Earthquakes: Causes, Measurement and Seismicity
CSIR NET Earth Sciences: elastic rebound, focal depth, location, magnitude and moment, intensity, Gutenberg-Richter, focal mechanisms, inter- and intraplate.
- 20 min
On this page · 14 sections
Prerequisites
- P, S and surface waves and their speeds from Seismology Basics and Earth’s Internal Structure.
- Stress, Coulomb failure, pore pressure and fault types from Stress, Strain, Folds, Joints and Faults.
- Logarithms and powers of ten. If they feel slow, revise Numerical Ability; log–linear plots are covered in Graphs of Functions and Scientific Plots.
The Hook: The Fence That Moved
What an Earthquake Is
An earthquake is ground shaking caused by a sudden release of stored energy, almost always by rapid slip on a fault. The point inside the Earth where rupture starts is the focus (hypocentre); the point on the surface directly above it is the epicentre. The rupture then spreads over a fault area that, for a great earthquake, can be hundreds of kilometres long.
Elastic rebound and the earthquake cycle
graph LR
A[Plates move steadily] --> B[Locked fault: elastic strain builds up]
B --> C[Stress reaches fault strength]
C --> D[Sudden slip: earthquake, seismic waves]
D --> E[Stress drops; aftershocks]
E --> B- Interseismic period: slow loading, measured today by GPS and InSAR.
- Coseismic slip: seconds to minutes; slip of centimetres to tens of metres.
- Postseismic period: aftershocks, afterslip and viscous relaxation.
Faults can also slip without earthquakes (aseismic creep), or slowly over days to weeks in slow-slip events, often with weak tremor. Rocks that weaken as they slip (velocity-weakening friction) are needed for earthquakes; velocity-strengthening fault zones creep.
Causes
| Type | Mechanism | Example or note |
|---|---|---|
| Tectonic | slip on faults driven by plate motions; by far the most common and the largest | plate-boundary and intraplate earthquakes |
| Volcanic | magma movement, pressure changes, volcano-tectonic faulting; usually small | swarms before eruptions; harmonic tremor |
| Collapse | roof collapse of caves and mines; small | karst areas, old mines |
| Induced | human activity changes stress or pore pressure | reservoir-triggered (Koyna, 1967), mining, fluid injection and wastewater disposal, geothermal stimulation |
| Explosions | nuclear tests, quarry blasts | mostly isotropic P radiation, weak S and surface waves; used to discriminate tests |
| Impacts | meteorite impact | rare |
Reservoir-triggered seismicity combines the load of the water with pore-pressure diffusion that lowers the effective normal stress on faults already near failure (the Coulomb mechanism on the previous page). Seismicity at Koyna, Maharashtra, has continued since impoundment of the reservoir in the early 1960s, and the 1967 Koyna earthquake (magnitude about 6.3) is widely cited as the largest reservoir-triggered earthquake.
Focal depth
| Class | Depth | Where |
|---|---|---|
| Shallow | 0–70 km | everywhere; most earthquakes and most damage |
| Intermediate | 70–300 km | subduction zones |
| Deep | 300–700 km | subduction zones only; none deeper than about 700 km |
Intermediate and deep earthquakes outline the dipping Wadati–Benioff zone of a subducting slab. Deep earthquakes cannot be ordinary frictional failure at such pressures; proposed mechanisms include dehydration embrittlement (intermediate depths) and transformational faulting in metastable olivine (deep). Their cut-off near 700 km lies close to the 660 km discontinuity.
Measuring Earthquakes
Seismometers
A seismometer is an inertial sensor: a mass on a spring (or pendulum), which tends to stay still while the frame moves with the ground. The relative motion is recorded (a seismogram). Response depends on the natural period and damping:
- for ground periods much shorter than the natural period, the instrument records displacement;
- near its natural period it resonates unless damped (usually close to critical damping).
Modern broadband force-feedback seismometers record from about 0.01 s to hundreds of seconds in three components (vertical, N–S, E–W). Accelerometers (strong-motion instruments) stay on scale in very strong shaking near the source.
