Monday, February 29, 2016

Leap Day and Quakes

Yay! Today is "leap day" - a day that only comes around once every four years, and which poses some unique challenges for kids born on that day :P  To celebrate, Google put up this awesome cute doodle on their homepage  :D

In other news, we had another nasty quake last night here in Christchurch :(   It struck at 3:32am on Leap Day, measuring "4.3" on the Richter Scale (TBH, this scale really sucks and is almost meaningless a lot of the time), but it was really shallow at 5km deep and had an epicenter that was just 5km from the center of the CBD, under a row of houses around the foot of the Port Hills  (and apparently nearly right under some poor lady's house!). So, what we had was basically a shallow quake very close to the city == Some rather scary shakey shakey action in the middle of the night!

Fortunately it wasn't a damaging one... but it was quite scary to have to go through another one like this, given that, like the one that kicked off all this seismic activity in the first place, it struck in the middle of the night! There's nothing quite like being fast asleep, then suddenly waking to find that the house is shaking with a high intensity/velocity vertical jitter that seems to ramp up and build up instead of going away, and to hear stuff falling on to the ground around you, sight unseen in the dark.

Indeed, many people were woken up (or "shoken up" as the radio hosts were discussing this morning) and had trouble getting back to sleep. Personally, I tried going back to sleep, couldn't do so after 5 minutes, and then proceeded to do some late night reading to calm the nerves a bit before finally getting tired enough to sleep an hour later at around 4:40am. I also couldn't help but think how accurate the description of "the kick" in Inception is... It was strange how last night, I remember vaguely dreaming about being in some tree-lined park (not unlike Hagley Park), and then starting to feel some rumbling/shaking, and waking to find that we were once again in the middle of an earthquake! Damn... these shakey shakeys are nasty little buggers, especially when they sneak up to you in the middle of the night!

For those of you wondering, no it hasn't really been like this all the time for the past 5 years. In fact, it's been really quite calm and "back to normal" for a good 1 to 2.5 years already... Up until the recent 5.7 quake on February 14 (which I rode out in the car for the first time ever; surprisingly, it didn't feel that bad... it was just like a gentle side-to-side wobbling motion, as if the rear left tire had either blown out, or had somehow just "fallen" off the car!), we really hadn't seen any major or frequent seismic activity for quite a long time. At long last, it seemed that maybe things were calming down and we would finally be good again for another 50-100 years (with maybe one slightly bigger "4" every year or so). But no. The Valentine's day quake seems to have just kicked it all off again... gah!


In news coverage of this today, the duty geologist asked to talk about the quake noted that the "peak ground acceleration" last night was very similar to that which we experienced in the Valentine's day quake - at about 0.2G IIRC.  Doing a bit of digging around, it does seem to me like practically all the engineers seem to believe that PGA (peak ground acceleration) is actually a better measure of the intensity/effects of seismic activity than using the Richter Scale ("energy released" magnitudes on a log scale) that are more commonly reported.

Specifically, all the seismic engineering models/standards/etc. all seem to use the "horizontal components" of PGA (PGA actually has 3 components - 2 horizontal, and one vertical. with the vertical usually ignored as it has a smaller amplitude... that said, IMO, most engineering tests currently overlook the fact that the ground surface often moves in waves, and not just in routine lateral "shunts", though really both of those are equally nasty).

However, the main reason they don't talk about PGA that much is that, although it can actually be directly measured (and used to create an overall score by averaging data from several reference sites), is that it depends a lot on the underlying geological structure of the land on which the quake is occurring in, as well as where the detectors are on the surface relative to the epicenter. As a result, it seems that Richter is still used more, as it has a longer history of usage, and is more generally "understood" of explaining and comparing seismic activity from different regions.


