Warning signals for Indian Ocean seaquake missed

Over a decade has passed since the 26 December 2004 Indian Ocean seaquake that triggered a devastating tsunami off the coast of south East India and Sri Lanka. Indian geologists, scrutinising whether they missed any warning signals, now say the plate near Andaman islands had started to creep downwards silently way before the earthquake – a vital clue that might have gone unnoticed¹.

C P Rajendran, professor at the geodynamics unit of the Jawarharlal Nehru Centre for Advanced Scientific Research (JNCASR), Bangalore says that the research community must admit to have failed to anticipate the event lurking along the eastern seaboard of India.

The earthquake that triggered the Asian tsunami led to several studies on slips and displacements in the Sumatra-Andaman plate during and after the earthquake. The quake started near Sumatra and travelled upwards towards Andaman, the studies found, and there was an additional slow slip that lasted 50 minutes.

To see if scientists might be any closer to forecasting possible quakes, Rajendran and John Paul from the University of Memphis, US, poured over previous data from the Andaman and Nicobar capital Port Blair, which was left tilted by the quake. Global Positioning Satellite System (GPS) and tide gauge data revealed that Port Blair had already started to subside between 2003 and 2004.

GPS data before the December 2004 earthquake showed a “possible precursory deformation signal prior to the 2004 Sumatra Andaman earthquake,” they report. The precursory signal includes both vertical subsidence and horizontal reorientation. This part of the fault line showed slow slipping both before and during the December 2004 event.

The scientists say that Port Blair was emerging during the ‘inter-seismic’ period between two quakes, but the area near port Blair started subsiding one to two years before the earthquake.

The site of data collection showed an uplift of 11.4 mm per year, between 2002 and 2003. But sometime between 2003 and 2004, the nearby area started subsiding at a rate of about 70 mm per year. Also, barely four months before the 2004 earthquake, the site subsided at a rate of 130 mm per year.

“The subsidence at Port Blair can be explained by slip of 1.5 m on the upper portion of the fault, spread across 6–12 months prior to the great earthquake,” the report says. Tide gauge measurements east of the GPS site at Port Blair also revealed pre-earthquake subsidence thus complimenting GPS results.

The inter-seismic GPS observations at Port Blair suggest that the region showed a long-term upward movement for some years before the earthquake, says Rajendran. Visual observations made 15 months prior to 2004 indicated signs of emergence of near-shore corals near Port Blair.

“Thus at Port Blair we witnessed a slow emergence coupled with a slow subsidence just before the earthquake followed up by a sudden subsidence during the earthquake. We have estimated this pre-earthquake slow downward slip as equivalent to an earthquake with a magnitude of 6.3. The magnitude of the slip may be higher towards the plate boundary located 200 km west of Port Blair.”

Rapid slip along a fault during an earthquake is usually on the top portion, which scientists refer to as ‘up-dip’. But the analysis shows that there was slip in the top portion for about a year before the great earthquake. Earlier scientists thought that slow slip occurred at far deeper at 35-kms but the Port Blair data shows it can occur at shallow depths too.

“The big question is whether this precursory signal observed at Port Blair has global significance in terms of its usefulness as a forewarning tool in major subduction zone earthquake zones, elsewhere,” the scientists say.

Geological indicators from the Alaskan coast also suggest a short-term subsidence prior to the 1964 quake, which makes the scientists ponder whether a slow slip is not “a localized phenomenon” but a “globally applicable signal, at least in some subduction zones.”

Scientists, however, say that while the theory is interesting, data from other regions marked for potential quakes should be similarly assessed to establish a clear link. And more than one kind of data – apart from slow slip – are needed.

“The discovery of slow slip at shallower depth is a significant discovery. However I don't think we can draw too many conclusions about the global significance of slow slip as a precursor at this stage,” says Ian Main, professor of seismology and rock physics and director of research at the school of geosciences at the University of Edinburgh.

Main says another study of slow slip prior to the 2011 Tohoku-Oki earthquake in Japan², based on seismological data, “is likely very relevant” to the observations from India.

“The main problem with identification of earthquake precursors after a big earthquake is that the sample is biased. This is a bit like a doctor examining patients who get better after a new drug is introduced in a trial. We don't know how many slow slip events in the Andaman area did not end in a large earthquake (false alarms), or how many have no slow slip precursors (misses).”

“Any deformation before an earthquake can be definitely treated as a precursor,” says M L Sharma, professor of civil engineering at the Indian Institute of Technology, Roorkee and president of the Indian Society of Earthquake Technology. “But in the case of the Andaman quake, the GPS observatory is 1000 kms away from the source of the quake and so this case cannot be treated as a precursor.”

Sharma says that the research is “certainly important” as in the near future GPS data will be very helpful in finding precursors. But there are constraints and unknowns, especially about differential motion, or movements at various depths that need to be addressed, he says.

N Purnachandra Rao, chief scientist at the National Geophysical Research Institute (NGRI) in Hyderabad says the major problem with most precursors is the lack of consistency and repeatability both in space and time. “Aseismic creep and slow earthquakes are well known phenomena which have been investigated rigorously in recent times. It is however, too premature to jump to any conclusion about their utility for a general earthquake prediction model.”