Locating an earthquake
- S–P time. P and S leave together; the delay grows with distance:
For crustal values \( V_P = 6.0 \), \( V_S = 3.5 \) km s\(^{-1}\) the factor is 8.4 km s\(^{-1}\), the “8 km per second of S–P” rule of thumb.
- Three stations give three distance circles that intersect at the epicentre. Locating the hypocentre and origin time needs four unknowns \( (x, y, z, t_0) \), so at least four arrival times; in practice many stations and least-squares inversion.
- Wadati diagram. Plot \( t_S - t_P \) against \( t_P \) for several stations: a straight line with slope \( V_P/V_S - 1 \), which meets \( t_S - t_P = 0 \) at the origin time. It needs no velocity model.
- Depth. The delay of the surface-reflected depth phase pP behind P is the most reliable measure of focal depth for distant earthquakes.
Magnitude
Magnitude measures the size of the source, one number per earthquake. It is logarithmic: one unit means 10 times the wave amplitude.
| Scale | Measured from | Note |
|---|---|---|
| Local, \( M_L \) (Richter, 1935) | maximum amplitude on a standard Wood–Anderson torsion seismometer, corrected for distance: \( M_L = \log_{10}A - \log_{10}A_0(\Delta) \) | defined for southern California, distances under about 600 km |
| Body-wave, \( m_b \) | amplitude and period of short-period (about 1 s) P waves: \( m_b = \log_{10}(A/T) + Q(\Delta, h) \) | for distant and deep events; saturates around 6–6.5 |
| Surface-wave, \( M_s \) | Rayleigh waves of about 20 s period: \( M_s = \log_{10}(A/T) + 1.66\log_{10}\Delta + 3.3 \) (\( A \) in μm, \( \Delta \) in degrees) | shallow events; saturates around 8 |
| Moment, \( M_w \) (Kanamori, 1977; Hanks and Kanamori, 1979) | seismic moment \( M_0 \) | does not saturate; the standard for large earthquakes |
Saturation. Large ruptures radiate most energy at long periods. A scale that reads short-period waves stops growing even as the earthquake gets bigger, so \( m_b \) and \( M_s \) underestimate great earthquakes.
Seismic moment is a physical measure of the faulting:
$$\boxed{M_0 = \mu A D}$$with \( \mu \) the shear modulus (about \( 3\times 10^{10} \) Pa in the crust), \( A \) the rupture area and \( D \) the average slip. Units: N m. Then
$$\boxed{M_w = \frac{2}{3}\left(\log_{10}M_0 - 9.1\right)} \qquad (M_0 \text{ in N m})$$Energy. The Gutenberg–Richter energy relation is
$$\boxed{\log_{10}E = 1.5M + 4.8} \qquad (E \text{ in J})$$so each unit of magnitude means about \( 10^{1.5} \approx 31.6 \) times more energy, and two units mean 1,000 times. \( M_0 \) also scales as \( 10^{1.5M_w} \).
Intensity
Intensity describes the effects of shaking at a place: on people, buildings and ground. One earthquake has one magnitude but many intensities, which generally decrease away from the epicentre.
| Scale | Levels | Note |
|---|---|---|
| Modified Mercalli (MMI) | I–XII | widely used, especially in the Americas |
| MSK-64 (Medvedev–Sponheuer–Kárník) | I–XII | used in India and much of Eurasia |
| EMS-98 (European Macroseismic Scale) | I–XII | successor to MSK in Europe; accounts for building vulnerability |
| JMA (Japan) | 0–7 |
Lines joining places of equal intensity are isoseismals; the zone of highest intensity is the meizoseismal area. Intensity depends on magnitude, distance, depth, local site conditions (soft sediments amplify shaking; saturated sand can liquefy) and construction.
Instrumental measures of shaking are peak ground acceleration (PGA), velocity and the response spectrum, used in seismic design. In India, the seismic code IS 1893 (Part 1) divides the country into seismic zones; the 2016 edition uses zones II to V with zone factors 0.10, 0.16, 0.24 and 0.36. Check the current BIS edition before quoting zone details.
