# The 1977 earthquake, over today's red-dot buildings

How the wave from Vrancea crossed Bucharest, what was recorded on each square kilometre, and which buildings on the red-dot list stand there today.

2.4 seconds. That is how long the wave took to pass over the 2,532 buildings on the list that we could place on the map; the wave reached nine in ten within a total interval of less than a second. Bucharest's ground sways once every 1.4–1.6 seconds, so the whole city was in motion in less than two swings.

The order in which the wave arrives does not say which building suffers. In 1977 place and building type mattered: a survey of about 18,000 buildings found very different damage from one square kilometre to the next and from low houses to tall blocks. The map overlays the measurements from the 1982 book on the positions of today's buildings.

The dot is the class an engineer gives a building after an assessment, and the 1977 record is in squares of one kilometre. The page shows what happened around a building. What would happen to it is a question for an engineer.

## The quake, second by second, over today's city

Press Play or drag the slider. The columns show the damage measured in 1977 in each square kilometre, for buildings with the period chosen below. The boxes are the buildings on the list: grey until the wave reaches them, then in the colour of their class.

## Find your building or street

Type a street. The map moves there and shows the buildings on the list. For a building, the card gives the time the wave arrives, its probable period and what engineers measured in 1977 for buildings of similar periods in the same square kilometre. It does not say what would happen to it.

## How long the wave takes to arrive

The 4 March 1977 focus lay about 94 kilometres below Vrancea, roughly 183 kilometres from central Bucharest in a straight line. The P wave, faster and weaker, reaches the city about 23.7 seconds after the rupture begins. The S wave, the one that shakes, arrives at about 41.5. The gap between them, 17.8 seconds in the calculation (between 17.1 and 18.4, depending on the ratio of the speeds), matches the 18 seconds of weak, almost purely vertical motion at the start of the record made at INCERC (Bălan and others, 1982). The speeds used, 4.4 and 7.7 km/s, are ordinary upper-mantle values, not measured for Vrancea; the match shows only that they are reasonable.

The city is small next to the distance to the focus. The S front passes over the 2,532 buildings in 2.4 seconds, and it reaches nine in ten within a total interval of 0.9 seconds. The 'Slow motion' button stretches these seconds so you can see the order. The order is real, but shorter than one sway of the ground, which in 1977 had a period of 1.4–1.6 seconds.

What follows lasts much longer: the strong phase of the record held 10–15 seconds of a minute or more, and long, low-amplitude waves appear after 30–40 seconds. The earthquake was a multiple event, with the focus migrating south-west. The calculation treats it as one source, at the USGS location (19:21:54 UTC, that is 21:21:54 local time), so absolute seconds may differ by a few.

## What was measured in 1977, by square kilometre

Between 4 and 28 April 1977, more than 200 students of the Institute of Construction and the Institute of Architecture, led by the city's seismic microzonation commission, recorded the damage to about 18,000 Bucharest buildings. The city was divided into squares of about one square kilometre; in 66 of them about 300 buildings each were surveyed. Each got a damage degree from 0 (none) to 5 (collapse) and was put in a band of its own period. The result is an average per square, per band and per category: masonry with flexible floors, masonry with rigid floors, reinforced-concrete frames.

The authors' conclusion deserves reading before the colours. The same survey gives different MSK intensities in the same square, depending on the buildings' period, and flexible concrete-frame buildings were more damaged than rigid masonry ones, the reverse of what the scale assumes. So there is no single intensity per street; there is one value per period band. Bulevardul Magheru, judged the worst hit right after the quake, shows small damage for low buildings and very large damage for those with a period over one second.

Squares without a column were not surveyed or have no buildings in the chosen band. We did not fill the gaps and do not interpolate between squares: Bălan shows that surface layers do not explain the damage and that the real distribution depends on Bucharest's deep geology, then little known.

## What the USGS model says

USGS reconstructed the 1977 earthquake as a ShakeMap, from 747 intensity observations, three stations and a source extended along a fault. Over Bucharest it gives MMI intensity between 7.2 and 8.0 and peak acceleration between 0.21 and 0.27 g; at INCERC 2.1–2.2 m/s² was measured, about 0.22 g. The model is smoothed to about 1.3 by 1.9 kilometres and lacks the local effects the damage depends on.