Ever since the earlier periods of quakes back in 2010/2011, like many Cantabrians, I've been interested/keen in the prospect of having some actual useful quake predictions. I'm sure I've talked about this on this blog in the past before, but the general consensus from the geologist community these days remains that they cannot actually predict when individual quakes will happen, how big they will be, and/or where. No doubt, the trial of a bunch of Italian scientists whose quake predictions for an ancient town that got pummeled by a quake appararently didn't go so well, would have quite a chilling impact on others being too keen to take a similar risk.  Instead, the "best" they do these days is hand out rough "broad scale maps of long term probabilities of quakes of certain sizes occurring". So, things like, "in the next 6 months, there's x% chance of a > magnitude 5 quake, and y% of a > magnitude 4 quake, etc."

I remember looking into some of the data off the USGS (US Geological Survey, which reports big quakes all around the world), and that thing does show that there are a LOT of big quakes happening all across the globe all the time. So, maybe it may not really be quite as simple as looking out for patterns of "there are quakes across the world, and now there are no quakes across the world".

Further compounding matters is that there does seem to be some truth to the notion that for any given quake, there is always the possibility that one small tremor can quickly escalate, building up and forming/triggering a larger and more damaging quake. Having lived through so many of these things, you do learn to notice how this seems to happen sometimes, where a quake may start of as just a gentle vertical jitter (3's), before transitioning into a stronger jitter with some horizontal components (4's), before then building up further into the nasty 3D "jelly bowl phase" (5+). The "jelly bowl" phase ones are the nasty, damaging ones that cause liqufaction, are impossible to stand up in, and which result in buildings collapsing (TBH, why wouldn't they? Different parts of the ground are moving at different speeds, and are displaced by different amounts - pulling some pillars closer, others further apart, while simultaneously tilting things forward/back like a drunk log-thrower struggling to balance a giant totara trunk... in effect the building ends up getting stretched and distorted in various ways, some of which may go well "beyond spec")

Anyways, despite all of these challenges, the CS part of my brain keeps saying: "I'm sure these patterns are ultimately able to be modelled + simulated on a computer". IMO, this is one of those things that maybe, just maybe some of the AI techniques being experimented with may actually be applicable here. Certainly it seems that this domain is quite ripe for investigation...  I do believe that part of the problem may well be that we're just currently modelling the wrong things or using the wrong metrics.

For example, apparently, while there's no good way to map from Richter to PGA, there are multiple ways to map from PGA to Richter or MM, when given prior knowledge about how the ground in those parts reacts in prior quakes... It seems that many of the attempts - at least the obvious attempts made by laypeople who haven't heard of anything other than the Richter stuff - would obviously have been hamstrung by the fact that they were using the Richter metrics instead. Sure, we may not actually have the accelerometers in enough places to actually be able to use PGA metrics reliably. However, perhaps if we used measure like this which are based on actual surface data that can be reliably collected like this, and which say something about the actual effects which are experienced by people on the surface, that maybe just maybe it turns out that it's possible to finally start seeing some effects now that we're looking in the right places with the right tools.

There's also the issue of the extent of the models used. Maybe we're better with localised models (using just the patterns for a particular area) to figure out what's likely to happen next. Or maybe, we need to use "whole earth" models to deal with the interconnected nature of the earth's crust, and the possibility that events on one end can trigger events on the other. Or maybe, it's necessary to use a hybrid: a coarse "global scale" one that contributes some residual influence, which is then combined with the localised model to determine where exactly stresses are most likely.

(Perhaps going in the conspiracy territory here, but something tells me that it's quite possible - as in, yes, it's happening, but just not widepsread - that someone is or has been trying at least some of these things already.  Heck, all the spying stuff that govt's around the world get up to is pretty good proof that all possibilities shouldn't be entirely discounted, even if they are quite far-fetched... Maybe they just haven't found anything yet that works often enough that they think it's safe to announce/start talking about. Or maybe, it's actually stuff that is known already, but just politically incorrect to talk about the existence of, as it has some rather widespread and dire consequences on a wider scale (which could lead to public fear/instability).  I'm aware of a few things that are a bit like that, with some potentially "industry disrupting" issues (at least from a PR perspective, if not managed well, and without some viable solutions in place yet) - pollution, effectiveness of products, etc.)

No comments:

Post a Comment