INFP catalogue · 2012 – October 2026 · Romania
The ground shook hundreds of times in a single month.
Fifteen episodes since 2012 in which the ground moved dozens or hundreds of times in a month — six followed a big shock, nine arrived with no shock to set them off. This is the complete list, county by county, town by town.
- Episodes recorded
- 15
- Quakes counted
- 5,464
- Counties affected
- 8
- Largest shock
- M 5.7 Vrancea, Nov 2014
The map
Every episode, on the map
Each circle marks a 15 km area (25 km for Gorj 2023) where the catalogue recorded an episode. Amber circles are swarms — no shock to set them off; blue circles followed a big quake.
Circle size = number of quakes in the episode. Select with the mouse, Enter or Tab; ← → keys move the selection.
Pick a point on the map
or search for your town below. Each point shows when the episode happened, how many quakes it had and how big the largest was.
—
—
A marker does not mean the area is dangerous today. It means that, in the window shown, the catalogue recorded an unusual number of quakes there.
Type a town name and pick from the list.
Why it matters
Two kinds of episodes, two different stories
After a big shock
First the shock, then the wave
When a big quake opens an episode, the episode fades on its own, following a known law: many aftershocks at first, then fewer and fewer. In Gorj, after two February 2023 shocks, the catalogue counted 1,303 quakes in February alone; by 2025, at night, only three above M 2 remained in a year.
Swarm
No shock to set it off
A swarm has no opening shock: dozens or hundreds of small quakes arrive in waves, with no "main" one. At Schela in 2013, the peak came only in October, after the second M 3.9. This is what the ground looks like when faults creep, not when they break.
What it means for you, if you live in one of these areas: nothing special, day to day. Quakes below magnitude 3 are not felt unless you are very close to them. A swarm does not announce a big earthquake — in Gorj the two big shocks came at the start, not at the end. And at Galați the swarm came back twice in twelve years without anything worse following.
What this list cannot tell you: whether a place is "safe" or "dangerous". It says only what the catalogue recorded, when and where. What you feel from a big earthquake depends on your distance from the Vrancea nest and on the ground under your feet — not on this list.
The Schela story
The question everyone in Galați asks
In the autumn of 2013, a few kilometres north of Galați, the ground began to move. The first quake in the catalogue dates from 15 August; INFP studied it between 23 September and 12 November and counted 406 located quakes, the largest, magnitude 3.9, on 29 September. Many were felt strongly in Izvoarele, Schela and Slobozia Conachi — intensity V–VI — and cracked walls.
Two hundred metres below the epicentres lies the Independența–Schela oil field. The question was inevitable: was it the wells? The National Institute for Earth Physics published a full study in 2017: the swarm occurred on a normal fault with a strike-slip component, between two major crustal fault systems — the New Trotuș system to the north-east and Peceneaga–Camena to the south — in an area studied seismically for the first time (Craiu et al., Acta Geodaetica et Geophysica 52, 53–67, 2017).
The Romanian Academy's Institute of Geodynamics went further: most of the events occurred well below the depth of the oil exploitations, in a graben-like structure transverse to the northern slope of the North Dobrogea Promontory, probably (re)activated by stresses in the Carpathian foreland (Beșuțiu, Diaconescu, Zlăgnean & Craiu, Pure and Applied Geophysics 176, 65–95, 2019).
Nobody claims today that the extraction caused the swarm. The overlap on the map remains, and asking about it is fair. What we do know: the mechanisms are normal-faulting, the depth is below the level of exploitation, and the swarm came back — two packets in August–September 2024 and December 2024–January 2025, the same ground, eleven years later.
The full list
All fifteen episodes
| Episode | Window | Kind | Quakes | Max M | Peak month |
|---|
"Quakes" = distinct events at depth < 45 km, in the episode months, within 15 km of the centre (25 km for Gorj). For Gorj the total from the mainshocks through 13 October 2026 is 4,580. Episodes marked "across the border" are in Ukraine and Serbia; they are shown because we felt them too.
Method and sources
How the list was made
The catalogue
- Source: the INFP FDSN service (doi:10.7914/SN/RO), downloaded on 8–9 November 2026, 38,662 records, 2004–13 October 2026. The page publishes derived counts and maps, not the raw catalogue.
- Cleaning: 80 pairs of duplicate solutions removed; the 6 February 2023 "Vrancea earthquakes" at 01:17–04:00 UTC were the Turkish quake's waves misread as local events — removed; 2005–09 automatic solutions with no USGS counterpart — removed.
- Real rates are computed only from 2012 on: 2005–09 uses a different magnitude scale and 2010–11 is nearly empty. The two scales differ visibly, from 1.5–2.0 (2005–09) to 2.5–2.75 (2012+).
- 2012 through September 2026 for complete monthly series; October 2026 appears only in the Gorj total (through 13 Oct).
How each episode was found
- A swarm = a 0.2° cell with ≥ 8 quakes in a month (night only, 21:00–05:00, to exclude quarry blasts), where the median of the other months is ≤ 0.5. Twenty-six cell-intervals came out; those sharing ground and window were merged.
- Adding day hours outside the blast cells reproduces the same places (a sensitivity rerun, in a separate script).
- Classification: if the first shock fell within the first tenth of events and exceeded the second by ≥ 0.3, the episode is "after a big shock"; otherwise it is a "swarm".
- Places are taken from the 2025 UAT outlines, not from memory. Episode centres are medians of the epicentres, not survey points.
What not to read into it
- This is not a seismic-risk map. Nothing here makes an area "safer" or "more dangerous" to live in.
- Depths below 45 km are not reported (values of 2–10 km are often implicit); no geology is inferred from default numbers.
- Small magnitudes depend on distance from stations: since 2022, the temporary AdriaArray stations have heard the Gorj and Bor areas better — part of the rise in small numbers there is new hearing, not necessarily more motion.
- The label "swarm" describes the shape of the sequence, not its cause.
Checks performed
- The smallest episode in the list (Nereju, August 2020, 15 events) was printed row by row from the catalogue and read by hand: 13 of the 15 came within 48 hours on 28–29 August, opening with an M 1.8 at 20:10 and peaking with an M 3.6 at 21:21.
- 60 points from the locality index were checked to lie inside their UAT polygons: 60 of 60 correct.
- The 15 episode centres were verified against the UAT polygons: 13 inside the named UATs, 2 on the border (marked as such).
Sources
- INFP — FDSN service, event catalogue (citation: National Institute for Earth Physics, Romanian Seismic Network, doi:10.7914/SN/RO). Published data here are derived counts and maps.
- Craiu A., Craiu M., Diaconescu M., Mărmureanu A. (2017). 2013 Seismic swarm recorded in Galati area, Romania: focal mechanism solutions. Acta Geodaetica et Geophysica 52:53–67. DOI:10.1007/s40328-016-0161-9.
- Beșuțiu L., Diaconescu M., Zlăgnean L., Craiu A. (2019). Structural and Geodynamic Ideas on the Galati-Izvoarele Seismic-Prone Area (Eastern Romania). Pure and Applied Geophysics 176:65–95. DOI:10.1007/s00024-018-1956-0.
- UAT outlines 2025: geoBoundaries ROU ADM1, World Bank / ANCPI, CC BY 4.0; names: SIRUTA, semester 1 2026 (INS, via data.gov.ro).
- Fonts: Fraunces, Newsreader, Plex Mono (SIL Open Font License).