Slashdot Mirror

← Back to Stories (view on slashdot.org)

Why Are We So Bad at Predicting Earthquakes? (telegraph.co.uk)

Posted by EditorDavid on from the rumbles-in-the-jungles dept.

In the wake of major earthquakes in both Japan and Ecuador, one British newspaper asks: Why are we so bad at predicting earthquakes? In 2015 seismologists told Vice, "The more we study them, the harder they look to predict, and "there's a shortage of instrumenation." But today the Telegraph newspaper concludes that we actually have two problems: first, "science is hopeless at predicting earthquakes and, second, we keep building cities on major fault-lines..." They cite a new book called Earth-Shattering Events which reports that nearly half the world's large cities are in earthquake-prone areas, adding, "we don't just build our cities on fault-lines, we also tend to rebuild them, in the same place, but no more robust, time and time again." In 1976 one quake in China killed more than 750,000 people, while a 2004 quake in Indonesia killed 170,000. "The Earth will move and there's not a thing we can do to stop it," the Telegraph concludes, arguing that we need to learn more from our past.

26 of 174 comments (clear)

  1. Because by Anonymous Coward · · Score: 2, Insightful

    Because we haven't prosecuted enough scientists for failing to predict earthquakes. Italy is on the ball. Get with it, world!

    1. Re:Because by shaitand · · Score: 2

      The scientists were wrongly prosecuted. They reported their assessment based on a review of the data available to them. They did not state whether or not there would be an earthquake.

  2. SImple answer... by QuietLagoon · · Score: 4, Interesting
    Because we do not have appropriately accurate modeling of the parts of the earth which cause earthquakes occur.

    .
    Why don't we have the appropriate modeling? Because we do not have enough information to create those models.

    Why don't we have the information to create the models? Because we do not explore the earth enough.

    1. Re:SImple answer... by ewibble · · Score: 5, Insightful

      Simpler answer,
      We don't know very much.

      This applies to most things, not just earthquakes,
      Medicine, (why do we need so much testing? because we are taking a stab in the dark and seeing if it works)
      Weather, how many are forecasts inaccurate.

      In fact any system that is even mildly complex we blunder our way through, even an area like programming, where we know exactly how the system works, and all inputs, we still need to test rigorously in to ensure that we haven't made to may mistakes.

      Also earth quakes are probably a chaotic system so we probably cannot get close to even knowing enough inputs to predict them, no matter how much we explore. (It doesn't mean we shouldn't try)

    2. Re:SImple answer... by rasmusbr · · Score: 3, Insightful

      Honest question: What would it take to make that happen? Money, time or technology?

      It's not even know whether or not it is possible in principle to predict when a major earthquake will happen. If you look at a time series of magnitude measurements at a particular fault it looks like something coming out of a random number generator. It might be predictable, but it's not obvious that it is.

      Some systems are fundamentally unpredictable because their long-term behaviour depends on arbitrarily small differences in the initial state of the system.

    3. Re:SImple answer... by PsychoSlashDot · · Score: 4, Interesting

      Because we do not have appropriately accurate modeling of the parts of the earth which cause earthquakes occur.

      Actually, it's because we're bad at predicting, period, end of sentence

      Wonderful things like reinforcement bias and the availability heuristic meddle with our accuracy. There's a reason why earthquake insurance sells the best days after an earthquake, when statistically that's when there's the least risk. When something horrible happens, people suddenly remember, and Do Something, but as the years add up since the last event, we become lax.

      That's the personal level. I recommend reading Risk by Dan Gardner to lean a bunch about this in general. From that book, he moved on with Philip Tetlock and wrote Future Babble and then Superforecasting, which more closely refine the general psychology covered in Risk down to mathematical models and expert opinions. Risk is a fascinating read useful to individuals to understand why grocery stores have sales with "limit 10 per customer", and why more extra Americans died driving the year after September 11th than died in the attacks themselves, and the other two get (much) more into "why are economists wrong so often" (Future Babble), and finally "why are some people so good at analyzing data and predicting things like complex geopolitical events and the like, and how can we learn to be like them" (Superforecasting).

      I leave locating the books to the audience as you know what bookstores you prefer and the author and titles are very clear.

      --
      "Oh no... he found the .sig setting."
    4. Re:SImple answer... by plopez · · Score: 2

      That is only part of the story. More instrumentation may not help if you are measuring the wrong things. Secondly you are dealing with complex systems modeled by non-linear equations with no closed form solutions. Ever. There may be a plethora of potential solutions but they may represent local solutions as opposed to global solutions. Past a certain point Math can't help you and if Math can't help you neither will a computer model.

      Weather models have gotten better at long range global trends and very short range predictions, about 48 to 72 hours. But 100% accuracy is impossible. But perhaps earthquake models might make it to the 2 or 3 day warning point. But I am skeptical.

      --
      putting the 'B' in LGBTQ+
    5. Re:SImple answer... by sysrammer · · Score: 2

      Earthquakes != Tectonics, as Weather != Climate.

