Earthquake Sources, Moment Tensor and Focal Mechanisms
CSIR NET geophysics: double-couple source, seismic moment, moment tensor decomposition, radiation patterns, source spectra, stress drop, fault-plane solutions.
- 21 min
On this page · 15 sections
- Prerequisites
- The Hook: Two Forces Are Not Enough
- Kinematic Description of a Fault Source
- The Moment Tensor
- Radiation Patterns
- Fault-Plane Solutions from First Motions
- Source Spectrum, Corner Frequency and Stress Drop
- Finite Rupture: Duration and Directivity
- Memory Tricks
- Don’t Confuse
- Common Mistakes
- Practice
- Quick Revision Table
- Key Takeaways
- What’s Next
Prerequisites
- Earthquakes: Causes, Measurement and Seismicity — elastic rebound, magnitude scales, \( M_0 = \mu AD \) and beach balls at Part B level.
- Stress, Mohr Circle, Failure Criteria and Fault Reactivation — principal stresses, Anderson’s fault types, effective stress.
- Seismometry — how the waves used here are recorded.
- Linear algebra: eigenvalues and eigenvectors of a symmetric 3 × 3 matrix.
The Hook: Two Forces Are Not Enough
Kinematic Description of a Fault Source
A tectonic earthquake is a shear dislocation: the two sides of a fault surface \( \Sigma \) slip past each other by a displacement discontinuity \( \mathbf{D} \) (the slip vector), which lies in the fault plane. Seismology describes the source by equivalent body forces that would radiate the same waves in an unfaulted medium.
| Geometric quantity | Symbol | Meaning |
|---|---|---|
| Fault normal | \( \hat{\mathbf n} \) | unit vector perpendicular to the fault, pointing towards the hanging-wall (positive) side |
| Slip direction | \( \hat{\mathbf d} \) | direction in which the positive side moves relative to the negative side |
| Strike \( \phi_s \) | 0–360° | azimuth of the fault trace, measured clockwise from north, with the fault dipping to the right when looking along strike |
| Dip \( \delta \) | 0–90° | angle of the plane below the horizontal |
| Rake \( \lambda \) | −180° to 180° | angle in the fault plane from the strike direction to the slip vector of the hanging wall |
Rake conventions (Aki and Richards):
| Rake | Faulting |
|---|---|
| 0° | left-lateral strike-slip |
| ±180° | right-lateral strike-slip |
| +90° | pure thrust (reverse) |
| −90° | pure normal |
| between 0 and 180 | reverse component; between 0 and −180, normal component |
Seismic moment
The scalar seismic moment is the strength of each couple of the double couple:
$$\boxed{M_0 = \mu A \bar D}$$with \( \mu \) the rigidity of the source region, \( A \) the rupture area and \( \bar D \) the average slip. It is measured from the long-period amplitude of the waves (the flat part of the source spectrum, below) and gives the moment magnitude:
$$\boxed{M_w = \frac{2}{3}\left(\log_{10}M_0 - 9.1\right)}\qquad (M_0\ \text{in N m})$$Inverting: \( M_0 = 10^{1.5M_w + 9.1} \) N m. One unit of \( M_w \) is a factor of \( 10^{1.5} \approx 31.6 \) in moment; 0.2 units is a factor of 2.
The Moment Tensor
A general point source of seismic waves (with no net force or torque) is described by a symmetric second-order moment tensor \( M_{ij} \). The far-field displacement is a linear combination of its six independent components, so \( M_{ij} \) can be found by linear inversion of waveforms; that is how modern catalogues (such as the Global CMT project’s centroid moment tensors) are made.
For a shear dislocation,
$$\boxed{M_{ij} = M_0\,(\hat n_i\hat d_j + \hat n_j\hat d_i)}$$Because \( \hat{\mathbf n} \) and \( \hat{\mathbf d} \) enter symmetrically, swapping them (fault plane ↔ auxiliary plane) gives the same tensor. That is the fault-plane ambiguity in one line.
