The devastating magnitude 6.9 earthquake that ripped through Sikkim on September 18, 2011, reportedly killing more than 100, did not catch everyone by surprise.

Since earthquakes occur suddenly with devastating consequences, earthquakes prediction is of great interest to the general public. However, the term “earthquake prediction” is often used to mean two different things for the public and the seismologists. [TOPIX]

Earthquake Predictions

From public’s perspective “earthquake prediction” means a highly reliable, short-term (within hours to weeks) prediction that can be used for emergency measures such as evacuation. The issue is then whether the quality of prediction as currently available is good enough for this type of prediction. The current science has definitely not reached this level of confidence to issue such a prediction. For the seismologists, “earthquake prediction” is a statement regarding the future seismic activity in a region. Since the basic physics of the earthquakes is now reasonably understood and high quality seismic data are available, it is quite possible to make some prediction regarding the future seismic activity in a region. However, this type of prediction is different from the short-term prediction usually asked by the general public.

In fact, the earthquake prediction is a matter of intense debate among the seismologists and earthquake scientists. There is much disagreement between the two sides. But both sides agree that it is impossible to make definitive predictions about earthquakes accurately. There are two extreme views on earthquake prediction – on one hand, scientists believe that earthquakes are a “stress accumulation and release process driven by relative plate motion”, in which case it may be possible to find out when faults will slip and earthquakes occur, and on the other hand, earthquakes are random phenomena and impossible to accurately predict in terms of when they will occur.

What have been learned in the last 40 years or so is that the earthquake process is indeed highly complex. The number of parameters involved in an earthquake process is so large and difficult to measure. For instance, the earthquake source (hypocenter) is deep (10 to 200 km) in the earth and is not directly accessible for direct observation, and the state of stress, which is related to impending earthquake size, cannot be measured directly.

Regarding prediction, we are basically interested in 3 things about earthquakes: one is where will earthquakes occur, the second is how big will they be, and the third question is when will they occur. The seismologists reasonably know the answers to the first two questions. But the third one is the really a difficult problem. Scientists have reasonable idea of where an earthquake is likely to strike and how big it will be, but there is still no way to tell exactly when it will happen.

Map showing 3 seismic gaps in Himalayas. Stars denote past earthquakes of magnitude 8 or greater. Source: Bilham et al 2001

Sketch showing Indian Plate going underneath Tibetan Plate causing growth of Himalayas & earthquakes

How earthquakes occur is well known now. The Earth’s crust is broken into several pieces known as tectonic plates that constantly bump and grind or slide past each other. The plate movement happens slowly. It is only about 2-5 cm per year in the Himalayas. Even then, over several decades, there is enough stress build-up and the rocks suddenly slip, releasing huge energy that we feel as earthquake. Scientists now know where these fault lines are, and maps of where earthquakes are most likely to occur are also readily available. Most large earthquakes occur on long fault zones.

Earthquake magnitude depends on the size of a fault segment that ruptures, the stiffness of the rocks, and the amount of accumulated stress. Where faults and plate motions are well known, the fault segments most likely to break can be identified and likely magnitude estimated. However, only long-term forecast is possible for the timing of the earthquake.

Seismologists have long warned that a large earthquake is overdue in the Himalayan region based on the Seismic Gap theory, which is the most frequently used method for long-term earthquake forecast. A seismic gap is a zone along a tectonically active area, where no earthquakes have occurred recently although continuous stress build-up is known to be occurring. This probability forecast is based on the idea that a fault builds up stress until it reaches a critical point, and release the strain as an earthquake. Then the whole process starts over. According to this theory, the faults that have not had an earthquake in the longest time are most at risk and those that just had large earthquakes are least at risk.

Based on the Seismic Gap theory, 3 seismic gaps, which are the areas expected to have large earthquakes in the near future, have been identified in the Himalayas. The seismic gap theory has been conventional wisdom for the last 40 years. However, it did come under attack recently when UCLA seismologists scrutinized a global set of forecasts made in 1979 and found that areas thought to be at low risk of earthquakes - the ones that had recently had quakes - actually experienced 5 times as many shocks as perceived high-risk areas. The seismology community is still debating the issue.