      --
      His ignorance covered the whole earth like a blanket, and there was hardly a hole in it anywhere. - Mark Twain
    6. Re:SImple answer... by silentcoder · · Score: 2

      We can predict the next eclipse (the next several in fact) - and the solar system is chaotic. Chaotic systems are not wholly unpredictable. They have what's known as a prediction-horizon, a maximum period after which predictions become impossible as errors in measurement add up.
      The more accurately you can measure - the further you can push the prediction-horizon. For the solar system the prediction horizon is about 2 million years. For weather it's about a week. For climate - about a million years (climate is far, far infinitely far simpler than weather - still chaotic but far less so).

      Earthquakes however fall into the category where we can accurately predict *if* but not *when* - partly due to insufficient measurement. I can say with absolute certainty that Japan will get hit by another big earthquake and probably before the century is out. Nobody can yet tell you when. We can do the *where* very well, any city on a fault line WILL get hit every now and then. Other places it is exceedingly rare and if you account for mining and groundwater depletion sources you can usually predict those too.

      We can't predict where the next flood will be, or when, but I can predict with absolute certainty that a major flood coupled with millions dead from choking on CO2 will happen in central Africa in the near future. We know there is a massive CO2 bubble building under a lake. We know that when such bubbles eventually burst they send out a shockwave that sends the water over the banks and floods everything nearby in a pool of mud while a cloud of CO2 too thick to breath settles over the landscape and kills everything in a 20 mile radius.
      We know - because it happened already - within the last century in fact, and we know it will happen again because we can see the build-up happening.
      We don't know when because we can't see how big the build-up already is - we can't see because the same geological forces that is making the bubble is also boiling the lake... you can't scuba dive in a boiling lake, especially a very deep one. We can't go down to look and we don't have much camera equipment designed to survive being boiled.

      We can't extend the prediction horizon for earthquakes or make the predictions more accurate because we lack the means to accurately measure. Just like the boiling lake problem - we can see where the problem areas are, but we lack the technology to get the data to tell us when those problems will happen.

      That doesn't mean there is nothing we can do. As we speak the Kenyan Government is busy reinforcing the banks of the boiling lake with concrete and steel, hoping to strengthen them enough to prevent flooding when the bubble bursts and hopefully contain most of the gas. It may not work - but knowing it's coming, we can take measures to mitigate the potential harm. Even if it doesn't work, it's better to try and fail than not to try - again. And that's what we actually need to do about earthquakes. We need to make the cities built on likely strike-zones far more capable of keeping their citizens alive when it comes.

      --
      Unicode killed the ASCII-art *
  3. What straw will break the camel's back by Kjella · · Score: 5, Interesting

    We know where earthquakes will be - along fault lines. When will it happen? When the tension gets too high. It's like asking what pebble will start a landslide, what snowflake will start an avalance or what straw will break the camel's back. The problem isn't the lack of an answer, the problem is that we're expecting an exact result to a complex and chaotic process. Would you also like a map of where lighting is going to strike?

    --
    Live today, because you never know what tomorrow brings
    1. Re:What straw will break the camel's back by Mr+D+from+63 · · Score: 2

      We can't directly measure those stresses. We probably never will be able to, so we are left with measuring the symptoms. Also, creating an accurate model of subsurface lamination (if that is the correct term) is likely also very hard to do without drilling a huge number of sample holes.

      The fact that we build some cities near fault lines is a completely different topic, however, one could speculate that with more people living and working near fault lines, there is more money being spent on quake research than there otherwise would have been, so in that sense it is helping us solve the problem.

    2. Re:What straw will break the camel's back by leftover · · Score: 2

      Likely the only way we will 'predict' earthquakes is when we learn enough to cause them. Why would we do that -- to make several small ones instead of one big one.

      Fracking-related (actually wastewater reinjection) quakes demonstrate it is feasible with current technology.

      --
      Bent, folded, spindled, and mutilated.
  4. That's not really true by phantomfive · · Score: 4, Insightful

    "we don't just build our cities on fault-lines, we also tend to rebuild them, in the same place, but no more robust, time and time again."

    I can't speak for everywhere, but in California, the construction standards are much higher after the 1989 earthquake.

    Boston, however, is in serious danger because of so many tall buildings built of brick (and other reasons).

    --
    "First they came for the slanderers and i said nothing."
    1. Re:That's not really true by Mr+D+from+63 · · Score: 2

      You'll have to forgive the author. Accuracy isn't important when trying to make a point.

  5. We can predict earthquakes ... by Alain+Williams · · Score: 5, Insightful

    we know the sort of geology that is found in earthquake areas, so we can predict where they will happen; by measuring strain, etc, we can get an idea of when they will happen and what sort of magnitude. The trouble is that we (== common people, non scientists) expect answers that fit in with my everyday rulers and clocks (ie a few miles and days), but geological events are measured differently: hundreds of miles and decades/centuries; so the margins of error are too great for what we want.