Eigenvalues and principal axes
For a pure double couple the eigenvalues are
$$\lambda = (+M_0,\ 0,\ -M_0)$$| Eigenvector | Eigenvalue | Axis | Relation to the fault |
|---|---|---|---|
| \( (\hat{\mathbf n} + \hat{\mathbf d})/\sqrt 2 \) | \( +M_0 \) | T axis (tension) | 45° from both nodal planes, in the compressional first-motion quadrant |
| \( (\hat{\mathbf n} - \hat{\mathbf d})/\sqrt 2 \) | \( -M_0 \) | P axis (pressure) | 45° from both nodal planes, in the dilatational quadrant |
| \( \hat{\mathbf n}\times\hat{\mathbf d} \) | 0 | B (null) axis | intersection line of the two nodal planes |
The P and T axes are often taken as estimates of the principal stress directions \( \sigma_1 \) and \( \sigma_3 \), but this is an approximation: a fault that slips may be a pre-existing weak plane not at 45° to \( \sigma_1 \). Inversions of many mechanisms for a regional stress tensor are better than a single mechanism.
Decomposition: isotropic, double couple and CLVD
Any moment tensor splits into an isotropic part and a deviatoric part:
$$M_{ij} = \tfrac13 M_{kk}\,\delta_{ij} + M'_{ij}$$- The isotropic part \( M_{kk}/3 \) represents volume change: positive for explosions, negative for implosions (for example, some deep-mine collapses).
- The deviatoric part, with eigenvalues \( m_1, m_2, m_3 \) summing to zero, is further split into a double couple (DC) and a compensated linear vector dipole (CLVD): a dipole of strength \( 2 \) along one axis balanced by dipoles of strength \( -1 \) along the other two, with no volume change.
A common measure of non-DC content uses the deviatoric eigenvalues ordered by absolute value, \( |m_{min}| \le |m_{int}| \le |m_{max}| \):
$$\boxed{\varepsilon = -\frac{m_{min}}{|m_{max}|}}$$| \( \varepsilon \) | Source |
|---|---|
| 0 | pure double couple (eigenvalues \( M_0, 0, -M_0 \)) |
| ±0.5 | pure CLVD (eigenvalues in ratio 2 : −1 : −1) |
| between | mixture; percentage DC \( \approx (1 - 2\lvert\varepsilon\rvert)\times 100 \) |
Non-double-couple sources to recognise: explosions and cavity collapses (isotropic); volcanic events, ring-fault slip and magma-filled crack opening (CLVD and isotropic parts); tensile cracks in geothermal and injection settings; and complex ruptures on several faults with different orientations, which sum to an apparent non-DC tensor. Small non-DC percentages are often just noise and modelling error.
Radiation Patterns
For a double couple in a homogeneous medium, the far-field displacement at distance \( r \) is (Aki and Richards form):
$$\mathbf u^P = \frac{1}{4\pi\rho\alpha^3}\,\frac{1}{r}\,\dot M_0\!\left(t - \frac r\alpha\right)\,\sin 2\theta\cos\phi\;\hat{\mathbf r}$$$$\mathbf u^S = \frac{1}{4\pi\rho\beta^3}\,\frac{1}{r}\,\dot M_0\!\left(t - \frac r\beta\right)\left(\cos 2\theta\cos\phi\;\hat{\boldsymbol\theta} - \cos\theta\sin\phi\;\hat{\boldsymbol\phi}\right)$$with \( \theta \) measured from the fault normal, \( \phi \) from the slip direction in the fault plane, \( \alpha \) and \( \beta \) the P and S velocities.
What to read from these:
- Far-field displacement is proportional to the moment rate \( \dot M_0(t) \): a step in moment gives a pulse of displacement. The pulse duration is the source duration.
- P radiation has four lobes with alternating sign; it is zero on both nodal planes (the fault plane and the auxiliary plane). Maximum P amplitude lies along the P and T axes, 45° from the planes.
