Earthquake Depth Explained: Why Shallow Quakes Do More Damage

Quick Answer: Why Does Earthquake Depth Matter?

Because depth adds distance to the initial source. If an earthquake starts 10 km beneath your feet, you are 10 km from its hypocenter, the point where rupture begins. At 50 km depth, that initial point is 50 km away. A large earthquake ruptures an extended fault, however, and parts of that rupture can lie much closer to you, sometimes reaching the surface.

Depth	USGS Class	What It Usually Means at the Surface
0 to 70 km	Shallow	Intense shaking concentrated near the epicenter. Most damaging earthquakes live here.
70 to 300 km	Intermediate	Weaker shaking spread across a wider area. Still capable of killing.
300 to 700 km	Deep	Felt across continents, but rarely destructive.

Roughly four out of five earthquakes above magnitude 5 are shallow, which is why "earthquake" and "shallow earthquake" are nearly synonymous in practice.

The one-line version: magnitude tells you how big the earthquake was, depth tells you how much of it reaches you.

Enlarge

Depth Is Distance, and That Is Most of the Story

For someone standing directly above an earthquake's hypocenter, the distance to that starting point is the focal depth. Treating the source as a single point is a useful approximation when the fault rupture is small compared with the distance to the observer.

For large earthquakes nearby, distance to the rupturing fault matters too. A rupture beginning 10 km down can extend upward or sideways, putting some observers much closer to slipping rock than the hypocentral distance suggests. USGS finite-fault models describe that extended source. Greater depth generally weakens surface shaking, but focal depth alone is not a minimum distance to every part of the rupture.

Three separate effects then weaken the waves along that path:

1. Geometric spreading. Energy from the source expands over a growing spherical wavefront, so body-wave amplitude falls off roughly as one over the distance. Triple the path length and you cut the amplitude to about a third before the rock has absorbed anything at all.
2. Anelastic attenuation. Rock is not perfectly elastic. It absorbs wave energy and turns it into heat through internal friction, and it does this unevenly across frequencies: high frequencies get stripped out faster than low ones over the same path.
3. Scattering. Waves bounce off small-scale irregularities in the rock, dispersing energy that would otherwise arrive together.

That second effect explains something you may have noticed. Shaking close to an earthquake feels like a sharp jolt or a bang, because the high frequencies survive the short trip. Shaking far from the same earthquake feels like slow, queasy rolling, because only the low frequencies made it.

The Textbook Pair: Northridge and Nisqually

Two American earthquakes make the depth effect about as clean as reality ever gets.

	Northridge, CA (1994)	Nisqually, WA (2001)
Magnitude	M6.7	M6.8
Depth	18.2 km	51.8 km
Max ShakeMap intensity	8.8 MMI	6.9 MMI
Deaths	57	None listed by USGS
Injuries	More than 9,000	About 400
Losses	About $20 billion	$1 to $4 billion

Nisqually was slightly larger in magnitude but produced much lower reported losses. Its greater depth was an important factor; exposure, ground conditions, and building vulnerability also affect losses.

USGS collects public "Did You Feel It?" reports and bins them by hypocentral distance, the straight-line distance from the point where rupture began. Comparing these events at matched hypocentral distances illustrates the importance of distance:

Hypocentral distance	Northridge median MMI	Nisqually median MMI
54.8 km	5.9	6.1
73 km	5.7	6.0
97.4 km	5.3	5.5
129.9 km	4.6	4.8

These bins show similar reported intensities at comparable distances from each hypocenter, with Nisqually slightly higher in each of the four bins.

Northridge's nearer reports reach a median intensity of 8.2 in the 17.3 km bin, while Nisqually's nearest listed bin is 6.1 at 54.8 km. Those are summaries of reports, not theoretical limits on shaking or distances to the entire fault rupture. They support the role of depth without isolating it from all the other differences between these earthquakes.

Nisqually's deeper source increased the distance seismic waves had to travel to reach the surface.

Same Magnitude, Opposite Worlds

2015 ran the experiment even more starkly, with two M7.8 earthquakes five weeks apart.

25 April, Gorkha, Nepal. M7.8 at 8.2 km. Maximum ShakeMap intensity 8.7. USGS issued a red alert. At least 8,669 killed (a figure that includes the M7.3 aftershock in May), 17,866 injured, and more than 500,000 houses destroyed.