    If I place a vase of roses outside on a summer's day I can expect the flowers to be visited by bees, but I cannot predict which flower will be visited first or the minute when the first bee will come.

    1. Re:We can predict earthquakes ... by sysrammer · · Score: 2

      "...so the margins of error are too great for what we want". Good explanation.

      --
      His ignorance covered the whole earth like a blanket, and there was hardly a hole in it anywhere. - Mark Twain
  6. "[W]e keep building cities on major fault-lines"? by 93+Escort+Wagon · · Score: 2

    We "keep doing" this? Are there some newly-built cities we've intentionally located on major faults?

    --
    #DeleteChrome
  7. Suppose we could... by R3d+M3rcury · · Score: 4, Interesting

    That brings up an interesting question--what do we do about it?

    "We've determined that on May 17, 2015, there will be a 6.0-level earthquake in San Francisco. We believe the accuracy of this prediction to be 90%."

    Great! We've got a month's warning. It's going to be major. We're pretty sure it's going to happen--but there's a possibility that it won't.

    Now what? Evacuate 800,000 people? Start building shelters that can withstand the earthquake? Do you tell people so that they can prepare?

    1. Re:Suppose we could... by sysrammer · · Score: 4, Funny

      Forecasting? It doesn't look like you can even get postcasting right.

      --
      His ignorance covered the whole earth like a blanket, and there was hardly a hole in it anywhere. - Mark Twain
    2. Re:Suppose we could... by evilviper · · Score: 3, Insightful

      "We've determined that on May 17, 2015, there will be a 6.0-level earthquake in San Francisco. We believe the accuracy of this prediction to be 90%." [...] Now what? Evacuate 800,000 people?

      Absolutely, the city would be evacuated. That's much higher prediction accuracy than we can get with hurricanes, yet cities are evacuated for those, sometimes even several times per year.

      Of course, the minority, whor are living in structures engineered to withstand very high earthquake loads would remain behind. But at least they'd have plenty of warning to strap down appliances before-hand, and stay away from dangerous spots on that day.

      Not to mention there would be a rush on home-improvement stores and contractors, as everyone rushes to reinforce their shoddy old buildings.

      --
      Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
  8. Re:For solidarity! by khallow · · Score: 4, Insightful

    More seriously, cities typically are there for a reason. And the reason usually doesn't go away just because there was an earthquake.

  9. Structured Criticality by Fragnet · · Score: 5, Informative
    It may not be possible to predict them. There's such a thing as Structured Criticality. That is to say:

    ... a property of complex systems in which small events may trigger larger events due to subtle interdependencies between elements. This often gives rise to a form of stratified chaos where the general behaviour of the system can be modelled on one scale while smaller- and larger-scale behaviours remain unpredictable.

  10. Re:The hubris of man by Impy+the+Impiuos+Imp · · Score: 2

    Earthquakes are man-made disasters, not natural ones. The vast majority of deaths are caused by collapsing buildings.

    We know to, and how to, make the buildings stronger.

    --
    (-1: Post disagrees with my already-settled worldview) is not a valid mod option.
  11. Re:predicting earthquakes is useless by ShanghaiBill · · Score: 4, Insightful

    I mean, what are you going to do, have everyone move houses and buildings to other locations? What a waste of time.

    Given a minute or so of warning, you can:
    1. Stop trains, so they don't come off the track.
    2. Stop additional cars from entering tunnels.
    3. Pretension dampers in tall structures
    4. Sound an alarm to warn people in warehouses and stores to move away from shelves.
    5. Pull up automatic safely webbing to prevent pallets from falling off racks.
    6. Stop and lower cargo on forklifts.
    7. Start powering down heavy machinery
    8. Stop people from entering elevators
    9. Open fire station doors, so they don't jam closed.
    10. Shutdown the flames in furnaces and water heaters
    11. Start reducing gas pressure in pipelines.
    12. Warn people on beaches to start moving to higher ground.
    13. Start backup diesels for emergency services.
    14. Retract the control rods from nuclear reactors.

    Since seismic waves travel about 5 km / sec, you can give useful warning to people further away just by quick detection. Of course, prediction would be better.

  12. Re:The hubris of man by pixelpusher220 · · Score: 3, Interesting

    Amazingly it depends on the height of the building. Shorter 'boxes' can ride out quakes pretty well being super rigid. Super tall need the flexibility. The problem is the 9-15 story buildings that aren't tall enough for the flexibility to help but too big to be rigid against the shaking.

    I work in a 12 story :(

    --
    People in cars cause accidents....accidents in cars cause people :-D
  13. Not true by infernalC · · Score: 3, Interesting

    There is a risk market for earthquakes. Actuaries would love to have better predictability for earthquakes to better calculate the risk so that insurance products can be priced more accurately. If someone is going to fund earthquake research, real property insurers are your best bet.