- S radiation is maximum on the nodal planes where P is zero, and zero along the P, T and B axes.
- Far-field S is larger than P by about \( (\alpha/\beta)^3 \). For a Poisson solid, \( \alpha/\beta = \sqrt3 \) and the factor is about 5.2.
- Amplitudes fall as \( 1/r \) (geometric spreading of body waves in a whole space), before attenuation.
Near-field terms decay faster (as \( 1/r^2 \) and \( 1/r^4 \)) and record the permanent static displacement close to the fault; they are what GPS and InSAR measure.
Fault-Plane Solutions from First Motions
Procedure
- For each station, pick the polarity of the first P motion: up (compression, C) or down (dilatation, D) on the vertical component.
- Compute the ray’s azimuth from source to station and its take-off angle \( i_h \) at the source (from the velocity model, using \( p = \sin i_h/v_h \)).
- Plot each polarity on a lower-hemisphere stereographic (Wulff) or equal-area (Schmidt) projection of the focal sphere.
- Fit two orthogonal great circles (the nodal planes) that separate compressions from dilatations.
- Read the strike and dip of each plane. The pole of each plane is the slip vector of the other. P lies in the middle of the dilatational quadrants, T in the middle of the compressional ones, both 45° from the planes; B is where the planes intersect.
Take-off angles are measured from the downward vertical. A ray that leaves the source upward (take-off angle greater than 90°, common for nearby stations above a deep focus) is plotted on the lower hemisphere at the antipodal point: azimuth + 180° and take-off angle 180° − \( i_h \). Its polarity is unchanged, because the radiation pattern is symmetric through the source.
Other data that constrain mechanisms: S-wave polarisation directions, amplitude ratios (S/P, SH/SV), and full waveform inversion for the moment tensor (works with fewer stations and gives \( M_0 \) at the same time).
Reading beach balls
| Mechanism | Lower-hemisphere pattern | P axis | T axis | Rake |
|---|---|---|---|---|
| Thrust | compressional (shaded) centre | horizontal | vertical | about +90° |
| Normal | dilatational (white) centre | vertical | horizontal | about −90° |
| Strike-slip | four quadrants, both planes vertical (a cross) | horizontal | horizontal | about 0° or 180° |
| Oblique | asymmetric | inclined | inclined | intermediate |
Source Spectrum, Corner Frequency and Stress Drop
The \( \omega^2 \) model
Real ruptures have finite size and duration. The far-field displacement spectrum of a body wave is, to a good approximation (the omega-squared model of Aki and of Brune):
$$|\Omega(f)| = \frac{\Omega_0}{1 + (f/f_c)^2}$$- Below the corner frequency \( f_c \): flat, with level \( \Omega_0 \propto M_0 \). This is where \( M_0 \) is measured.
- Above \( f_c \): falls as \( f^{-2} \).
- \( f_c \) is inversely proportional to the source dimension (and duration). Bigger earthquakes have lower corner frequencies.
This is also why amplitude magnitudes saturate: \( m_b \) is read at about 1 s and \( M_s \) at about 20 s. Once \( f_c \) drops below the measurement frequency, the amplitude there stops growing in proportion to \( M_0 \).
Brune’s model
For a circular fault of radius \( r \), Brune related the S-wave corner frequency to the radius:
$$\boxed{r = \frac{2.34\,\beta}{2\pi f_c} \approx \frac{0.37\,\beta}{f_c}}$$and the static stress drop for a circular crack (Eshelby) is
$$\boxed{\Delta\sigma = \frac{7}{16}\,\frac{M_0}{r^3}}$$Typical earthquake stress drops lie between about 0.1 and 100 MPa, with many in the 1–10 MPa range, and are roughly independent of size. Constant stress drop implies self-similar scaling: \( M_0 \propto r^3 \propto f_c^{-3} \), so \( \log M_0 \) increases by 3 when \( \log f_c \) decreases by 1.
