BhuKompo — Bangladesh earthquake risk, in plain language

01 Context · why people are scared right now

What feels unusual. Since late 2025, Bangladesh has felt 4 noticeable quakes in quick succession: M5.4 near Dhaka (Nov 2025), M4.3 Paltan, M4.1 Dhaka, M5.0 Satkhira (Feb 2026). Before that, years could pass without a widely felt event. The clustering feels alarming — but statistically it's within normal variation for an active plate boundary.

Clustering is normal, not a sign of the Big One. Earthquakes don't tick like a metronome — they cluster. Three in six months, then none for three years. It's the same reason you can flip heads four times in a row: it doesn't mean the coin is broken. The long-term rate hasn't changed.

02 Data frame · probabilistic scenarios

Based on USGS ComCat (2024–2026), GSHAP/GEM PSHA models, Gutenberg-Richter statistics (b≈0.98) and known fault geometries. Return period = average time between events of this size. Annual probability ≈ 1 / return period.

Scenario	Magnitude	Source zone	Return period	Annual prob.	50-yr prob.	Max PGA near epicentre	MMI in Dhaka	Key impact
Frequent crustal	M4.0–4.5	Madhupur / Sylhet / local	~1–2 yr	50–100%	~100%	0.05–0.15g	II–IV (felt, minor)	Widely felt, minor cracks, panic
Moderate damaging	M5.0–5.5	Madhupur, Dauki	~10–15 yr	7–10%	30–40%	0.15–0.30g	V–VI (moderate)	Non-structural damage, old buildings at risk
Major city-affecting	M6.0–6.5	Madhupur (Dhaka proximal)	~50–80 yr	1.3–2%	5–8%	0.30–0.50g	VII–VIII (heavy)	Structural damage, casualties possible
Great subduction	M7.5–8.0+	Chittagong–Myanmar Arc	~200–500 yr	0.2–0.5%	1–2.5%	0.25–0.45g coastal	VI–VII (far-field)	Widespread damage, tsunami risk
Max credible	M8.5+	Full Sunda megathrust	~800–1500 yr	<0.1%	<0.6%	0.30–0.50g	VII (far-field)	Catastrophic regional event

How to read this without a statistics degree. "50-yr probability" means: live in Bangladesh for 50 years and this is the chance you experience at least one such quake. M5–5.5 (~30–40%) = likely in a lifetime. M6.5+ (~5–8%) = a roughly 1-in-15 chance. M8+ (<1%) = very unlikely, massive consequences. These are long-term averages — the fault doesn't "know" it's been 60 years. It could happen tomorrow, or not for 200 years.

03 Data frame · city-level shaking hazard (PSHA)

City	Population	Nearest major fault	Distance	PGA 10%/50yr	PGA 2%/50yr	Design target	Risk level
Dhaka	~22M	Madhupur Fault	~20 km	0.18g	0.35g	0.18g (BNBC 2020)	Elevated
Chittagong	~5M	Ctg–Myanmar Arc	~60 km	0.22g	0.42g	0.22g	Elevated
Sylhet	~1M	Dauki / Sylhet	~15 km	0.28g	0.50g	0.28g	High
Mymensingh	~0.6M	Madhupur Fault	~10 km	0.25g	0.45g	0.25g	High
Rajshahi	~1M	Madhupur / Himalayan	~80 km	0.14g	0.28g	0.14g	Moderate
Khulna	~1.5M	Far-field	>150 km	0.12g	0.24g	0.12g	Moderate
Barisal	~0.5M	Far-field	>150 km	0.10g	0.20g	0.10g	Moderate
Rangpur	~0.8M	Himalayan (distant)	>200 km	0.09g	0.18g	0.09g	Moderate

What "0.35g" means for your building. g = gravity (9.8 m/s²), so 0.35g ≈ 3.4 m/s² of horizontal acceleration. Modern BNBC 2020 buildings are designed not to collapse at the 475-year shaking level — but "not collapse" ≠ "no damage": cracks and falling plaster start around 0.1–0.15g. Pre-1993 and informal construction has far less capacity.

04 Model · the science, for non-experts

Gutenberg-Richter law

b ≈ 0.98

"Choto choto bhukh beshi hoy, boro bhukh kom hoy." Every one-step rise in magnitude means ~10× fewer quakes. M4 ~10×/yr → M5 ~1×/yr → M6 every ~10 yr → M7 every ~100 yr. Our 2024–26 data lands at b≈0.98 — almost exactly the textbook value.

PSHA

HAZARD MATHS

"Shobar probability jog kore — kotota shaking ashte pare?" Map every fault, give each a rate, compute the shaking each possible quake sends to your spot, then add up all the probabilities. Result: "a 10% chance in 50 years that shaking here exceeds 0.18g."