Earthquake Statistics
Gutenberg–Richter law
In a region over a period, the number \( N \) of earthquakes with magnitude at least \( M \) follows
$$\boxed{\log_{10}N = a - bM}$$- \( a \) measures the overall activity rate of the region.
- \( b \) (the b-value) is close to 1 for most tectonic regions: each step down of one magnitude unit gives about ten times as many earthquakes. Higher \( b \)-values are reported for volcanic swarms and some induced seismicity.
- The recurrence interval of events of size \( M \) or more is \( 1/N(M) \).
If earthquakes are random in time (Poisson), the probability of at least one event with rate \( \nu \) per year in \( t \) years is \( P = 1 - e^{-\nu t} \).
Aftershocks and foreshocks
- Omori’s law (in the modified form of Utsu): the aftershock rate decays as \( n(t) = \dfrac{K}{(c + t)^p} \) with \( p \approx 1 \).
- Båth’s law: the largest aftershock is typically about 1.2 magnitude units smaller than the mainshock.
- Foreshocks precede some mainshocks, but they cannot be recognised as foreshocks until afterwards.
- A swarm is a sequence with no single dominant mainshock (common in volcanic and geothermal areas).
Forecasting, not predicting
No method reliably predicts the time, place and size of a damaging earthquake days in advance. What exists:
- Long-term forecasts: recurrence intervals from palaeoseismology (trenching across faults), geodetic strain rates and catalogues; seismic gaps are segments of an active boundary that have not ruptured for a long time.
- Probabilistic seismic hazard maps (probability of exceeding a given shaking level).
- Earthquake early warning: detects the fast, weaker P wave near the source and warns before the damaging S and surface waves arrive farther away; lead times are seconds to tens of seconds.
Focal Mechanisms
The pattern of first-motion polarities around an earthquake shows how the fault slipped. P waves leave as compressions (first motion up on a vertical seismometer) in two opposite quadrants and dilatations (first motion down) in the other two. The quadrants are separated by two perpendicular nodal planes: one is the fault plane, the other the auxiliary plane. First motions alone cannot tell which is which; aftershock alignments, surface rupture or geology decide.
On a lower-hemisphere “beach ball”, compressional quadrants are usually shaded:
| Mechanism | Beach ball | Stress |
|---|---|---|
| Normal | white (dilatational) centre, shaded sides | P axis near vertical |
| Thrust (reverse) | shaded centre, white sides | T axis near vertical |
| Strike-slip | four-quadrant cross, nodal planes vertical | P and T horizontal |
The P axis (pressure) lies in the middle of the dilatational quadrant and the T axis (tension) in the middle of the compressional quadrant. This sounds backwards but follows from the motion: material is pushed outward (compression first motion) in the direction it is being pulled towards.
Modern catalogues use moment tensors inverted from whole waveforms. They also capture non-double-couple sources such as explosions (isotropic) and some volcanic events.
Where Earthquakes Happen
Global belts
| Belt | Character |
|---|---|
| Circum-Pacific (“Ring of Fire”) | the largest share of global seismicity and seismic energy; subduction zones and transforms; all depth classes |
| Alpine–Himalayan (Mediterranean–Trans-Asiatic) | continental collision and its spreading margins; mostly shallow, with intermediate depths in places (Hindu Kush, Burma arc) |
| Mid-ocean ridges | narrow belts of shallow, small to moderate earthquakes |
| Continental rifts and interiors | East African Rift; scattered intraplate events |
The narrow belts of seismicity were among the strongest evidence for plate tectonics: they outline the plates (see Paleomagnetism, Sea-Floor Spreading and Plate Tectonics).