General form for any shape: \( \Delta\sigma = C\mu\bar D/\tilde L \), with \( \tilde L \) a characteristic dimension (radius or width) and \( C \) a shape factor; for a circular crack, \( C = 7\pi/16 \).
Energy and apparent stress
- Radiated seismic energy \( E_s \) is measured by integrating the squared velocity spectrum, so it depends on the high frequencies above \( f_c \).
- Apparent stress \( \sigma_a = \mu E_s/M_0 \) links energy and moment. Kanamori’s \( M_w \) was built on the assumption \( E_s/M_0 \approx 5\times10^{-5} \), which follows from \( E_s \approx (\Delta\sigma/2\mu)M_0 \) with \( \Delta\sigma \) about 3 MPa and \( \mu \) about \( 3\times10^{10} \) Pa.
- The energy magnitude \( M_e \) is computed from \( E_s \); it can differ from \( M_w \) for events with unusual stress drop or slow rupture.
Finite Rupture: Duration and Directivity
A rupture of length \( L \) propagating at rupture velocity \( v_r \) (typically about 0.7–0.9 of \( \beta \)) lasts about \( L/v_r \). For unilateral rupture, a station at angle \( \psi \) from the rupture direction sees the pulse duration
$$\boxed{\tau(\psi) = \frac{L}{v_r} - \frac{L}{c}\cos\psi}$$where \( c \) is the wave speed (P or S). In the rupture direction (\( \psi = 0 \)) the pulse is short and large (energy arrives compressed, like a Doppler effect); behind the rupture (\( \psi = 180° \)) it is long and small. The area under the displacement pulse (proportional to \( M_0 \)) is the same at all azimuths. This directivity explains asymmetric damage around many large earthquakes.
Other finite-source terms: rise time (time for slip at one point to complete), source time function (moment rate against time), slip distribution (from inversion of seismic, geodetic and tsunami data), asperities (high-slip patches) and barriers. Very large ruptures, such as megathrust events, last minutes; this is why \( M_w \) must be computed from very long periods or the W phase for great earthquakes.
Memory Tricks
Don’t Confuse
| This | vs | That |
|---|---|---|
| Single couple: net torque, wrong S pattern | Double couple: no net torque, equivalent to shear slip | |
| Fault plane | Auxiliary plane: identical radiation; needs geology, aftershocks or directivity to resolve | |
| P axis (compression, eigenvalue \( -M_0 \)) | T axis (tension, eigenvalue \( +M_0 \)) | |
| P and T axes from one mechanism | Principal stresses \( \sigma_1, \sigma_3 \): only approximately aligned | |
| Isotropic component: volume change | CLVD: no volume change, dipole 2 : −1 : −1 | |
| Static stress drop \( \Delta\sigma \) (from \( M_0 \) and size) | Apparent stress \( \mu E_s/M_0 \) (from radiated energy) | |
| Corner frequency: source size | Instrument natural frequency: sensor property | |
| Hypocentre: where rupture starts (first arrivals) | Centroid: moment-weighted average location (CMT) | |
| Rupture velocity \( v_r \) (below \( \beta \) usually) | Slip velocity (particle velocity of fault faces, about m/s) |
Common Mistakes
Practice
All questions below are original practice questions written for this page. They are not from any CSIR NET paper.
Level 1: Foundation
- Give the eigenvalues of a pure double-couple moment tensor of scalar moment \( M_0 \).
- What rake describes pure normal faulting?
- In which quadrant of a beach ball does the T axis lie?
Answers
- \( +M_0,\ 0,\ -M_0 \).
- −90°.
- The compressional (shaded) quadrant.