In the squares where buildings fell, the model's intensity averages 7.89; in the others, 7.84. It does not tell them apart. The 'USGS model' layer on the map lets you compare.

## Where the blocks fell, against the record

We put the 25 buildings that fell, wholly or in part, in 1977 (the INCDFP layer used on the earthquake page) on the squares and asked whether they sit in the squares with the highest damage for tall buildings. Tall buildings have periods above 0.7 seconds; for them we have only the reinforced-concrete column of table V.1, in 16 squares, all in the centre.

In 5 of these 16 squares (numbers 21, 31, 34, 51, 52), the mean degree for periods between 0.7 and 1.3 seconds is 4 or 5: on average, collapsed or nearly collapsed buildings. Of the 25 fallen buildings, 9 sit in one of them. With a 350-metre margin for the position of the squares, it is 18.

The right baseline is the 16 squares that have any measurement for long periods. The share of squares with a degree of 4–5 is 31% there, so chance would give about 7.8 of the 25 buildings. The count is 9. Against all 67 squares, where the share is 7% and chance would give about 1.9, the result would look much better, but most squares have no long-period measurement at all. The record shows that the centre was where the heavy damage was; it does not tell central squares apart.

Where it fails. Of the fallen buildings, 4 sit in squares with no bar at all for long periods. Squares 31, 34, 51 have a degree of 4–5 and contain none of the 25; two of them, 31 and 51, lie by our placement near the zones Bălan names as having maximum intensity, Cotroceni and north-east of Piața Unirii. Block OD16, in Cotroceni, falls in a square with a single value, for low buildings. And for masonry buildings and for periods under 0.7 seconds the record does not separate the squares with fallen buildings at all: in the 0.50–0.70 s band they average 1.2, the others 1.67.

The likely explanation: the averages are over about 300 buildings chosen along streets, while the tall blocks that fell were a few dozen, rare and at corners, so poorly represented. A square can show a degree of 5 from a single building.

## What the page cannot say

- Whether a particular building would stand. The dot is an administrative class given by an engineer; RsI shows that in the design earthquake the building is liable to collapse; what happens to a particular building is settled by an assessment.
- What the ground was like, street by street. The record is in squares of about a kilometre, placed on today's map with an error of about 300 metres, and has nothing between them.
- What a building is made of. The period is estimated from the storey count with the European code's formula; masonry or concrete is unknown.
- Buildings that are not on the list. The list has 2,796 rows; a World Bank estimate quoted by AMCCRS speaks of about 20,000 vulnerable buildings. Absence from the list means only that no assessment is on record.
- What was strengthened, and how. 'Strengthened' rows appear on the map, but the list does not say what work was done.
- The 1940 earthquake. It has no record by square; the macroseismic maps of the time cover the whole country. That is why the page has 1977 only.
- The next earthquake. It will not be 1977: another epicentre, another duration, another city.
- An assessment. For a particular building, ask the administrator or AMCCRS for the assessment report.

## What the classes mean

A class is set only by a technical assessment from a certified engineer. Law 212/2022 defines them by what would happen to the building in the design earthquake, the strong quake new buildings are calculated for.

- **RsI**: May collapse, wholly or in part. City Hall puts a notice panel at the entrances of a residential building in this class: that is the 'red dot'.
- **RsII**: Major damage that endangers people, but collapse is unlikely.
- **RsIII**: Moderate damage that may endanger people.
- **RsIV**: Behaves like a building designed to today's codes.
- **U1, U2, U3**: 1990s codes under repealed rules: strengthening deadlines of 2, 5 or 10 years. City Hall says they cannot be converted automatically into RsI–RsIV.

The list has 415 RsI buildings, 387 of them on this map.

## Method, in short

Recorded, calculated, illustrative. Recorded are the damage degrees in map V.25 a–e and in table V.1 of Bălan's book (the reinforced-concrete-frame column), entered twice, independently: the two readings matched in 139 of 140 boxes, and the table column matched the map in 51 of 56 comparable values. Calculated are the arrival seconds, each building's period and the position of the squares on today's map. Illustrative are the curve by which the columns grow in time, their height and the size of the boxes.

The squares. Figure V.23 of the book has the grid of squares on a 1977 plan of the city. We matched the plan's landmarks (the main roads) to OpenStreetMap roads; the fit has an error of about 300 metres. Each square comes out at about 1.05 by 0.98 kilometres. The numbers in map V.25 attach to squares through the numbers in V.23.

