The Haicheng Earthquake Prediction

Summary

On 4 February 1975 a magnitude 7.3 earthquake wrecked Haicheng, in north-east China, hours after local officials had begun moving people out of their homes. It was hailed as the first major earthquake ever predicted, and a 2006 study of declassified documents found no official short-term prediction at all.

What the 2006 study did find was a warning driven by an unusually pronounced burst of small earthquakes, acted on by county and provincial officials, and an evacuation that was uneven. Less than eighteen months later the Tangshan earthquake killed more than 240,000 people. The United States Geological Survey now says that no scientists "have ever predicted a major earthquake."

What is documented

The background. Earthquake prediction in China expanded rapidly from 1966, during the political and social turmoil of the Cultural Revolution (Chen and Wang, 2010). Kelin Wang and colleagues (2006), who studied declassified Chinese documents and interviewed key witnesses, found that the Haicheng earthquake had been the subject of two official middle-term predictions.

The foreshocks. An "unusually pronounced foreshock sequence" preceded the earthquake (Chen and Wang, 2010). Lucile Jones and colleagues (1982), locating the foreshocks from six local stations, found them in a tight cluster about 2 kilometres across, which after the largest foreshock spread into a zone 6 kilometres long. The mainshock began 6 kilometres south of the cluster and several kilometres shallower. By their analysis, the increase in shear stress caused by the foreshocks was very small. According to the USGS, "the increase in foreshock activity triggered the evacuation warning."

Other reported anomalies. Wang and colleagues found that geodetic deformation, changes in the level, colour and chemistry of groundwater, and peculiar animal behaviour "also played a role", behind the foreshocks. The USGS lists changes in land elevation and groundwater levels and widespread reports of peculiar animal behaviour.

The warning. Wang and colleagues found "no official short-term prediction." On the day of the earthquake a county government issued a specific evacuation order, and the actions of provincial scientists and officials "effectively constituted an imminent prediction." The evacuation was "extremely uneven across the disaster region", and critical decisions were often made at very local levels. A delegation of visiting scientists, reporting in Eos in 1977, placed the evacuation on the afternoon of 4 February.

The earthquake. It struck at 19:36 local time (11:36 Universal Time). Contemporary accounts give magnitude 7.3 (Adams, 1976; Eos, 1977); the USGS lists it as magnitude 7.0. About 90 per cent of Haicheng, a city of 90,000 people, was destroyed or seriously damaged (Eos, 1977).

The toll. The figures conflict. NOAA's significant-earthquake database records 300 deaths; the USGS says the earthquake "caused more than 2,000 fatalities", and a USGS page on earthquake prediction gives the toll exactly: "2,041 people died, 27,538 were injured." The Eos report estimated that "in excess of 100,000 casualties" would ordinarily have been expected, and the USGS that deaths and injuries "would have exceeded 150,000" without prediction and evacuation. Wang and colleagues concluded that the warnings "saved thousands of lives", but that local construction style and time also kept the death toll down.

The reception. Robin Adams, who visited the area about seven and a half months after the earthquake, called it "the first major earthquake anywhere in the world known to have been predicted with enough certainty for people to have been warned" (1976). The Eos delegation called the prediction "an extraordinary achievement". In the words of Wang and colleagues, the publicised four-stage prediction "once generated worldwide fascination."

Tangshan. At 03:42 local time on 28 July 1976 a magnitude 7.5 earthquake struck Tangshan, followed by a major aftershock 15 hours later. The USGS gives 242,769 deaths from the two, notes estimates as high as 655,000, and counts over 799,000 injured. Encyclopaedia Britannica gives the official toll as 242,000. A USGS note on the Chinese forecast adds: "The next large Chinese event was unheralded and many thousands of Chinese died." Chen and Wang (2010) date China's "diminishing confidence" in prediction from the catastrophic Tangshan earthquake.

Since. Before the magnitude 7.9 Wenchuan earthquake of 12 May 2008, despite massive precursor monitoring networks and scheduled prediction conferences, no anomalous pattern that would have enabled a prediction was identified (Chen and Wang, 2010). The international commission on earthquake forecasting set up after the L'Aquila earthquake of 2009 wrote that similar schemes "have not led to other comparable successes" (Jordan and colleagues, 2011).

Leading explanations

An unusual foreshock sequence, acted on by local officials. The leading account. Wang and colleagues found the foreshocks the most important precursor, and Chen and Wang attribute the prediction mainly to the foreshocks and to the "extraordinary readiness of some local officials at the time to issue an imminent warning." The account rests on declassified documents and witnesses, and it explains how the warning came about.

Other precursors (reported; a supporting role; none has proved reliable). Groundwater, deformation and animal behaviour were part of the picture, by Wang and colleagues' reconstruction. On animals the USGS is plain: consistent and reliable behaviour before earthquakes, "and a mechanism explaining how it could work, still eludes us."

Circumstance (a contributing factor). Wang and colleagues credit local construction style and the time at which the earthquake struck with part of the low death toll.

Why this earthquake had such foreshocks (not established). Most do not. The USGS notes that such activity "is rarely followed by a large earthquake" and that most earthquakes "have no precursory events whatsoever." Robert Geller (1997) argued that faulting is so sensitive to fine details of the state of the Earth that any small earthquake has some probability of cascading into a large one, and that reliable alarms of imminent large earthquakes appear to be effectively impossible. In 1977 Christopher Scholz proposed that a deformation front travelling about 1,000 kilometres across north-east China at about 110 kilometres a year triggered the earthquake. It is a hypothesis. In 2026 Zhang and colleagues, imaging the crust beneath the source region, interpreted the upward migration of crustal fluids as "a key deep controlling factor" for the frequent foreshocks and the groundwater anomalies. That is also a hypothesis.

Current status

Partially explained. How the warning came about is now documented: an unusually pronounced foreshock sequence, and officials at county and provincial level ready to act on it, with no official short-term prediction. What is not explained is why this earthquake gave such a warning when most give none, and whether the other reported anomalies carried any information. No method built on Haicheng has repeated it: Tangshan came less than eighteen months later, and before Wenchuan in 2008 no usable anomaly was identified. The case would change with a precursor that succeeds in repeated tests made before the earthquakes it predicts.

Last reviewed: September 2026. Records are provisional. Where the evidence changes, the entry changes. Found an error? Tell us.