Level 2: Part B Style
Practice Q1. A rupture of 40 km × 15 km with average slip 1.0 m in crust of rigidity \( 3\times10^{10} \) Pa has a moment magnitude of about:
(a) 5.8 (b) 6.4 (c) 6.8 (d) 7.4
Answer: (c) 6.8
\( A = 4\times10^4\times1.5\times10^4 = 6\times10^8 \) m² (600 km²). \( M_0 = 3\times10^{10}\times6\times10^8\times1.0 = 1.8\times10^{19} \) N m.
\( \log_{10}M_0 = 19.255 \); \( M_w = \frac23(19.255 - 9.1) = \frac23(10.155) = 6.77 \).
Using the area in km² (600) instead of m² gives an absurdly small moment; always convert to SI first.
Practice Q2. Which moment tensor (eigenvalues, same units) is closest to a pure explosion?
(a) \( (1, 0, -1) \) (b) \( (2, -1, -1) \) (c) \( (3, 3, 3) \) (d) \( (5, -1, -4) \)
Answer: (c)
Equal positive eigenvalues mean a purely isotropic expansion. (a) is a pure DC, (b) a pure CLVD, (d) a deviatoric DC + CLVD mixture.
Practice Q3. For a double-couple source, which statement about far-field radiation is correct?
(a) P amplitude is maximum on the fault plane. (b) S amplitude is zero on the nodal planes. (c) P amplitude is zero on both nodal planes and maximum along the P and T axes. (d) P and S have the same maximum amplitude for the same moment rate.
Answer: (c)
\( \sin2\theta\cos\phi \) vanishes on both planes and peaks at 45°. S peaks on the planes; far-field S is about \( (\alpha/\beta)^3 \approx 5 \) times larger than P.
Practice Q4. The S-wave corner frequency of an earthquake is 0.5 Hz and \( \beta = 3.6 \) km/s. The Brune source radius is about:
(a) 0.67 km (b) 1.3 km (c) 2.7 km (d) 7.2 km
Answer: (c) 2.7 km
\( r = 2.34\beta/(2\pi f_c) = 2.34\times3.6/(2\pi\times0.5) = 8.424/3.142 = 2.68 \) km.
Level 3: Part C Style
Practice Q5. An \( M_w \) 6.0 earthquake has a Brune radius of 4.0 km. A second earthquake in the same region has the same stress drop and \( M_w \) 7.0. Which statements are correct?
- The stress drop of the first event is about 8.6 MPa.
- The moment of the second event is about 31.6 times that of the first.
- The radius of the second event is about 12.6 km.
- The corner frequency of the second event is about one-third that of the first.
(a) 1 and 2 only (b) 2 and 3 only (c) 1, 2, 3 and 4 (d) 1, 2 and 4 only
Answer: (c) 1, 2, 3 and 4
- \( M_0 = 10^{18.1} = 1.259\times10^{18} \) N m. \( \Delta\sigma = 0.4375\times1.259\times10^{18}/(4\times10^3)^3 = 0.4375\times1.259\times10^{18}/6.4\times10^{10} = 8.6\times10^6 \) Pa. True.
- One unit of \( M_w \): \( 10^{1.5} = 31.6 \). True.
- Same \( \Delta\sigma \): \( r \propto M_0^{1/3} \), so \( r_2 = 4.0\times31.6^{1/3} = 4.0\times3.16 = 12.6 \) km. True.
- \( f_c \propto 1/r \): ratio \( 1/3.16 \approx 0.32 \), about one-third. True.
Practice Q6. A moment tensor in (x = north, y = east, z = down) coordinates is
$$\mathbf M = \begin{pmatrix}2 & 0 & 0\\ 0 & 2 & 0\\ 0 & 0 & -4\end{pmatrix}\times10^{16}\ \text{N m}$$Which statements are correct?
- Its isotropic part is zero.
- It is a pure CLVD with \( \varepsilon = 0.5 \) in magnitude.
- Its major dipole is vertical with negative eigenvalue, i.e. vertical compression balanced by horizontal extension.
- It is a pure double couple.
