Earthquake and Tsunami Forecast 2026–2036

This article is part of The Purple Report September 2026 | Disaster Forecasting and Predictions and focuses specifically on earthquake and tsunami risk.

Earthquake forecasting is where the word prediction most easily becomes misleading.

Earthquake scientists can identify active faults, measure plate motion, reconstruct past ruptures, estimate recurrence, calculate shaking hazard and model tsunami propagation. They can also publish short-term aftershock forecasts after a major event.

What they cannot currently do is give a reliable exact date, location and magnitude for the next major earthquake.

The earthquake forecast therefore asks where large earthquakes are physically credible and consequential—not which fault will “go next”.

Quick Read

The September 2026 baseline contains 15 earthquake/tsunami watch systems.

Nankai Trough, Japan
Eastern Taiwan / Hualien–Taitung
Cascadia Subduction Zone
Puget Sound / Seattle–Tacoma
San Francisco Bay Area
Greater Los Angeles / Southern San Andreas system
Alpine Fault, New Zealand
Hikurangi–Wellington, New Zealand
Sunda Strait megathrust, Indonesia
Mentawai–Siberut megathrust, Indonesia
Metro Manila / West Valley Fault
Marmara Sea / Istanbul
Central Himalaya / Kathmandu
Central Peru coast / Lima
Central–southern Chile coast

These are not ranked from “most likely” to “least likely”. Some have quantified official probabilities. Others belong because the hazard/exposure/cascade combination is structurally important.

Earthquake Evidence Boundaries

Active fault ≠ imminent earthquake.
Seismic gap ≠ countdown clock.
Long elapsed time ≠ proof that rupture is “overdue”.
Probability over 50 years ≠ probability over 10 years unless the model supports conversion.
Earthquake magnitude ≠ damage. Distance, depth, ground motion, building vulnerability and basin effects matter.
Tsunami source ≠ tsunami catastrophe. Bathymetry, warning, evacuation and coastal geometry shape the consequences.
Scenario ≠ forecast. A planning scenario asks what could happen if a defined event occurs.

How Long-Term Earthquake Forecasting Actually Works

Large earthquakes are rare compared with weather events. That makes direct statistical forecasting difficult.

Scientists therefore combine several evidence families:

plate motion and geodesy — how fast crustal blocks move relative to one another;
fault geometry — which structures can rupture and how large a rupture could become;
paleoseismology — evidence of ancient earthquakes in sediments, coastlines, tsunami deposits or fault trenches;
historical seismicity — instrumented and documentary earthquake records;
ground-motion modelling — how shaking propagates through rock and sediment;
recurrence models — how often particular event classes have occurred under explicit assumptions;
site response — how local basins, soft soil and topography can amplify shaking.

The USGS National Seismic Hazard Model is a good example. It does not forecast one named earthquake. It calculates ground-shaking hazard across many possible earthquake sources and probability levels so building codes and infrastructure planning can use the result.

Why Nankai Is Different From a News Headline

Japan’s Earthquake Research Committee revised its Nankai Trough long-term evaluation in September 2025. The committee continues to place Nankai in its highest probability rank.

The official guide explains that the previous 30-year estimate was around 80% as of January 2025 and that the updated evaluation considers alternative calculation approaches rather than presenting one magical countdown number.

The correct civilisation interpretation is:

Nankai is a persistent high-priority preparedness problem. It is not a valid exact-date prediction.

The cascade is why it belongs prominently in the Purple Forecast: tsunami evacuation, Pacific-coast industry, ports, rail, roads, power, communications, hospitals and national supply chains can all be affected by one regional rupture.

Japan Earthquake Research Committee — Nankai Trough long-term evaluation guide

Taiwan Gives Us a Genuine 10-Year Earthquake Product

Taiwan is especially useful for this 2026–2036 project because the Central Weather Administration already publishes earthquake potential maps on 10-, 30- and 50-year horizons for M6.0, M6.7 and M7.0 events.

That does not mean a map pixel predicts where the next earthquake will occur.

It means this forecast project can preserve a native ten-year product rather than inventing one.

Taiwan also teaches another lesson: earthquake hazard does not stop when the shaking stops. Extreme relief and steep roads mean an earthquake can become a landslide and access event. A typhoon can later remobilise earthquake-weakened slopes.

Taiwan Central Weather Administration — 2026 earthquake potential maps

Cascadia: Low Annual Frequency, Enormous Cascade

USGS estimates a 10–15% chance of an approximately magnitude-9 full-margin Cascadia earthquake in the next 50 years.

That is not a high annual probability.

Yet Cascadia remains a major civilisation hotspot because of consequence.

near-field tsunami;
coastal subsidence;
bridges and road isolation;
ports and marine logistics;
power and fuel;
communications;
water and wastewater;
regional supply-chain disruption.

The same USGS work gives Puget Sound a different seismic profile: an 85% chance of M6.5+ deep earthquake and a 17% chance of M6.5+ crustal earthquake over 50 years.

These hazards overlap geographically but should not be collapsed into one earthquake object.

USGS — Earthquake probabilities and hazards in the U.S. Pacific Northwest

Alpine Fault: Probability Plus Isolation

New Zealand’s AF8 programme reports a 75% probability of an Alpine Fault earthquake within 50 years and a four-in-five chance that the next rupture will be magnitude 8 or greater.

The Purple Forecast gives this system special attention because the main civilisational problem is not simply shaking intensity.

The South Island is long, mountainous and crossed by roads, rail, power and communications corridors that can fail in multiple places simultaneously.

