An earthquake occurred on July 17, 2006, at 15:19:27 local time along a subduction zone off the coast of west and central Java, a large and densely populated island in the Indonesian archipelago. The shock had a moment magnitude of 7.7 and a maximum perceived intensity of IV (Light) in Jakarta, the capital and largest city of Indonesia. There were no direct effects of the earthquake's shaking due to its low intensity, and the large loss of life from the event was due to the resulting tsunami, which inundated a 300 km (190 mi) portion of the Java coast that had been unaffected by the earlier 2004 Indian Ocean earthquake and tsunami that was off the coast of Sumatra. The July 2006 earthquake was also centered in the Indian Ocean, 180 kilometers (110 mi) from the coast of Java, and had a duration of more than three minutes.
An abnormally slow rupture at the Sunda Trench and a tsunami that was unusually strong relative to the size of the earthquake were both factors that led to it being categorized as a tsunami earthquake. Several thousand kilometers to the southeast, surges of several meters were observed in northwestern Australia, but in Java the tsunami runups (height above normal sea level) were typically 5–7 meters (16–23 ft) and resulted in the deaths of more than 600 people. Other factors may have contributed to exceptionally high peak runups of 10–21 m (33–69 ft) on the small and mostly uninhabited island of Nusa Kambangan, just to the east of the resort town of Pangandaran, where damage was heavy and a large loss of life occurred. Since the shock was felt with only moderate intensity well inland, and even less so at the shore, the surge arrived with little or no warning. Other factors contributed to the tsunami being largely undetected until it was too late, and although a tsunami watch was posted by an American tsunami warning center and a Japanese meteorological center, no information was delivered to people at the coast.
The island of Java is the most densely populated island on Earth, and is vulnerable to both large earthquakes and volcanic eruptions, due to its location near the Sunda Trench, a convergent plate boundary where the Australian tectonic plate is subducting beneath Indonesia. Three great earthquakes occurred in the span of three years to the northwest on the Sumatra portion of the trench. The 2004 M9.15 Sumatra–Andaman, the 2005 M8.7 Nias–Simeulue, and the 2007 M8.4 Mentawai earthquakes produced the largest release of elastic strain energy since the 1957/1964 series of shocks on the Aleutian/Alaska Trench.
The southeastern (Java) portion of the Sunda Trench extends from the Sunda Strait in the west to Bali Basin in the east. The convergence of relatively old oceanic crust is occurring at a rate of 6 centimeters (2.4 in) per year in the west portion and 4.9 cm (1.9 in) per year in the east, and the dip of the Benioff Zone (the angle of the zone of seismicity that defines the down-going slab at a convergent boundary) is around 50° and extends to a depth of approximately 600 kilometers (370 mi). Pre-instrumental events were the large to very large events of 1840, 1867, and 1875, but unlike the northwestern Sumatra segment, no megathrust earthquake has occurred on the Java segment of the Sunda Trench in the last 300 years.
The earthquake was the result of thrust faulting at the Sunda Trench. A rupture length of approximately 200 kilometers (120 mi) (and an unusually low rupture velocity of 1–1.5 km (0.62–0.93 mi) per second) resulted in a duration of about 185 seconds (just over three minutes) for the event. The shock was centered 50 kilometers (31 mi) from the trench, and about 180 kilometers (110 mi) from the south coast of the island. A comparison was made with the earlier 2002 Sumatra earthquake, a M7.5 submarine earthquake of a similar size that also occurred along the Sunda Arc and at a shallow depth, but one that did not result in a tsunami.
The large and damaging tsunami that was generated was out of proportion relative to the size of the event, based on its short-period body wave magnitude. The Indonesian Meteorological, Climatological, and Geophysical Agency assigned a magnitude of 6.8, and the United States Geological Survey (USGS) reported a similar value of 6.1 (both body wave magnitude) that were calculated from short-period seismic waves (1–2 seconds in the case of the USGS). The USGS then presented a moment magnitude of 7.2 that was calculated from 5–100-second surface waves, and Harvard University subsequently revealed that a moment magnitude of 7.7 had been resolved based on even longer 150-second surface waves.
In tsunami prone regions, strong earthquakes serve as familiar warnings, and this is especially true for earthquakes in Indonesia. Previous estimates of the tsunami hazard for the Java coastline may have minimized the risk to the area, and to the northwest along the Sumatran coast, the risk is substantially higher for tsunami, especially near Padang. Previous events along the coast of Java in 1921 and again in 1994 illustrate the need for an accurate assessment of the threat. The July 2006 earthquake had an unusually slow rupture velocity which resulted in minor shaking on land for around three minutes, but the intensity was very light relative to the size of the tsunami that followed.
The earthquake produced shaking at Pangandaran (where the M6.3 2006 Yogyakarta earthquake was felt more strongly) of intensity III–IV (Weak–Light), intensity III at Cianjur, and II (Weak) at Yogyakarta. Further inland and farther from the epicenter, intensity IV shaking made tall buildings sway in Jakarta, but at some coastal villages where many of the casualties occurred, the shaking was not felt as strong. An informal survey of 67 people that were present at the time revealed that in at least eight cases, individuals stated that they did not feel the earthquake at all (a typical M7.7 earthquake would have been distinctly noticed at those distances). The unusually low felt intensities, along with the short period body wave magnitudes, were components of the event that narrowed its classification into that of a tsunami earthquake.
Tsunami earthquakes can be influenced by both the presence of (and lack of) sediment at the subduction zone, and can be categorized as either aftershocks of megathrust earthquakes, like the M7 June 22, 1932 Cuyutlán event in Mexico, or as standalone events that occur near the upper portion of a plate interface. Northwestern University professor Emile Okal imparts that in the aftershock scenario, they can occur as a result of stress transfer from a mainshock to an accretionary wedge or a similar environment with "deficient mechanical properties", and as standalone events they can occur in the presence of irregular contacts at the plate interface in a zone that lacks sediment.
One of the initial characterizations of tsunami earthquakes came from seismologist Hiroo Kanamori in the early 1970s, and additional clarity materialized following the 1992 Nicaragua earthquake and tsunami, which was evaluated to have a surface-wave magnitude of 7.0 when analyzing short period seismic signals. When longer period signals of around 250 seconds were investigated, the shock was reevaluated to have a moment magnitude of 7.6, with a hypothesis that the slow nature of the slip of the event may have concealed its substantial extent. Sediment was thought to have contributed to a slower rupture, due to a lubrication effect at the plate interface, with the result being an earthquake signature that had abundant long period seismic signals, which could be an important factor in the tsunami-generation process.
A tsunami warning system was not in operation at the time of the shock, but the Pacific Tsunami Warning Center (operated by the National Oceanic and Atmospheric Administration in Hawaii) and the Japan Meteorological Agency posted a tsunami watch, based on the occurrence of a M7.2 earthquake. The bulletin came within 30 minutes of the shock, but there was no means to transmit the warning to the people on the coast that needed to know. Many of those who felt the earthquake responded by moving away from the shore, but not with any urgency. The withdrawal of the sea that exposed an additional 5–10 meters (16–33 ft) of beach created an even more significant warning sign, but in some locations wind waves on the sea effectively concealed the withdrawal that signalled the approach of the tsunami.