(a) 1, 2 and 3 only (b) 1 and 4 only (c) 2 and 3 only (d) 4 only
Answer: (a) 1, 2 and 3 only
- Trace \( = 2 + 2 - 4 = 0 \). True.
- Eigenvalues \( 2, 2, -4 \): ratio 1 : 1 : −2, a CLVD. Ordered by absolute value, \( m_{min} = 2 \), \( m_{max} = -4 \): \( \varepsilon = -2/4 = -0.5 \), magnitude 0.5. True.
- The largest-magnitude eigenvalue \( -4 \) is along z (vertical): compressional dipole vertical, balanced by tension horizontally. True. Such mechanisms are reported, for example, for some volcanic and ring-fault sources.
- A DC needs a zero eigenvalue. False.
Practice Q7. A unilateral rupture of length 60 km propagates at 3.0 km/s. S waves travel at 3.5 km/s. Which statements are correct?
- The rupture lasts 20 s.
- A station in the rupture direction records an S pulse of about 2.9 s.
- A station directly behind the rupture records an S pulse of about 37 s.
- The area under the far-field displacement pulse is 13 times larger in the forward direction.
(a) 1, 2 and 3 only (b) 1 and 4 only (c) 2, 3 and 4 only (d) 1, 2, 3 and 4
Answer: (a) 1, 2 and 3 only
- \( L/v_r = 60/3 = 20 \) s. True.
- Forward: \( 20 - 60/3.5 = 20 - 17.14 = 2.86 \) s. True.
- Backward: \( 20 + 17.14 = 37.14 \) s. True.
- The pulse area is proportional to \( M_0 \), the same in every direction; only its height and width change (the forward pulse is about 13 times taller). False.
Quick Revision Table
| Idea | Fact or formula |
|---|---|
| Equivalent force | double couple (no net torque) |
| Seismic moment | \( M_0 = \mu A\bar D \), N m |
| Moment magnitude | \( M_w = \frac23(\log_{10}M_0 - 9.1) \); ×31.6 per unit; ×2 per 0.2 |
| Shear-source tensor | \( M_{ij} = M_0(n_id_j + n_jd_i) \); symmetric in \( n, d \) |
| DC eigenvalues | \( +M_0 \) (T), 0 (B), \( -M_0 \) (P) |
| Decomposition | isotropic \( M_{kk}/3 \) + DC + CLVD |
| \( \varepsilon \) | \( -m_{min}/\lvert m_{max}\rvert \); 0 DC, ±0.5 CLVD; DC% ≈ \( 1 - 2\lvert\varepsilon\rvert \) |
| Rake | 0 left-lateral, 180 right-lateral, +90 thrust, −90 normal |
| Radiation | P \( \propto \sin2\theta\cos\phi \), zero on planes; S max on planes; S/P ≈ \( (\alpha/\beta)^3 \) |
| Far field | \( \propto \dot M_0(t - r/c)/r \) |
| Beach balls | thrust shaded centre; normal white centre; strike-slip cross |
| \( \omega^2 \) model | flat below \( f_c \) (\( \propto M_0 \)), \( f^{-2} \) above |
| Brune | \( r = 2.34\beta/(2\pi f_c) \); \( \Delta\sigma = \frac{7}{16}M_0/r^3 \) |
| Stress drop | about 0.1–100 MPa, often 1–10 MPa; self-similar \( M_0 \propto f_c^{-3} \) |
| Apparent stress | \( \sigma_a = \mu E_s/M_0 \); \( E_s/M_0 \approx 5\times10^{-5} \) |
| Directivity | \( \tau = L/v_r - (L/c)\cos\psi \) |
Key Takeaways
What’s Next
- Seismotectonics: Himalayan, SCR and Reservoir-Induced Earthquakes: focal mechanisms put to work on India’s plate boundary and interior.
- Seismic Hazard and Earthquake Prediction: from source size and rate to shaking probabilities.
- Structural Geology and Geotectonics: fault mechanics and stress regimes behind the mechanisms.
- Seismology: Quick Revision.