Humans today are able to forecast atmospheric weather quite accurately, which allows us to alert communities about the arrival of a hurricane, a snowstorm or a possible flood. Scientists can also predict the risk of a volcanic eruption or the effects of solar flares. It would seem, therefore, that nowadays we are able to foresee with greater or lesser accuracy the different types of natural disasters that stalk us—except for one. Every time the earth trembles we are caught completely off guard, often suffering great devastation and loss of life. Why has 21st century science still not managed to accurately predict when a quake will strike?
“We can see inside weather systems from below, from above and from inside them. We understand the laws of physics and the mathematical equations that control weather and climate,” Terry Tullis, former chairman of the National Earthquake Prediction Evaluation Council of the United States Geological Survey, tells OpenMind. “For earthquakes it is harder for many reasons. The Earth is opaque, so we can’t see inside it to figure out what is going on,” says Tullis. The seismologist notes that it is possible to observe certain geophysical parameters, but that earthquakes generally originate at considerable depths where it is impossible to place measuring instruments.
Predicting seismic movements is an age-old aspiration of human beings, even before the development of modern science. In ancient times it was believed that animals sensed earthquakes. As the physicist-chemist and writer Helmut Tributsch recounted in When the Snakes Awake (The MIT Press, 1984), many historical anecdotes speak of changes in the behaviour of animals before a tremor. The oldest event collected by Tributsch dates from 373 BC, when snakes, ferrets and mice left the Greek town of Helice two days before the occurrence of an earthquake and tsunami that submerged the city.
Tributsch is not the only one who believes in this ability of animals. The British biologist Rachel Grant has published several studies documenting how different species in different places, including toads and cows in Italy or rodents and tapirs in the Andes, have modified their behaviour before an earthquake. With the collaboration of geophysicist Friedemann Freund from the NASA Ames Research Center and the SETI Institute, Grant has linked these changes with disturbances in the ionosphere due to electrical phenomena in the rocks under stress, which in turn can alter the chemistry of the water.
Forecasting or prediction
But in the absence of more studies to verify how much is true in this hypothesis, seismologists have tried for more than a century to detect geophysical changes that help anticipate the tremors. In the 1970s, there was an idea that it would soon be possible to make accurate predictions, but the repeated failures cooled the enthusiasm. The efforts have been remarkable in Japan, where there is even a Committee for the Prediction of Earthquakes. However, most experts today differentiate between forecasting and prediction. “When using probabilities we usually use the term “forecasting” rather than “prediction” since prediction carries the connotation that we have more definite knowledge of what will happen than what is actually the case,” says Tullis.
The current consensus among scientists dictates that prediction, in the sense of predicting in detail when, where and with what magnitude a tremor will take place, is impossible today. Some think it will always be so. Robert Geller, an American seismologist at the University of Tokyo and a critical voice of the Japanese earthquake prediction program, says it is a settled debate: “Everyone knows earthquakes can’t be reliably and accurately predicted,” he tells OpenMind. For Geller, those who still defend this possibility are either scientists who use this idea as a “fund raising slogan” or “amateurs or very poor professional scientists who are deluded.”
Forecasts, on the other hand, attempt to estimate “probabilities of occurrence of a given sized earthquake in a given area in a given time period,” in the words of Tullis. And in this sense progress is being made, as in the case of modelling the string of aftershocks that occurs after a large earthquake. As an example, a recent study published in Science Advances has analysed the sequence of seismicity that took place in the Italian region of Amatrice-Norcia between 2016 and 2017, whose main tremor caused about 300 deaths on August 24, 2016.
“The main conclusion is that we have models that can describe probabilistically with accuracy the evolution of complex seismic sequences with different bursts of seismicity,” summarizes the study’s director, Warner Marzocchi, of the Istituto Nazionale di Geofisica e Vulcanologia de Roma, for OpenMind. “We provide probabilities and not deterministic predictions,” he clarifies. The seismologist adds that these models are already being applied to other regions such as New Zealand or the US, so they will help to refine forecasting. “For now the largest weekly probabilities of large earthquakes are about a few percent; when we are able to move above 10% or even larger, it will be a great step forward, and it will facilitate the adoption of mitigation measures”.
Very-last-minute warning systems
The fact that the likelihood of a tremor for a period of days or weeks is still very small deters the taking of steps that might help to mitigate the effects of the quakes, beyond the permanent ones, such as the design of safer buildings. According to Tullis, the so-called earthquake “early warning systems” are still in fact “very-last-minute warning systems.”
These are devices that detect the start of a tremor and calculate how long it will take the waves to reach certain places, so that the surrounding areas can be alerted. But Tullis warns that, in many cases, it is not even possible to know in the first moments how much the magnitude of an earthquake will grow. For example, in the case of the 2011 earthquake in Tohoku (Japan), which caused the accident at the Fukushima nuclear power plant, the movement grew disproportionately after the initial detection, so the warning issued underestimated the violence of the shocks.
“Unlike the weather forecasting case, we don’t know enough about the physical behaviour of the material in fault zones to be able even to have good equations to describe how earthquakes work,” laments Tullis. However, the scientist is confident that remarkable advances will be achieved over the next decades. “I’m hopeful that we will make significant progress, but earthquake prediction will never be as reliable as weather prediction for example,” he concludes. “Someone once said that it’s hard to make predictions, especially about the future,” he jokes.