Fault behaviour

LOCK & SNAP

Madhupur: locked, loading, last big one ~400+ yr ago, capable of M6.5–7.0. Dauki: M7+ potential, rarer but larger. Ctg–Myanmar Arc: subduction, M8+ possible, century-scale recurrence. Sylhet: active thrust, M6–6.5 typical.

Probability ≠ prediction. Prediction — "M6.5 hits Dhaka on 15 March 2027 at 2:30pm" — is impossible with current science. Probability — "~6% chance per year of M6+ somewhere in the region" — is what science gives us. Probability helps you prepare. Prediction would help you evacuate. We don't have the latter.

05 Model · three realistic scenarios

A · M5.5 Madhupur, near Dhaka

~8% / YR

"Likely in a lifetime." MMI VI–VII in Dhaka. Cracks in old buildings, fallen plaster, panic, injuries from non-structural falls. Recent analog: Nov 2025 Tungi M5.4 — 989 felt reports, 10 deaths, 629 injured.

B · M6.5 Madhupur, under Dhaka

~1.5% / YR

"The credible worst case for the city." MMI VIII–IX. Collapse of vulnerable structures (pre-1993, informal), heavy damage to mid-rise frames, potentially thousands of casualties depending on time of day. Historic analog: the 1885 Bengal earthquake (M7.0+).

C · M7.8 Ctg–Myanmar subduction

~0.3% / YR

"Regional catastrophe." MMI VIII+ in Chittagong, VI–VII in Dhaka with long shaking, tsunami risk on the Bay of Bengal coast. Historic analog: the 1762 Arakan earthquake (M8.8) — tsunami, widespread destruction.

The time-of-day factor is enormous. An M6.5 at 3am (people asleep in vulnerable buildings) versus 11am can mean a 10× difference in casualties. That's why building resilience beats timing luck — always.

06 Data · what the 2025–26 cluster does and doesn't mean

What the data shows. 8 quakes in Bangladesh since 2024 (M4.1–5.4) · 4 widely felt in 14 months · all shallow crustal (10–45 km) · consistent with known fault systems · GR statistics still hold (b≈0.98).

What it does NOT mean. Not "pressure released, big one cancelled." Not "the fault is awake, the big one is imminent." Not "an accelerating sequence." Not "aftershocks will keep growing." Only ~5% of quakes are foreshocks.

The "stress release" myth. A M5.4 releases about 1/30th the energy of an M6.5 and 1/1000th of an M7.5. The 2025–26 cluster released a negligible fraction of the Madhupur Fault's locked strain — a few small cracks in a dam don't stop the dam from breaking.

07 Action · what this means for decisions

Individuals & families

DO NOW

Secure heavy furniture, water heaters, shelves
Learn Drop, Cover, Hold On — practice with kids
Pick safe spots per room — sturdy table, away from windows
Emergency bag: water, meds, documents, cash, torch
Family meetup plan if separated
Ask whether your building follows BNBC 2020

Building owners

RETROFIT

Retrofit pre-1993 buildings — soft storeys, weak columns
Brace partitions, ceilings, facades
Soil investigation before construction — Dhaka alluvium liquefies
Peer-review structural designs for high-rises
Post-earthquake inspection protocol

Policy & planners

SYSTEMIC

Enforce BNBC 2020 strictly — no exceptions
Seismic microzonation for Dhaka, Chittagong, Sylhet
Retrofit schools, hospitals, fire stations first
Public early warning — even 10–20 seconds saves lives
Tsunami evacuation routes for the coast
Insurance incentives for retrofit

08 Reference · the probability cheat sheet

Question	Data-driven answer	Confidence
Another felt quake in Dhaka this year?	~60–70% chance of M4+ felt in the metro	Medium
A damaging quake (M5.5+) within 10 years?	~50–60% (M5.5–6.0 range)	Medium
Dhaka MMI VII+ within 30 years?	~15–25% — mainly a Madhupur M6.5+	Medium
Is the recent cluster a foreshock sequence?	No evidence — only ~5% of quakes are foreshocks	High
Can Dhaka get a tsunami?	No — too far inland. Chittagong / Cox's Bazar: yes	High
Are Dhaka high-rises safe?	Post-2008 BNBC compliant: generally yes. Pre-1993: vulnerable	Medium
Should I sleep outside / leave the city?	No. Panic evacuation causes more injuries than quakes	High
Is Bangladesh prepared?	Improving (BNBC 2020, DRR Act) — big gaps in retrofit, warning, awareness	Medium

Sources: USGS ComCat API (2024–2026) · GSHAP Global Seismic Hazard Map · GEM Foundation PSHA models · BNBC 2020 · BMD seismic catalog · Steckler et al. 2016 · Alam et al. 2021.
Models: Gutenberg-Richter (b=0.98, a=2.1) · PSHA (3 GMPEs, 4 source zones) · characteristic-earthquake model for Madhupur/Dauki.
Not a prediction. Long-term rates, not short-term forecasts. Use for preparedness, not panic.

Back to the console All BhuKompo instruments