P2P software uses hard drives to detect, warn of tsunamis

News outlets and weather stations alike have been searching for more responsive (and foretelling) methods to detect massive storms before they make landfall, and it seems a simple hard drive or two could help out substantially. While certain folks have tried to get fancy and implement mobile warnings to alert citizens of incoming tsunamis, Michael Stadler has devised a P2P software solution that uses plain 'ole hard drive vibration detectors to not only keep the read / write heads aligned, but to feed analytic software those same quaking measurements. By having numerous supernodes compare vibration levels, the software can reject false alarms and substantiate actual threats based on the uniformity of the data -- if an actual tsunami looks to be rushing inward, all connected clients could be immediately informed to brace for impact. While Stadler's software is still in an "experimental stage," it recently caused quite a stir at the Ars Electronica exhibition, and could probably entice a few proactive governments to fund future development.

[Via Slashdot]

[leeeeshad @ Sep 8th 2006 5:24AM]

surly they should use water sensers aswell

[LazyMegaMan @ Sep 8th 2006 5:36AM]

Not really sure why they didn't have this in mind too, but earthquakes would be nice to know about, especially in California where many major cities reside right next to a huge fault line.

[realist @ Sep 8th 2006 7:34AM]

Tsunamis are caused by large quantities of mass moving and displacing water. This is most likely to be due to earthquakes, but also includes landslides or volcanic activity.

Purpose built seismometers can detect minute vibrations caused by the passage of seismic waves from an earthquake on the other side of the planet. The stations where they are based are specially constructed to isolate the seismometers from any sort of local environmental noise.

What will actually help with real-time tsunami warnings is faster and more accurate analysis of the triggering earthquake. Knowing the displacement that occurs along a fault means you can make a more accurate estimate of water displacement, and hence tsunami potential. From that point, you can model the propogation and size of a resulting tsunami, to allow better warnings to be distributed to shorelines.

Trying to analyse hard drive vibrations and is somewhat optimistic. For a start, seismometers have a very wide frequency-response range to vibrations, and are purpose-designed for this - the same can not be said for hard-drives.

While there are clearly many, many times more hard drives out there than dedicated seismic stations, the supposed benefit of having so much data would most likely be cancelled by the massively increased time required to extract any sort of useful signal, especially from such a narrow frequency response, that cannot distinguish noise from signal (eg; a truck driving past your house, and actual seismic signals). And this is without even getting to the fact that tsunamis are vibrations in a liquid - when vibrations from tsunami reach your location, that's because the tsunami is already there.

This seems like a case of someone with technical knowledge/ability not grasping the physical science basics behind the actual problem. Someone's noticed that PC's have (very limited) vibration sensors in them, and jumped on the shared p2p bandwagon to solve the problem.

Improved tsunami warnings will come from improvments in dedicated seismic monitoring for early warnings of a potential tsunami in the local region near the earthquake. Followed by fast, accurate modelling of tsunami propogation, to allow reliable warnings to be sent to coastlines at greater distance. One of the problems with early warning systems for natural hazards, is that if they aren't accurate, then their effectiveness in the future is reduced.

Sending data from actual seismic monitoring stations to a p2p network might help with improved earthquake models. But the time-critical nature of the problem means that such analyses are probably better left to dedicated supercomputers or clusters.

Actually, I think I'm going to resurrect my old slashdot account and post this there as well :)

PS: Engadget is a great news site - but stop using 'Drop' to mean relases - it means the opposite!

[psxp @ Sep 8th 2006 8:58AM]

realist, one of the best posts I have read on this site. Yes, the point that "This seems like a case of someone with technical knowledge/ability not grasping the physical science basics behind the actual problem." is very true!!!

[chris @ Sep 8th 2006 2:18PM]

Realist said-- "PS: Engadget is a great news site - but stop using 'Drop' to mean releases - it means the opposite!"

Amen! Sometimes Engadget tries so hard to sound cool that they sound lame. I've even seen two posts in a row with drop meaning "release" in one and "discontinue" in the other.

[Franssu @ Sep 8th 2006 3:21PM]

This looks interesting, but my friends at the *AA told me that P2P is bad and should be outlawed...

[realist @ Sep 8th 2006 8:20PM]

The problem isn't the p2p concept - there's no shortage of seismic data out there that would benefit from having all that processing power to analyse it with.

The first problem is using vibration detectors in PC hard disks, as opposed to actual, purpose built, wide-frequency-response seismometers, is like trying to study cosmology using a pair of reading glasses. With enough results, at _very_ best you might, possibly, be able to provide a very rough estimate of the location of an earthquake, by trying to look for vibrations spreading out from some central point. But a rough estimate is useless in the context of tsunami warning systems. You couldn't be sure if the event is even offshore, which defeats the purpose somewhat.

But dedicated seismic networks can already do this (and have been doing so for decades) in very short periods of time, through a combination of incredibly sensitive seismometers, and trained obeservers.

P2P networks might provide some use in helping improve models of earthquake ruptures and tsumani propogation, but only when proper data from seismometer networks has been pre-processed and then distributed.

The basic concept of hard drive vibration sensors as seismometers-lite, is unfortunately very unlikely to be of any use whatsover.

[Positivist @ Sep 11th 2006 2:24AM]

Useful idea. I say let's be positive about it and let them try and see how and for what purpose it works best (unless there's some Big Brother practice involved).
Millions of computers are still millions of computers, and the internet is a huge and fast network, so let's not underestimate its powers.