30 May, Bonin Islands, Japan. M7.8 at 664 km. Maximum ShakeMap intensity 3.4. Green alert. Twelve people injured in Tokyo, some trains stopped, 400 houses briefly lost power. Nobody died.

Same magnitude. Same year. One of them was a national catastrophe and the other was a news item.

Deep Earthquakes Are Felt Farther, Not Harder

Depth does not only weaken shaking. It spreads it out.

The M8.3 Sea of Okhotsk earthquake of May 2013, about 600 km deep, was reported felt in Moscow, Delhi, Tokyo, Vancouver, Honolulu, Seattle and California. Its maximum reported intensity anywhere on Earth was V. Nobody was killed and nothing significant was damaged. The 1994 M8.2 Bolivia earthquake at 631 km was felt in Toronto, Boston, Chicago and Minneapolis, the first South American earthquake from that region believed to have been felt in North America.

Two things widen that felt area:

• Geometry. At 600 km depth, moving 200 km sideways along the surface barely changes your distance to the source. The intensity field goes nearly flat: low everywhere, but low across an enormous area, with no sharp peak above the epicenter.
• Weak surface waves. Deep sources excite surface waves only feebly. Surface waves are what normally carry damaging regional shaking from a shallow quake, so their near-absence removes the main mechanism for concentrated destruction. Published corrections to surface-wave magnitude for deep events run to roughly half a magnitude unit, which is a measure of how much weaker that wave type gets.

It is worth resisting the popular shorthand that "the mantle transmits waves better than the crust." It is not uniformly true, and the asthenosphere is in fact the most strongly attenuating layer in the upper mantle. Geometry and weak surface-wave excitation carry the explanation on their own.

In subduction zones there is a third, stranger effect: the cold subducted slab acts as a waveguide for high-frequency energy, so deep earthquakes beneath Japan routinely produce a band of unusually strong shaking along the Pacific coast hundreds of kilometres from the epicenter, with intensity contours stretched along the slab instead of ringed around the source.

Why Do Deep Earthquakes Exist at All?

This is a genuinely unsolved problem, and it is more interesting than the textbook version suggests.

A shallow earthquake is brittle failure: stress builds on a fault until friction gives way and the rock slips. That mechanism should be impossible at 600 km. Down there, pressure clamps faults shut with enormous normal stress, and temperature is high enough that rock responds to stress by creeping slowly, like extremely stiff putty, rather than snapping.

To be clear about what is not happening: the mantle at those depths is solid. It is not molten. It deforms by slow ductile flow, and that is precisely why sudden brittle fracture should not be available as an option.

Yet the depth distribution of earthquakes is not a simple decline. It is bimodal: a huge shallow peak, a minimum near 300 km, and then a distinct second peak near 600 km. Something down there is making earthquakes on purpose.

Three mechanisms are on the table, and none has been ruled out:

Proposed mechanism	The idea
Transformational faulting	Olivine trapped in a metastable state past its stability field transforms suddenly to a denser mineral phase, and the transformation itself nucleates a fault.
Dehydration embrittlement	Water released from hydrous minerals raises pore pressure, unclamping faults enough for brittle failure to occur.
Thermal shear runaway	Localized shear heats a narrow zone, which weakens it, which concentrates more shear, in a feedback that ends in sudden slip.

The most recent comprehensive review of the field declines to reject any of the three. Anyone who tells you the question is settled is ahead of the evidence.

What we can say confidently is where: deep earthquakes occur inside subducting slabs, always. And they are astonishingly concentrated, with something over 70 percent of deep-focus events occurring in the Tonga-Kermadec-Fiji region alone.

Below roughly 700 km they essentially stop. Note carefully what stops: earthquake generation stops, not the slab. Subducting slabs demonstrably continue down into the lower mantle in several regions. They simply stop producing earthquakes.

As for the exact floor, honesty demands a range rather than a number. Dedicated studies of the deepest events found no reliably determined focal depths beyond about 680 to 690 km. The familiar 700 km is a round-number convention. A 2021 paper reported aftershocks near 751 km and drew heavy press coverage, but a 2025 reanalysis of the same sequence found no evidence for it. Treat any single "deepest earthquake ever" figure with suspicion.

When Shallow-Is-Worse Breaks Down

Depth is a strong modifier. It is not the main control, and reality is full of cases where something else won.

