Please note that the magnitude scale for the older earthquakes is not well known – its basically estimated from the sediment depth layers from core drilling the turbidity canyons (which are now easily seen in the latest Google Earth images off the Oregon coast). These canyons form when the continental shelf shakes like crazy and lots of sediment rolls down the continental slope (at high velocity) and carve these canyons. The magnitudes of the larger quakes could be higher (like around 9.5) for some occurrences.
A very good article on all of this, which comes outside the normal channels, can be found here
It’s even harder to determine the timing and magnitude of athletic scandals, as we learned from Gottfredson’s cover-up.
Hippo
10/14/2015
I would guess that the interarrival times for scandals, unlike earthquakes, is truly memoryless (exponential), since our institutional memory is so limited and responses so short-term.
Zaq
10/14/2015
I have a serious question about earthquake prediction: is there any evidence of the imminent “big one” other than historical trends? This whole business seems a bit too inductive for me to really take the threat seriously, but I know nothing about earthquake science.
With academic scandals we can deduce the data-generating process based on the incompetence of the administration. Can we do anything of the sort with earthquakes?
Dog
10/14/2015
@Zaq
Two points
1. Clearly there have been major earthquakes along the Cascadia zone and clearly the last one occurred on a longer timescale than
the period between major quakes. So, statistically another one should strike now (now being now +/- 50 years).
2. Google on Cascadia Initiative to find that we have 12,000 sensors out there monitoring the plate for deformation as we are likely to have 2-3 weeks of advanced warning. In addition. there are current active volcanoes (undersea) on the main pressure ridge line, so the engine has not stopped.
XDH
10/14/2015
Dog – for a biologist (I presume) you sure are versed in your Pacific NW geology!!!
Zaq
10/14/2015
Thanks for the reply, Dog.
Being statistically “due” for an earthquake is just trend fitting though, right? i.e. purely inductive?
I understand that the “engine” is still running, but do we know anything about the engine itself? Can we measure how “fast” it is running independently of how many earthquakes it produces? To put it another way: how would we know if the engine *stopped,* other than by seeing a reduction in major earthquakes (or movement via sensors)?
Dog
10/15/2015
we would know if the engine stopped when the when things like
the axial seamount stopped occurring
Regarding your first point: you seem to say that the arrival of earthquakes is not Poissonian (otherwise, the probability of the next earthquake would be independent of how long ago the previous one was). Are there papers on what the waiting time or arrival time distribution might be for this type of earthquakes?
Hippo
10/14/2015
The process is obviously non-Markovian. Tension in plates builds up over time, making the probability of an event depend greatly on distance to previous earthquake. I am too lazy to do a search, but yes there is huge literature saying that earthquakes do not follow a Poisson process.
Dog
10/15/2015
One more reference
this is a comprehensive statistical treatment of occurrences (over the last 10,000 years) that also focuses on the notable gaps between occurrences
Please note that the magnitude scale for the older earthquakes is not well known – its basically estimated from the sediment depth layers from core drilling the turbidity canyons (which are now easily seen in the latest Google Earth images off the Oregon coast). These canyons form when the continental shelf shakes like crazy and lots of sediment rolls down the continental slope (at high velocity) and carve these canyons. The magnitudes of the larger quakes could be higher (like around 9.5) for some occurrences.
A very good article on all of this, which comes outside the normal channels, can be found here
http://geodesygina.com/Cascadia.html
It’s even harder to determine the timing and magnitude of athletic scandals, as we learned from Gottfredson’s cover-up.
I would guess that the interarrival times for scandals, unlike earthquakes, is truly memoryless (exponential), since our institutional memory is so limited and responses so short-term.
I have a serious question about earthquake prediction: is there any evidence of the imminent “big one” other than historical trends? This whole business seems a bit too inductive for me to really take the threat seriously, but I know nothing about earthquake science.
With academic scandals we can deduce the data-generating process based on the incompetence of the administration. Can we do anything of the sort with earthquakes?
@Zaq
Two points
1. Clearly there have been major earthquakes along the Cascadia zone and clearly the last one occurred on a longer timescale than
the period between major quakes. So, statistically another one should strike now (now being now +/- 50 years).
2. Google on Cascadia Initiative to find that we have 12,000 sensors out there monitoring the plate for deformation as we are likely to have 2-3 weeks of advanced warning. In addition. there are current active volcanoes (undersea) on the main pressure ridge line, so the engine has not stopped.
Dog – for a biologist (I presume) you sure are versed in your Pacific NW geology!!!
Thanks for the reply, Dog.
Being statistically “due” for an earthquake is just trend fitting though, right? i.e. purely inductive?
I understand that the “engine” is still running, but do we know anything about the engine itself? Can we measure how “fast” it is running independently of how many earthquakes it produces? To put it another way: how would we know if the engine *stopped,* other than by seeing a reduction in major earthquakes (or movement via sensors)?
we would know if the engine stopped when the when things like
the axial seamount stopped occurring
see
http://oregonstate.edu/ua/ncs/archives/2015/apr/researchers-think-axial-seamount-northwest-coast-erupting-%E2%80%93-right-schedule
@Doc
Regarding your first point: you seem to say that the arrival of earthquakes is not Poissonian (otherwise, the probability of the next earthquake would be independent of how long ago the previous one was). Are there papers on what the waiting time or arrival time distribution might be for this type of earthquakes?
The process is obviously non-Markovian. Tension in plates builds up over time, making the probability of an event depend greatly on distance to previous earthquake. I am too lazy to do a search, but yes there is huge literature saying that earthquakes do not follow a Poisson process.
One more reference
this is a comprehensive statistical treatment of occurrences (over the last 10,000 years) that also focuses on the notable gaps between occurrences
http://activetectonics.coas.oregonstate.edu/paper_files/Kulkarni%20et%20al.,%202013.pdf
Thanks dog, just what I’m looking for!