Interplate seismicity
Interplate earthquakes occur at plate boundaries and release the overwhelming majority of seismic energy.
| Boundary | Depth | Mechanism | Size |
|---|---|---|---|
| Mid-ocean ridge (spreading segments) | shallow (under about 10 km) | normal | small to moderate |
| Oceanic transform | shallow | strike-slip, only between ridge offsets | moderate |
| Subduction zone | shallow to about 700 km | megathrust (thrust) at the interface; normal and others within the slab and outer rise | the largest (\( M_w \) 9+) |
| Continental collision | mostly shallow, spread over a wide zone | thrust, with strike-slip and normal behind | large (\( M_w \) 8+) |
| Continental transform | shallow | strike-slip | large (\( M_w \) about 7–8) |
The size of the largest possible earthquake is set by the area of fault that can rupture at once. Shallow-dipping subduction interfaces offer the widest seismogenic zones, so they host the largest earthquakes; ridges, with thin hot lithosphere, host only small ones.
India’s plate-boundary earthquakes. The Himalayan arc (India–Eurasia collision) has had great earthquakes such as 1897 Shillong, 1905 Kangra, 1934 Bihar–Nepal, 1950 Assam (about \( M_w \) 8.6, among the largest continental earthquakes recorded) and 2015 Gorkha, Nepal (\( M_w \) about 7.8). The stretch between the 1905 and 1934 ruptures has been discussed as the central seismic gap. The Andaman–Sumatra subduction zone produced the 26 December 2004 earthquake and Indian Ocean tsunami.
Intraplate seismicity
Intraplate earthquakes occur away from plate boundaries, in stable continental regions (SCRs) and old oceanic lithosphere. They are rarer, but:
- they are usually shallow (upper crust) and can be large (magnitude 7 or more);
- strain rates are tiny, so recurrence intervals are long (often thousands of years) and hard to estimate;
- seismic waves travel efficiently through cold, old continental lithosphere (low attenuation), so a given magnitude shakes a larger area;
- they tend to occur on reactivated ancient structures, especially failed rifts and old sutures, where intraplate stress (transmitted from plate boundaries by ridge push and collision) exceeds the strength of a weak zone;
- local triggers include reservoir loading, fluid injection and erosion or unloading.
| Earthquake | Setting |
|---|---|
| New Madrid, USA (1811–1812) | Reelfoot rift, a failed rift in the central US |
| Koyna, Maharashtra (1967) | reservoir-triggered, Deccan Volcanic Province |
| Latur (Killari), Maharashtra (30 September 1993) | stable Deccan craton, no prior known activity; about \( M_w \) 6.1–6.2; very high death toll from poor masonry |
| Jabalpur (1997) | Narmada–Son lineament |
| Bhuj, Gujarat (26 January 2001) | Kachchh (Kutch) palaeo-rift, reverse faulting; about \( M_w \) 7.7 |
The Indian peninsula is an SCR, but it is under compression from the India–Eurasia collision, which helps explain its intraplate earthquakes.
Effects
- Ground shaking, the main cause of damage; surface rupture along the fault.
- Liquefaction of saturated loose sand, causing sand boils, lateral spreading and loss of bearing capacity.
- Landslides, especially in mountains.
- Tsunamis: caused by sudden vertical displacement of the seafloor (mostly by megathrust earthquakes), or by submarine landslides and volcanic collapse; in deep water they travel at \( \sqrt{gh} \), several hundred km per hour, with long wavelengths and small heights that grow on reaching shallow water.
- Fires and failure of dams and infrastructure.
Memory Tricks
Don’t Confuse
| This | vs | That |
|---|---|---|
| Magnitude: one value, source size, instrumental, logarithmic | Intensity: varies by place, effects, Roman numerals | |
| Focus (hypocentre): where rupture starts, at depth | Epicentre: point on the surface above it | |
| \( M_L, m_b, M_s \): amplitude-based, saturate | \( M_w \): moment-based, does not saturate | |
| Seismic moment \( M_0 = \mu AD \) (N m) | Radiated energy \( E \) (J) | |
| Fault plane | Auxiliary plane (both are nodal planes) | |
| P axis: in the dilatational quadrant | T axis: in the compressional quadrant | |
| Interplate: boundaries, frequent | Intraplate: interiors, rare, reactivated weak zones | |
| Earthquake forecast: probability over years | Prediction: time, place and size in advance (not yet possible) |
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
- What is the depth range of intermediate-focus earthquakes?