A major rupture could therefore create isolated communities even where buildings remain standing.

AF8 — Alpine Fault programme

Indonesia: Seismic Gap Is a Preparedness Signal, Not a Clock

BMKG’s discussion of the Sunda Strait and Mentawai–Siberut megathrust zones is almost a perfect statement of the boundary this project needs.

BMKG describes both as long-recognised seismic-gap concerns and explicitly warns the public not to interpret that language as an early warning that a large earthquake will happen in the near future.

BMKG also states plainly that science cannot currently predict exactly when, where and how strong a major earthquake will be.

So why include them?

Because a seismic gap is useful for mitigation priority.

It is not useful as a countdown clock.

BMKG — Sunda Strait and Mentawai–Siberut megathrust potential

Metro Manila: Scenario Is Not Prediction

PHIVOLCS’ West Valley Fault scenario models a magnitude-7.2 earthquake with severe impacts across Greater Metro Manila.

The scenario includes destructive shaking, liquefaction and very large building losses.

That is a planning tool.

It does not mean PHIVOLCS has predicted when the West Valley Fault will rupture.

The distinction is important because scenarios are enormously useful even without timing.

They reveal evacuation needs, hospital demand, building vulnerability, road loss, water interruption and governance continuity.

PHIVOLCS — earthquake scenarios and the West Valley Fault

Marmara, Kathmandu, Lima and Chile: Structural Watch Without False Timing

Some of the most important systems in this forecast do not have a simple public probability number that can be copied into a table.

Marmara contains active faulting beside one of the world’s largest urban economies.

Kathmandu lies inside the actively shortening Himalayan collision zone, with basin amplification and mountain-access consequences.

Lima and the central Peru coast sit over the Peru–Chile subduction margin, with tsunami and critical-infrastructure exposure.

Central and southern Chile have repeatedly experienced very large megathrust earthquakes and remain a major long-term seismic field.

The correct public state for these is STRUCTURAL WATCH.

The absence of a clean ten-year number should not be replaced by a number invented for visual symmetry.

Tsunami Risk Depends on Who and What Is Exposed

A megathrust earthquake offshore does not translate into the same tsunami consequence everywhere.

The consequences depend on:

rupture geometry and sea-floor displacement;
distance to shore;
bathymetry;
coastal shape;
local run-up;
warning time;
evacuation routes;
vertical evacuation options;
night/day population;
port and industrial exposure.

Near-field tsunami systems are especially unforgiving because people may have minutes rather than hours.

That makes public education—strong shaking, sudden sea-level change and unusual roaring—part of an effective warning system where official guidance supports it.

The Hidden Earthquake Cascades

Ground shaking is only the first layer.

Liquefaction can damage foundations, roads, ports and buried pipes.
Landslides can block mountain corridors long after the shaking ends.
Fire can follow broken gas and electrical networks.
Dam and reservoir damage can change flood risk.
Port closure can turn a local earthquake into a national logistics problem.
Water-system failure can lengthen the health emergency.
Hospital overload can cause indirect mortality.
Communications failure can make an otherwise manageable event harder to coordinate.

That is why the Purple Forecast does not judge an earthquake system by magnitude alone.

What Could Change the Watch State?

Long-term seismic hazard usually changes slowly.

But the Purple Report will still watch for state changes:

new official long-term probability updates;
new paleoseismic evidence;
new fault mapping;
major slow-slip or geodetic observations where authorities judge them relevant;
aftershock sequences after a large earthquake;
new tsunami inundation maps;
major changes in exposure, building vulnerability or evacuation capacity;
infrastructure repairs that reduce cascade potential.

A quiet month should not cause Nankai or Cascadia to disappear from the structural ledger.

A new scientific model can, however, legitimately change the representation.

September 2026 Seismic Ledger

System	September state	Native forecast language
Nankai Trough	WATCH	official long-term probability / highest rank
Eastern Taiwan	WATCH	official 10/30/50-year potential maps
Cascadia	WATCH	10–15% M~9 in 50 years
Puget Sound	WATCH	85% deep M6.5+ / 17% crustal M6.5+ in 50 years
San Francisco Bay	WATCH	national probabilistic shaking-hazard model
Greater Los Angeles	WATCH	national probabilistic shaking-hazard model
Alpine Fault	WATCH	75% in 50 years; next rupture M8+ four-in-five
Hikurangi–Wellington	WATCH	structural national hazard watch
Sunda Strait	WATCH	seismic-gap preparedness signal, not warning
Mentawai–Siberut	WATCH	seismic-gap preparedness signal, not warning
Metro Manila / West Valley Fault	WATCH	M7.2 planning scenario
Marmara / Istanbul	WATCH	structural seismic hazard
Central Himalaya / Kathmandu	WATCH	structural collision-zone hazard
Lima / central Peru coast	WATCH	structural subduction hazard
Central–southern Chile	WATCH	structural subduction hazard

The Forecasting Test for the Next Ten Years

The Purple Report does not win if one of these fifteen systems produces an earthquake.

These are already known hazardous systems.

The harder test is whether the project:

preserves the correct probability horizon;
does not sensationalise quiet periods;
captures changes in vulnerability and preparedness;
recognises secondary cascades;
does not miss major damaging earthquakes outside the baseline;
learns from every false assumption about where the worst consequence occurred.

For earthquakes, the forecast is primarily a map of justified preparedness—not a clock counting down to rupture.

Earthquake and Tsunami Forecast 2026–2036 | The Purple Report Global Seismic Hotspot Outlook