This section is a sort of asterisk * to a previous section about earthquake prediction. Before I proceed, I'll emphasize again that earthquake prediction is still not technically possible, despite legions of self-anointed amateur "expert" opinion to the contrary.
Most earthquake prediction research focuses on precursors: something, some clue, that might provide a warning. There are abundant stories out there of strange animal behavior, strange electrical phenomena, strange groundwater phenomena before an earthquake. In almost all cases this information has been gathered after an event, in an effort to understand what happened. In the vast majority of cases, there is no independent instrumental record to back up the precursor "discovery" - it's just the say-so of one or more individuals, after the fact.
An additional, more serious problem is is that these phenomena apply to one earthquake but not all earthquakes. Unfortunately each earthquake, like bears, humans, and gold deposits, is different - each has its own "personality," so to speak. Each has its unique rock, fault, and tectonic environment.
One physical concept that won't go away: tracking strain buildup. This concept is like crack cocaine for an earthquake research scientist. The idea is this: if you can see a change - especially an acceleration - in strain buildup, this might mean an impending earthquake is looming. Unless you can see an asymptotic (accelerating) pattern, however, you won't have much chance of saying when the fault will break. Measuring strain is technically difficult, and this makes the problem even less tractable. To do it correctly, you would need to measure strain as a function of time over a large horizontal region - and in a perfectly-funded world, also at depth. While there are proxies for measuring strain, they are different for each area, and hard to calibrate. Normally, professional scientists use - you guessed it - a strainmeter.
This latter is one of several goals of SAFOD: the San Andreas Fault Observatory at Depth. This is a hugely expensive endeavor, requiring much more energy by the (IMHO brilliant) principal investigators to just gather funding - than to even carry out the technical effort itself. A drillhole has already been punched down parallel to the San Andreas Fault, then at depth it was "whipped" (bent and re-directed) to pass through the fault (really a fault zone, a region of sheared-up rock several hundred meters thick) at depth. Besides sampling the rock, pulverized or otherwise at the fault zone itself, the plan was to install strainmeters in the fault zone.
Now the interesting news, published only recently (late 2011) in the Journal of Geophysical Research (Solid Earth). A group of scientists (Katsumata, principal investigator) investigated the earthquake catalog for events preceding the Magnitude 8 Tokachi-Oki earthquake of 2003 just off the Pacific coast of the Japanese island of Hokkaido. This is a nontrivial exercise, because there are hundreds of seismometers in Japan alone, plus they used everything in the general region - a lot of data to manage, much less wade through in a systematic fashion. The authors' statistical analysis shows a significant decrease in seismic activity in the region during the four years leading up to the earthquake. This can be construed as evidence of strain not being released by small events, but instead the strain was accumulating and perhaps even accelerating.
Of course, this is for just one fault, and statistics are famously arguable. This is also not the first effort to look for a quiescence signature. Therefor, you can expect much more additional work to be done on the earthquake catalog for, say, the Great Tohoku Earthquake of 2011.
~~~~~
Most earthquake prediction research focuses on precursors: something, some clue, that might provide a warning. There are abundant stories out there of strange animal behavior, strange electrical phenomena, strange groundwater phenomena before an earthquake. In almost all cases this information has been gathered after an event, in an effort to understand what happened. In the vast majority of cases, there is no independent instrumental record to back up the precursor "discovery" - it's just the say-so of one or more individuals, after the fact.
An additional, more serious problem is is that these phenomena apply to one earthquake but not all earthquakes. Unfortunately each earthquake, like bears, humans, and gold deposits, is different - each has its own "personality," so to speak. Each has its unique rock, fault, and tectonic environment.
One physical concept that won't go away: tracking strain buildup. This concept is like crack cocaine for an earthquake research scientist. The idea is this: if you can see a change - especially an acceleration - in strain buildup, this might mean an impending earthquake is looming. Unless you can see an asymptotic (accelerating) pattern, however, you won't have much chance of saying when the fault will break. Measuring strain is technically difficult, and this makes the problem even less tractable. To do it correctly, you would need to measure strain as a function of time over a large horizontal region - and in a perfectly-funded world, also at depth. While there are proxies for measuring strain, they are different for each area, and hard to calibrate. Normally, professional scientists use - you guessed it - a strainmeter.
This latter is one of several goals of SAFOD: the San Andreas Fault Observatory at Depth. This is a hugely expensive endeavor, requiring much more energy by the (IMHO brilliant) principal investigators to just gather funding - than to even carry out the technical effort itself. A drillhole has already been punched down parallel to the San Andreas Fault, then at depth it was "whipped" (bent and re-directed) to pass through the fault (really a fault zone, a region of sheared-up rock several hundred meters thick) at depth. Besides sampling the rock, pulverized or otherwise at the fault zone itself, the plan was to install strainmeters in the fault zone.
Now the interesting news, published only recently (late 2011) in the Journal of Geophysical Research (Solid Earth). A group of scientists (Katsumata, principal investigator) investigated the earthquake catalog for events preceding the Magnitude 8 Tokachi-Oki earthquake of 2003 just off the Pacific coast of the Japanese island of Hokkaido. This is a nontrivial exercise, because there are hundreds of seismometers in Japan alone, plus they used everything in the general region - a lot of data to manage, much less wade through in a systematic fashion. The authors' statistical analysis shows a significant decrease in seismic activity in the region during the four years leading up to the earthquake. This can be construed as evidence of strain not being released by small events, but instead the strain was accumulating and perhaps even accelerating.
Of course, this is for just one fault, and statistics are famously arguable. This is also not the first effort to look for a quiescence signature. Therefor, you can expect much more additional work to be done on the earthquake catalog for, say, the Great Tohoku Earthquake of 2011.
~~~~~
No comments:
Post a Comment