Distance beats depth. Canterbury, New Zealand ran this experiment twice. In September 2010 the Darfield earthquake, magnitude 7.1 in New Zealand usage and M7.0 per USGS, struck at 10 km depth about 40 km west of Christchurch at 4:35 in the morning. Severe damage, zero deaths. Five months later, in February 2011, a much smaller earthquake, 6.3 locally and M6.1 per USGS, struck at about 5 km depth within 10 km of the city centre at lunchtime on a working day. It killed 185 people. The smaller earthquake killed 185 and the larger killed nobody.

Soft ground beats depth. Nisqually was 52 km deep, yet its maximum intensity of VIII landed on soft ground in Olympia and Seattle's Pioneer Square, with liquefaction and sand blows. A USGS study of 35 Seattle sites found soft-soil sites on artificial fill and young alluvium amplified 1 Hz motion by factors of three to seven relative to soft rock, and that those amplifications matched where the damage actually occurred.

Basin resonance beats depth spectacularly. The 1985 M8.0 Michoacán earthquake had its epicenter 384 km from Mexico City. USGS records the maximum Modified Mercalli intensity as IX at Mexico City, equal to the maximum recorded on the Pacific coast directly above the rupture. The city's soft lakebed clay resonated with the arriving long-period waves. At least 9,500 people were killed, and USGS notes some sources go far higher. In 2017 it happened again: the M7.1 Puebla-Morelos earthquake was 48 km deep with its epicenter roughly 119 km from Mexico City, and it still killed at least 220 people in the city alone.

Deep does not mean safe. The October 2015 M7.5 Hindu Kush earthquake was 231 km deep, genuinely intermediate-depth, and it killed roughly 400 people across Afghanistan and northern Pakistan, destroying more than 29,000 buildings. Vulnerable unreinforced masonry turned moderate shaking into mass casualties.

Building codes beat depth. The 2010 Haiti earthquake was M7.0 at 13 km. The 2018 Anchorage earthquake was M7.1 at 47 km. Depth is part of that difference, but the decisive variable is that Anchorage builds to a modern seismic code. Anchorage recorded zero deaths. Haiti's toll is officially estimated at 316,000, though USGS itself flags that other estimates run far lower.

Landslides ignore depth entirely. The 2001 El Salvador earthquake was M7.7 at 60 km depth. It killed at least 844 people, of whom about 585 died in landslides at Nueva San Salvador and Comasagua.

There is also a subtlety that cuts the other way. Intermediate-depth earthquakes inside subducting slabs tend to have higher stress drops than shallow crustal earthquakes, radiating energy in a shorter, higher-frequency burst. A 50 km-deep intraslab earthquake is not simply "a crustal earthquake, further away." It hits harder per unit of distance in exactly the frequency band that damages short, stiff buildings.

Tsunamis: The One Place Shallow Is Unambiguously Worse

For tsunamis, depth is close to a hard gate.

A tsunami is not created by shaking. It is created by vertical displacement of the seafloor, which lifts or drops the entire water column above it. That requires a shallow marine earthquake, and thrust faulting is far more likely to do it than strike-slip motion.

A deep earthquake, however large, is simply too far beneath the seafloor to deform it meaningfully. The 2018 M8.2 Fiji earthquake at 600 km depth produced only a small tsunami. Tsunami warning centres reflect this operationally: they screen on magnitude, on an undersea or nearshore location, and on depth under about 100 km. That depth cutoff is a policy threshold chosen by warning centres, not a law of physics, but it encodes a real one.

The dangerous inversion is worth knowing. So-called "tsunami earthquakes" rupture slowly, produce surprisingly mild felt shaking, and generate outsized waves. Weak shaking near a coast is not evidence that you are safe.

The Depth in the First Report Is Often Not a Measurement

One practical warning before you start reading depth off a live feed.

USGS assigns 10 km as a default depth when the available data cannot constrain it. In the USGS catalog from 2015 through 2025, about 44 percent of M5.0+ earthquakes are listed at exactly 10.0 km. That clustering is far too sharp to be natural.

The reason is geometric. A useful rule of thumb from USGS is that a reliable depth requires the nearest seismic station to be closer to the epicenter than the earthquake is deep. For an offshore earthquake with no station within a hundred kilometres, that test fails immediately, and depth trades off against origin time in the solution.

