I know what he meant, but
"We want to make this a killer app" probably isn't an ideal quote for this sort of application.
There's a new smartphone app that could save your life by giving an early warning of an impending earthquake. The app, dubbed MyShake, was developed as a collaboration between UC Berkeley and Deutsche Telekom and uses the accelerometers built into modern smartphones to sense for earthquakes. "MyShake cannot replace …
It might be theoretically useful for those earthquake-prone countries with governments that so don't give a shit about most of the people that they can't be arsed to spend far less than, e.g. the cost of one of their president's executive helicopters, to provide the population with a marginally adequate seismic network.
But the chances are that the real utility is in the app itself providing Californian smartphone users with a minute or so's warning from the extensive set of high performance networked seismometers equipped with precision accelerometers installed in the state, rendering the limited sensing capabilities of the smartphones irrelevant.
Even 400 sensors is not a lot for big California. There will be many gaps where smart phones can fill in.
Also, while it's nice to have those 400 high quality datasets, It would also be desirable to have 50,000 low quality datasets too. Boffins will be able to produce an accurate time-simulation of a quake's wave motions and strengths over a big area with very many data points, and fully correlated to the high q data.
Who knows what insights may be gleaned with such a clear look at a quake?
Low quality data brings serious analysis problems. To me this sounds like a good idea but there is likely too much garbage included in the data to make it essentially useless.
Now if they are saying that a rejiggered smartphone could be used a cheap seismometer linked into the seismic monitoring network from fixed locations that might work.
> "To me this sounds like a good idea but there is likely too much garbage included in the data to make it essentially useless."
Not necessarily. The one data axis that's guaranteed to be accurate for smart phones is the timing of shake events. Even with all the garbage, the signal can be winkled out due to the timing and large number of data points. And while perceived shake strength will vary due to the type of connection between the phone and the shaking plate, a sufficiently large sample allows averaging that should be reasonably consistent. Those averages can then be normalized by use of the nearest high q data.
Even 400 sensors is not a lot for big California. There will be many gaps where smart phones can fill in.
Remember that earthquakes usually occur just along fault lines, so you only need to instrument those for prompt earthquake detection. You then just issue an alert to everyone within some strengh-dependent distance of it.
That way you don't have to leave the app running in the background all the time. A confirmed quake notification should ideally be sent out via the phone's emergency alert system (so it is on TV too) for maximum reach rather than just those who have the app running and their smartphone nearby.
I agree. For a first cut, this is worth contributing time to I believe. Since I generally leave my LG GPad 8.0 on the charger with the WiFi on, it gives them data to play with. Not that I expect it to be good data since the main train yard in Fresno County across the street.
Over the years I've done SETI@Home, Folding@Home, and a Few other projects. What the heck, I've the machines standing by between torture sessions. Put them to use.
I really can't see how this adds value in places like California with the current seismic network with hundreds of very, very sensitive sensors.
Earthquakes sometimes fo give some warning in that P waves travel faster than the S waves. The S waves cause the damage, so if the source is far enough away you get some warning by detecting and reacting to the P waves.
We just had a 5.7 here in Christchurch. No real damage. No warning either and no fancy app would have given us warning either.
The Christchurch earthquake wasn't the type that could've provided any practical warning. First, the epicenter of the quake was out to sea, so no one would be there to detect it. Second, it was only 15km offshore, meaning the seismic waves wouldn't have to travel very far. Any Primary/Secondary lag time would be measured in seconds.
Seeing as the accelerometers are being utilized, another algorithm could look for the tell tale signs of other calamities, like:-
1. Speed goes from zero to 70mph and back to zero very quickly.
Action. Inform emergency services, with GPS location, and a pre-record voice message giving your personal details, and/or rare blood type, allergies, etc.
2. Gone from vertical to horizontal while falling at 9.81m/s/s and then remained motionless for five seconds.
Action Bleep loudly, cos they've either dropped it and not noticed, or they might have collapsed, in which case run the possible emergency action when the annoying bleep gets no response.
3. Falls at a rate consistent with being dropped in liquid
Action Text Apple informing them you have voided your warranty.
Perhaps, not too well thought out, but I hope you get the gist of it.
Anyone got any proper ideas?
1. Speed goes from zero to 70mph and back to zero very quickly.
Action. Inform emergency services, with GPS location, and a pre-record voice message giving your personal details, and/or rare blood type, allergies, etc.
So it will only work if I boot it then slam on the brakes?
I was part of the Quake-Catcher Network for several years - small sensor mounted on the floor with my desktop analysing accelerations and sending packets to Stanford. Apparently proved the concept well, and my setup reported on several quakes, but maps of user locations showed that the network was over-represented where lots of tech people live and under-represented where most earthquakes happen. It seems that the grant ran out, and the network is no longer really active. Maybe this will replace it.
I guess we can consider ourselves fortunate that so few earthquakes are actually centered WITHIN a heavily-populated area because not only would there be no warning at all, not even a P wave because you're at the point they're being generated, but also maximum impact because all the energy's right there.
... we had a seismograph wired to send an alarm (sonalerts in all participants living quarters, ~100 participants) at the first sign of fairly low-level P-waves. After a year or so, not a single one of us managed to get out of the house before the S-waves got there.
Needless to say, the project was dropped as useless.
THAT said, as a many data-point gatherer, I'm all for it!