- Which magnitude scale does not saturate for great earthquakes?
- State the typical b-value of the Gutenberg–Richter law.
Answers
- 70–300 km.
- Moment magnitude, \( M_w \), based on seismic moment.
- About 1.
Level 2: Part B Style
Practice Q1. At a station the S wave arrives 10 s after the P wave. With \( V_P = 6.0 \) and \( V_S = 3.5 \) km s\(^{-1}\), the hypocentral distance is:
(a) 25 km (b) 60 km (c) 84 km (d) 140 km
Answer: (c) 84 km
\( d = \dfrac{\Delta t\,V_PV_S}{V_P - V_S} = \dfrac{10\times 6.0\times 3.5}{2.5} = 84 \) km.
(b) uses \( V_P\Delta t \); (a) uses \( (V_P - V_S)\Delta t \).
Practice Q2. Compared with a magnitude 5 earthquake, a magnitude 7 earthquake releases about how much more energy?
(a) 2 times (b) 100 times (c) 1,000 times (d) 10,000 times
Answer: (c) 1,000 times
\( E_2/E_1 = 10^{1.5\times 2} = 10^{3} \). (b) is the amplitude ratio.
Practice Q3. A fault ruptures an area of 100 km × 20 km with 2 m average slip; \( \mu = 3\times 10^{10} \) Pa. The moment magnitude is about:
(a) 6.6 (b) 7.0 (c) 7.3 (d) 8.0
Answer: (c) 7.3
\( M_0 = 3\times 10^{10}\times 2\times 10^{9}\times 2 = 1.2\times 10^{20} \) N m; \( M_w = \frac{2}{3}(20.08 - 9.1) \approx 7.32 \).
Using km² instead of m² for the area (\( 2\times 10^3 \)) would give a nonsensically small value; check units first.
Practice Q4. A lower-hemisphere focal mechanism has a shaded (compressional) centre and white sides, with both nodal planes striking roughly east–west. The earthquake was most likely:
(a) a normal-fault event under N–S extension (b) a thrust event under N–S compression (c) a strike-slip event (d) an explosion
Answer: (b)
A compressional centre means the T axis is near vertical and the P axis near horizontal: thrust faulting. With E–W striking planes the P axis is N–S, so the compression is N–S. An explosion gives compression first motions everywhere.
Level 3: Part C Style
Practice Q5. A region’s catalogue gives 200 earthquakes per year of magnitude 4 or more and 20 per year of magnitude 5 or more. Assuming the Gutenberg–Richter law holds to larger magnitudes and events are Poissonian, which statements are correct?
- The b-value is 1.
- The mean recurrence interval of magnitude 7 or larger events is 5 years.
- The probability of at least one magnitude 7 or larger event in 10 years is about 86%.
- The a-value is about 6.3.
(a) 1 and 2 only (b) 1, 2 and 4 only (c) 1, 2, 3 and 4 (d) 2 and 3 only
Answer: (c) 1, 2, 3 and 4
- \( b = \log_{10}(200/20)/(5 - 4) = 1 \). True.
- \( a = \log_{10}200 + 4 \approx 2.301 + 4 = 6.301 \). True.
- \( N(\geq 7) = 10^{6.301 - 7} = 10^{-0.699} = 0.2 \) per year, so recurrence \( = 1/0.2 = 5 \) years. True.
- \( P = 1 - e^{-0.2\times 10} = 1 - e^{-2} \approx 0.865 \). True.
The catch in real life: extrapolating to large magnitudes assumes the region has faults big enough to host them.