USGS states plainly that determining depth is harder than determining location, and that depth errors are larger than location errors. So an exact 10.0 km on a fresh event is often a placeholder. It frequently changes, as does the magnitude, which we cover in why earthquake magnitudes change after the first report.

How to Read Depth on a Live Tracker

1. Read depth and magnitude together. Neither means much alone. A M5.5 at 8 km under a town is a bigger problem than a M7 at 300 km under open ocean.
2. Treat exactly 10.0 km with suspicion on a fresh event, especially offshore.
3. Under about 70 km, expect the shaking to be concentrated. Above roughly 300 km, expect it to be broad and mild.
4. Ask what is above it, not just how deep it is. Population, soil, and building stock decide outcomes at least as much as depth does.
5. For tsunami potential, shallow plus marine plus large is the combination that matters. Depth alone rules things out more reliably than it rules them in.

Our live earthquake tracker shows depth alongside magnitude and felt reports for every event in the USGS feed, and links each one to its official USGS page.

FAQ

Why do shallow earthquakes cause more damage?

Shallower sources generally leave less distance for seismic waves to travel before reaching the surface, so less energy is lost to spreading and attenuation. Focal depth measures where rupture begins, not the distance to every part of the fault: a large rupture can extend closer to the surface. Shaking also depends on magnitude, distance to the rupture, ground conditions, and the rupture process.

What is considered a shallow earthquake?

USGS classifies earthquakes shallower than 70 km as shallow, 70 to 300 km as intermediate, and 300 to 700 km as deep. These are conventions rather than physical boundaries, and roughly four in five earthquakes above magnitude 5 fall in the shallow band.

Can deep earthquakes cause damage?

Yes. The 2015 M7.5 Hindu Kush earthquake was 231 km deep and killed roughly 400 people, and the 2017 M7.1 Puebla earthquake at 48 km killed at least 220 in Mexico City. Depth reduces shaking intensity, but vulnerable buildings and resonant soils can convert moderate shaking into a disaster.

What is the deepest earthquake ever recorded?

There is no clean answer. Dedicated studies of the deepest events found no reliably determined depths beyond about 680 to 690 km, and the familiar 700 km figure is a round-number convention. A 2021 study reported events near 751 km, but a 2025 reanalysis found no evidence for them.

Why do earthquakes stop at around 700 km?

Nobody is certain. At those pressures and temperatures rock should deform by slow creep rather than fracture, so ordinary brittle failure is ruled out and the deep earthquakes that occur need a special mechanism. Transformational faulting, dehydration embrittlement, and thermal shear runaway are all candidates, and the current review literature rejects none of them. Note that the subducting slabs themselves continue deeper; only the earthquakes stop.

Do deep earthquakes cause tsunamis?

Essentially never. A tsunami requires vertical displacement of the seafloor, and a deep earthquake is too far beneath it to move it meaningfully. Tsunami warning centres screen for shallow marine events, generally under about 100 km depth.

Why are so many earthquakes reported at exactly 10 km deep?

Because 10 km is the USGS default when the data cannot resolve depth. About 44 percent of M5.0+ events in the recent USGS catalog carry exactly that value. Reliable depth generally requires a seismic station closer to the epicenter than the earthquake is deep, which often fails offshore.

Always Verify with Official Sources

Earthquake depths are revised as analysis continues, and early values are the least reliable parameter in the solution. For authoritative event details use the USGS Earthquake Hazards Program, for tsunami alerts use the official tsunami warning centres, and follow your local emergency management agency for protective action.

Track Earthquake Depth with WeatherAI

WeatherAI shows depth alongside every earthquake it reports, on iPhone, iPad, Mac, and Vision Pro:

• Full event detail including depth, magnitude type, and location, with maps
• Magnitude and distance filtering so routine small events stay out of your way
• Nearby and global earthquake notifications with thresholds you set
• USGS or EMSC as your data source, whichever suits your region
• Tsunami bulletins surfaced alongside earthquake information

WeatherAI reports earthquakes that have already been detected, so it is monitoring rather than earthquake early warning, and it does not replace official emergency alerts.

Download WeatherAI for iPhone →

---

Keep reading: why earthquake magnitudes change after the first report, Richter vs moment magnitude, or look up focal depth, hypocenter and subduction zone in our glossary.