of watce atthe earth’s surface, as appropriate ‘Second, the horizontal principal stresses are almost never equal and may be less than or greater than the verti- cal stress. In fact, the relative magnitudes of the principal stresses can be simply related to the faulting style curreatly ‘active in a region. As illustrated in Fig. 2, (and follow- ing Anderson, 1951), characterizing a region by normal, strike-slip, or reverse faulting is equivalent to defining the horizontal principal stress magnitudes with respect to the vertical stress, When the vertical stress dominates in extensional deformational regions (S;=S,), gravity drives normal faulting. Conversely, when both horizontal stresses exceed the vertical stress (S3 = S,), compressional ‘deformation (shortening) is accomodated through reverse 146 faulting. Swike-slip faulting represents an intermediate ‘stress state (S;=5,), where the maximum horizontal stress is greater than the vertical stress and the min ‘imum horizontal stress is less than the vertical stress Sitges = 5,2 Sis). ‘The concept of effective stress is used to incorporate the influence of pore pressure at depth whece a component of| effective stesso; is elated 1 the total stess 5; via oy = (Si — dy Foy © here isthe Kronecker delta and P; isthe pore pressure. Laboratory studies of the frictional strength of faulted rock carried out over the past several decades indicate that, the Coulomb criterion describes the fictional strength of faults. Tat is, fault sippage will occur when T= St Uo oO where t is the shear stress acting on the fault, S. is the fault ‘cohssion, 11 is the coeficient of friction on the fault, and 2p is the effective normal stress acting on the fault plane. ‘The maximum shear stress is given by 1/2(61 ~ 54). Using the concept of effective stress at depth, we can extend Anderson's faulting theory to predict stress ‘magnitudes at depth through utilization of simplified two-dimensional Mohr-Coulom feilure theory. Two dimensional faulting theory assumes that failure i only «function of the difference between the lst and pretest principal effective stresses 0 and a3 as given by Jacger and Cook (1971) orfos = (Si — Fp)/MS— Py) =? +1)? +P, 8) ‘Thus, a third point about crustal stress that cam be de- rived from Andersonian faulting theory is thatthe mag- nitudes of the throe principal stresses at any depth are Timited by the strength of the earth's crust at depth. In ‘the case of normal faulting, stress magnitudes are con- trolled by 2%, and din, Which correspond to ¢; and 0, respectively Otnas. Which corresponds to is interme. diate in value betwen oy 8d Cig, does not influence faulting. Coulomb failure theory indicates tha fictional sliding occurs when the ratio shear stress to effective nor- smal stress.on preexisting fault planes is equal othe coelfi- ientoffiietion. As the coefficient of friction is relatively wel defined for most rocks and ranges between ~0,6 and 1.0 (Byeriee, 1978), Eq. (8) demonstrates that frictional sliding will occur when oj /o3 ~3.1~5.8. For the case of hydrostatic pore pressure and commonly observed frc- tion coefficients of 0.6 (e.g,, Townend and Zoback, 2000), in extensional areas Synin~0.6 Sy, in roverse faulting ar- 85 Shines ~ 2.3 Sy, and in strike-slip faulting areas (when S.~1/UStinax+ Suit) Shinar ~2.2 Syne As discussed below, these simple relationships have been confirmed by in sit stress measurements to depths of almost 8 km ata number of sits in intraplate areas, Sutin he Eas Libre Il, INDICATORS OF CONTEMPORARY ‘STRESS Information on te state of stress in the lithosphere comes from a variety of sources—earthquake focal plane mech- anisms, young geologic data on fault slip and volcanic alignments, in sit suess measurements, stress-induced wellbore breakoats, and diling-nduced tensile fractures. A stress measurement quality criterion for different types of stress indicators was developed by Zoback and Zoback (1989, 1991). This quality criterion was subsequently uti- lized in the Iniernational Lithosphere Program's World Stress Map Project, a large collaborative effort of data ‘compilation and analyses by over 40 scientists from 30 diffrent countries (Zoback, 1992). A special issue of the Journal of Geophysical Research (v.97, pp. 11,703- 12,014, 1992) summarized the overall results of this project and presented the individual cantibutions of many ofthese investigators in various regions of the wold. To day, the World Stress Map (WSM) database hos more than 9100 entries and is maintained at the Heidelberg, ‘Academy of Sciences and Humanities (Mueller er al 1997; hitp:hwww-wsm.physiuni-karlsrube de) Zoback and Zoback (1991) discussed the rationale for the WSM guilty criterion used in the WSM project in detail. The success of the WSM project demonstrates that ‘with careful attention to data quality, coherent stress pat- tems over large regions of the earth can be mapped with reliability and interpreted with respect to large-scale litho- spheric processes. ‘A. Earthquake Focal Mechanisms While earthquake focal plane mechanisms are the most ubiquitous indicator of stess in the lithosphere, the de- termination of principal stess orientations and relative magnitudes from these mechanisms must be done with appreciable caution. The pattern of seismic radiation from the focus of an earthquake permits construction of earth- quake focal mechanisms (right column of Fig. 2). Perhaps the most simple and straightforward information about Jn situ stress that is obtainable from focal mechanisms and in sine stress is that the type of earthquake (i.e, not- ral, strike-slip, or reverse faulting) defines the relative magnitudes Of Sina Sains Sy. In ation, the oie tation of the fault plane and the auxiliary plane (whieh bind the compressional and extensional quadrants ofthe focal plane mechanism) defines the orientation ofthe P (com- pressional), & (intermediate), and 7 (extensional) axes, ‘hese axes are sometimes incomectly assumed to be the same as the orientation of $y, Sx, and S} For cases in which laboratory-measured coeticiens of fault friction of ~0.6-1.0 are applicable to the crust, there isanontrvial enor of ~15-20"ifoneuses the P, B, and T Srosin be Ears Lideaphere axes as approximations of average principal stress orien- tations, especially ifthe orientation ofthe fault plane upon Which the earthquake occurred is known. I friction is neg. ligible onthe fits in question bat higher in surrounding rocks), there can be a considerable difference between the P, B, and axes and principal stress directions, An earth- ‘quake focal plane mechanism alwayshasthe P and T axes, at 45° to the fault plane and the B axis inthe plane of the fault, Wit a frictionless fautthe seismic radiation pattern is controlled by the orientation of the fault plane and not the in situ stress field (McKenzie, 1969). One result ofthis is that ust knowing the orientation ofthe P axis of earth- ‘quakes along weak, plate-bounding, strike-slip faalts (ike the San Andreas) does not allow one to define principal stress otientations from the focal plane mechanisms of the strike-slip earthquakes occurring on the fault (Zoback etal, 1987). For this reason, itis common practice to omit plate-boundary earthquakes from regional stress compils- tions (see below). Principal stess directions can be determined directly from a group of eérthquake focal mechanisms (or set of ‘ult striae measurements) through use of inversion tech- niques that are based on the slip kinematics and the as- sumption that falt slip will always occur inthe dizection ‘of maximum resolved shear stress on fault plane. Such inversions yield four parameters, the orientation of the thee principal stress and the relative magnitude ofthe in- termediate principal stress with respect to the maximum and minimum principal stress ‘The analysis of seismic waves rediating from an earth- ‘quake also can be used to estimate the magnitude of sts released in an earthquake (stress érop), although not abso lute stress levels. In general, stress drops of crustal earth- quakes are on the order of 1-10 MPa. Euation (8) can be used to show that such stress drops are only a small fraction ofthe shear stresses that actually causes fault slip if pore pressures are approximately hydrostatic at depth and Coulomb faulting theory with laboratory-

You might also like