Practice Q6. P and S arrival times at four stations, in seconds after 10:00:00 UTC:
| Station | \( t_P \) | \( t_S - t_P \) |
|---|---|---|
| A | 7.0 | 3.75 |
| B | 11.0 | 6.75 |
| C | 15.0 | 9.75 |
| D | 19.0 | 12.75 |
From a Wadati diagram, the origin time and \( V_P/V_S \) are:
(a) 10:00:00, 1.73 (b) 10:00:02, 1.75 (c) 10:00:02, 1.50 (d) 10:00:07, 1.75
Answer: (b) 10:00:02, 1.75
Slope: \( \dfrac{12.75 - 3.75}{19.0 - 7.0} = \dfrac{9.0}{12.0} = 0.75 = V_P/V_S - 1 \), so \( V_P/V_S = 1.75 \).
Origin time: where \( t_S - t_P = 0 \): \( t_P = 7.0 - 3.75/0.75 = 7.0 - 5.0 = 2.0 \) s, i.e. 10:00:02.
The corresponding Poisson’s ratio is \( \dfrac{1.75^2 - 2}{2(1.75^2 - 1)} \approx 0.258 \), close to a Poisson solid.
Practice Q7. Which statements about intraplate earthquakes in stable continental regions are correct?
- They commonly occur on reactivated ancient rifts or sutures.
- For the same magnitude, they are often felt over a larger area than plate-boundary earthquakes, because attenuation in old continental lithosphere is low.
- Their recurrence intervals on a given fault are usually short (decades), so hazard is easy to estimate.
- The 2001 Bhuj earthquake occurred within the Indian plate, in the Kachchh palaeo-rift region.
(a) 1 and 2 only (b) 1, 2 and 4 only (c) 2, 3 and 4 only (d) 1, 3 and 4 only
Answer: (b) 1, 2 and 4 only
Statement 3 is false: strain rates inside plates are very low, so recurrence intervals are long (often thousands of years) and poorly known, which is exactly what makes intraplate hazard hard to assess. The others are standard features of SCR seismicity; Bhuj is the classic Indian example.
Quick Revision Table
| Idea | Fact or formula |
|---|---|
| Elastic rebound | Reid, after the 1906 San Francisco earthquake |
| Depth classes | shallow 0–70, intermediate 70–300, deep 300–700 km |
| S–P distance | \( d = \Delta t\,V_PV_S/(V_P - V_S) \); about 8 km per second |
| Location | 3 stations for epicentre; 4 unknowns \( (x, y, z, t_0) \) for hypocentre |
| Wadati diagram | slope \( V_P/V_S - 1 \); intercept gives origin time |
| Focal depth | pP − P delay |
| Magnitudes | \( M_L \) (Richter 1935), \( m_b \) (1 s P), \( M_s \) (20 s Rayleigh), \( M_w \) (moment) |
| Moment | \( M_0 = \mu AD \); \( M_w = \frac{2}{3}(\log_{10}M_0 - 9.1) \) |
| Energy | \( \log_{10}E = 1.5M + 4.8 \) J; ×31.6 per unit |
| Intensity | MMI, MSK-64 (India), EMS-98: I–XII; isoseismals |
| Gutenberg–Richter | \( \log_{10}N = a - bM \), \( b \approx 1 \) |
| Omori, Båth | rate \( \propto (c + t)^{-p} \), \( p \approx 1 \); largest aftershock about 1.2 below |
| Focal mechanism | thrust: shaded centre; normal: white centre; P axis in dilatational quadrant |
| Largest recorded | Chile 1960, \( M_w \) about 9.5 |
| Interplate | ridges (normal, shallow), transforms (strike-slip), subduction (megathrust, all depths) |
| Intraplate (India) | Koyna 1967 (reservoir), Latur 1993, Jabalpur 1997, Bhuj 2001 |
Key Takeaways
What’s Next
- Paleomagnetism, Sea-Floor Spreading and Plate Tectonics: the plate framework that explains where earthquakes happen.
- Stress, Strain, Folds, Joints and Faults: revisit the fault mechanics behind focal mechanisms.
- Interior of the Earth, Deformation and Tectonics: Quick Revision: the whole chapter on one page.