The proponents of the Seismic Gap Theory however, argue that a number of recent large destructive earthquakes had occurred after an earthquake silence of long period supporting their seismic gap theory. For example, it was 600 years between the 2004 Indonesia earthquake (M9.0) and a previous destructive earthquake in the region; about 500 years to the 2008 China Wenchuan earthquake (M7.9); and about 200 years to the 2010 Haiti earthquake (M 7.0). However, the recorded earthquake history is relatively short since research in earthquake started only about 40 years ago and reliable seismometers did not exist before 1976.

One basic idea behind earthquake prediction is that faults send out subtle but detectable warnings before they slip. Scientists have looked at a host of potential warning signals, or “precursors,” including foreshocks, weird animal behavior, and changes in the water table, stream flow, well levels, and patterns of electrical currents in the ground, albeit without much success.

On February 3, 1975 a number of small earthquakes struck Haicheng Province in Manchuria (China). Believing them to be foreshocks (precursors) of a big earthquake, Chinese geophysicists issued a warning that a major earthquake would strike within the next two days. Sure enough, less than 24 hours later a major earthquake of magnitude M7.3 struck, but thousands of lives were saved because of the warning. It was the first successful prediction of a major quake ever made. However, to this day, the events surrounding this prediction are marred by controversy, with skeptics claiming it was little more than a coincidence. Their argument is based on the fact that the same Chinese team failed to make predictions for other earthquakes since then. For example, the team failed to save the lives of the 240,000 who died in the devastating Tang Shan earthquake (M7.9) the following year.

Many predictions are based on precursors, defined here as the non-threatening, initial phase of a natural hazard. It is now generally agreed that the earthquake precursor must be based on sound geophysical science, and perform significantly better than random chance. During an earthquake two seismic waves are produced: one is the Primary waves (P-waves) and the other, the Surface Waves. P-waves travel faster than surface waves and it does not cause damage. It is the surface waves, which causes the shaking and damage. Since P-waves travel faster than surface waves, it can be used to forecast the severe shaking of the earthquake. But this warning is only for less than a minute. Even so, the governments of Japan, Taiwan and Mexico have all decided even a few seconds warning are worth having. In 2004, a P-wave detection system in Japan gave 3 seconds warning of an earthquake, which was enough to allow the famous Bullet Trains to be stopped before the severe shaking.

Left: Seismometer for measuring earthquake. Right: Seismograph showing seismic waves (P waves and surface waves)

A lot is going on in this area of earthquake science, including efforts to drill down into major faults – to get a better sense of how earthquakes work. Still, the ability to predict earthquakes precisely is so far off in the distance that some organizations are advocating expending our resources on learning how to better deal with earthquakes when they happen. For example, US Geological Survey (USGS), which is on the forefront of earth science, states clearly that they do not make earthquake predictions. Instead, they will only make long-term forecasts about the likelihood of an earthquake as well as assessments of earthquake damage.

Presently, the appropriate question on earthquake prediction is not whether or not we can predict earthquakes; rather it is how well we can predict them. It may never be possible to predict the exact time when a damaging earthquake will occur, because even when enough stress has built up, a fault may become inherently unstable, and any small background earthquake may or may not continue rupturing and turn into a large earthquake. So there is room for optimism even on those regions where long-term forecast has been made.

While it may eventually be possible to accurately diagnose the stress state of faults, the precise timing of large events may continue to elude us. But the fact is, we already know what the results will be when a big earthquake strikes. Further, we have a pretty good idea where earthquakes occur and how big they could be. We also know that for places with a high rate of historic seismic activity such as our region of the Himalayas, the chance that an earthquake will hit in a future period of several decades can be quite high. So although we do not know exactly when the Big One is coming, we must prepare because the combination of a large fault ready to rupture and poorly prepared area can lead to wide devastation when an earthquake strikes as was the case in Haiti in 2010.

Earthquake preparedness can give us the best possible chance of minimizing the extent of the damage in the event of disaster. What needs to be done to reduce the vulnerability to an earthquake hazard is also very well defined. For example, efforts should be made to identify those areas that are most prone to earthquake through the generation of seismic hazard maps, so that informed decisions can be made about land-use and building codes. Most earthquake experts would agree that the authorities should focus on making homes, schools and dams as resistant as possible especially in the view of prevailing long-term forecast because Sikkim earthquake will definitely be not the last in the region. Surely, it will cost a lot of money to set up seismic stations, enforce stricter building codes and make disaster plans, but since we will have to spend it anyway – either in preparing for a future earthquake, or cleaning up after it – we really do not have much room to debate.