Building period: T = Ct · H^0.75, with H of 3.0–3.5 m per storey and Ct between 0.05 (masonry, other structures) and 0.075 (reinforced-concrete frames), after EN 1998-1. The range gives one or two bands; the card shows them all.

Wave arrival: a point source at the USGS location, speeds of 4.4 km/s (S) and 7.7 km/s (P). The columns grow in time along an illustrative curve that peaks 12 seconds after the S wave, the length of the strong phase in the INCERC record.

Buildings: rows of the AMCCRS list of 19 May 2026 with a point on the exact address, from OpenStreetMap or the DupăCutremur export (CC BY 4.0): 2,532 of 2,796. The page 'Bucharest's red-dot buildings, one by one' also places 74 buildings by the middle of a number range; we did not use them here because each building gets a square attached.

Licences: the scan of Bălan's book is CC BY 4.0; OpenStreetMap is ODbL; DupăCutremur is CC BY 4.0 (Re:Rise, geo-spatial.org); the USGS data are published by the US Geological Survey; the list is published by AMCCRS.

## Sources

- Bălan, Cristescu, Cornea (1982), the 4 March 1977 earthquake, Editura Academiei: ch. V (the 18,000-building survey, fig. V.23, V.25 a–e, table V.1), pp. 131, 147, 238–240. Scan CC BY 4.0 https://zenodo.org/records/20610342
- USGS ComCat, event usp0000mtc: 4 Mar 1977, 19:21:54 UTC, 45.772° N 26.761° E, depth 94 km, Mw 7.5 https://earthquake.usgs.gov/earthquakes/eventpage/usp0000mtc
- USGS ShakeMap Atlas v4 for the same event (747 intensity observations, 3 stations; source Radulian & Popa 1996) https://earthquake.usgs.gov/earthquakes/eventpage/usp0000mtc/shake-map
- ISC Bulletin, event 700695 (alternative location: 45.879° N 26.655° E, 85 km) https://www.isc.ac.uk/
- EN 1998-1:2004, §4.3.3.2.2: approximate fundamental period T = Ct·H^0.75 (Ct 0.05 and 0.075) https://www.cen.eu/
- Lungu, Arion (2012), Seismic risk and/or real estate risk for the Bucharest heritage buildings, 15 WCEE (Berg et al. 1980 and the 1940 earthquake, quoted there) https://www.iitk.ac.in/nicee/wcee/article/WCEE2012_1996.pdf
- AMCCRS (Bucharest City Hall), list of buildings with a technical expertise and a seismic risk class, table updated 19 May 2026 https://amccrs-pmb.ro/lista-imobile-2/
- OpenStreetMap: addresses, roads and administrative boundaries, downloaded 28–29 Sep 2026 (ODbL) https://www.openstreetmap.org/copyright
- DupăCutremur (Re:Rise and geo-spatial.org), points for the AMCCRS list buildings, export of 20 Jul 2026, CC BY 4.0 https://www.dupacutremur.ro/
- The buildings that fell in 1977: INCDFP's layer of buildings affected by earthquakes, as used in the 1977 earthquake page https://mariuscomper.uk/cutremurul-din-1977/#pe-loc
- Law 212/2022 on measures to reduce the seismic risk of buildings (arts. 1, 3) https://legislatie.just.ro/Public/DetaliiDocument/257526
- AMCCRS, clarifications on myths about seismic risk classes and technical expertises, 4 Mar 2025 https://amccrs-pmb.ro/2025/03/clarificari-esentiale-despre-miturile-privind-incadrarile-in-clase-de-risc-seismic-si-expertizele-tehnice/
- 'Bucharest's red-dot buildings, one by one': the same list, on each building's outline https://mariuscomper.uk/cladiri-cu-risc-seismic/
- 'The 4 March 1977 earthquake': the fallen blocks, in 3D https://mariuscomper.uk/cutremurul-din-1977/
- 'How strong?': seismic hazard by locality (Order MDLPA 1037/2026) https://mariuscomper.uk/cat-de-tare/
- INCDFP, 'The seismic risk of Bucharest': a 2D/3D app with hazard, historic damage and emergency routes https://prequake.infp.ro/a-new-webgis-app-the-seismic-risk-of-bucharest/