Seismology --- One of my other interests.....

[davenn]

One of my many major interests away from electronics (sort of) is Geology, varying from rock, mineral and fossil collecting through to volcanics and earthquakes. This interest culminated in doing a BSc in geology at my local university during the '90's.
I love the adrenaline rush from the feeling of the ground moving beneath my feet.
Unfortunately it can go way beyone that to major damage and deaths, as has been seen over the years around the world. Closest to home for me is all the activity in the Christchurch region of the South Island of New Zealand. My 2 early 20 yr old son and daughter and 3 yr old granddaughter live there. Its been quite an ordeal for them over the last 10 months with many large events rattling their nerves.

I got into seismology (the study of earthquakes) in 1974, when a M5.0 shook my home area of Dunedin city. It left a lifelong desire to find out what caused them etc. Over the last 40+ years I have felt many, many quakes ranging from a tiny M3.0 to a big M7.6

My dream of actually recording quakes finally came to pass in 1989, when I was able to set up an analog recording station. This page on my www site gives some background
http://www.sydneystormcity.com/g_phones.htm
The analog system consisted of a commercial geophone seismometer, a 3 stage, hi gain op-amp preamplifier and onto a rotating drum recorder. The drum rotates on a wormgear, once every 15 minutes and on the appropriate width sheet of paper wrapped around the drum gives 24hrs of recording. YUP, every 24hrs the paper has to be changed!

The quake you can see on the drum was a M5.5 from the SW of the Sth Is of NZ.

In ~ 1992, I met up (online) with a small group of guys in California that had a Public Seismic Net, PSN, running, The unofficial leader of the Group, Larry Cochrane, is a programming and electronics wiz. He had designed an analog to digital converter system, software to run that and then other software to analyse the recorded quake files.
(seriously... This guy is brilliant !! its relatively rare to find some one who is really good at both software and hardware) His site .... http://psn.quake.net/

By 1996 I had a digital system up and running concurrently with the old analog drum recorder. I was recording 3 digital channels and 1 analog channel. This was in operation through to the end of 1999 when I left NZ and came over to Australia.

Leap foreward 11 years to today.... Over the last couple of months I have finally got my system up and running again. At the moment I have just a single channel being recorded on the digital system. I am currently building a long period seismometer to compliment the short period geophone.
SP geophone natural period = 8Hz great for recording local and regional quakes out to 400km
LP seismometer natural period = 10 sec (0.1Hz) will record a M6.0 and up anywhere in the world

look here for a pic of what I am reproducing.... http://jclahr.com/science/psn/chapman/2008 lehman/index.html click on the image for a full size pic.

This unit has the well known nickname of a "Lehman" seismometer. Jim Lehman, another American, pioneered this style of sensor during the 1970's and it even made it into the Scientific American magazine in July, 1979 when it was presented by another person.
It is amazing how many times this basic design has been taken and revamped by many people around the world.

Ok that will do for "The Story Pt1"
more to follow

cheers
Dave

 

[(*steve*)]

Gee, you're into everything aren't you?

 

[davenn]

Gee, you're into everything aren't you?

haha and I havent even started on my astronomy activities

The crazy thing is, Steve, I really cant get away from electronics!! its deeply woven, in some form. throughout all my interests


D

 

[daddles]

That Lehman design looks interesting -- I have virtually all the stuff sitting in my shop to make such a thing (but I'd have to buy some magnet wires to make the coils) -- even a chunk of 3" diameter brass bar stock about a foot long! I like the design of the washer at the top left to hold the wire -- that's something I'll stick in my notebook, as it would be easy to make when needed. How are the two washers connected?

It looks like the main features are: the knife edge for the pivot (the photo appears to use a hardened ball as a pivot, which would be easier to make as all you'd need would be a center drill), the magnetic damping, and a coil and a magnet to generate the signal. How about some details on the coil, magnet's field, and some of the electronics you're planning on using?

Paper can be a PITA -- I used chart recorders a lot in the 70's and 80's and always wound up with lots of rolls of paper. OTOH, it was nice to have that permanent record. Last time I used one was in the late 80's when I borrowed a recorder from work to monitor the current from my well pump -- I used the data to prove to myself that our pump was running because of an underground leak between the well and the house. I had to hire a plumber to run a new pipe under the driveway and it was around a $1000 expense -- and I was gratified to see that the pump running for no apparent reason stopped after the pipe was replaced. It made me trust my troubleshooting skills a bit more.

If you're willing, I'd love to hear a short blurb from you on the analysis of the signals from the seismometer. I assume the seismometer primarily picks up one component of the motion (e.g., a shear wave). Or does it? I'd also imagine the relationship between the different parts of the signal let you determine a distance to the fault because of the (known) dispersion -- do you then need one or two other stations' results to pinpoint the location?

Yeah, I was into astronomy in the early 80's -- and that sucked up a bunch of hobby time...

 

[davenn]

SNIP
I like the design of the washer at the top left to hold the wire -- that's something I'll stick in my notebook, as it would be easy to make when needed. How are the two washers connected?

Not sure how they are connected..... I dont really see a need for 2 washers anyway
I will doing a slightly different arrangement

It looks like the main features are: the knife edge for the pivot (the photo appears to use a hardened ball as a pivot, which would be easier to make as all you'd need would be a center drill), the magnetic damping, and a coil and a magnet to generate the signal. How about some details on the coil, magnet's field, and some of the electronics you're planning on using?

I have sliced open some large ball bearings and removed some of the steel balls
The bottom pivot point will be 2 balls against each other.

One thing I can be sure of ... the project WONT be a work of art I have extremely limited workshop facilities and with my job coming to an end, the limited facilities I do have are going to disappear. I am looking more at functionality rather than an engineering delight

I will cover electronics etc in forthcoming posts

Paper can be a PITA -- I used chart recorders a lot in the 70's and 80's and always wound up with lots of rolls of paper. OTOH, it was nice to have that permanent record. Last time I used one was in the late 80's when I borrowed a recorder from work to monitor the current from my well pump -- I used the data to prove to myself that our pump was running because of an underground leak between the well and the house. I had to hire a plumber to run a new pipe under the driveway and it was around a $1000 expense -- and I was gratified to see that the pump running for no apparent reason stopped after the pipe was replaced. It made me trust my troubleshooting skills a bit more.

If you're willing, I'd love to hear a short blurb from you on the analysis of the signals from the seismometer. I assume the seismometer primarily picks up one component of the motion (e.g., a shear wave). Or does it? I'd also imagine the relationship between the different parts of the signal let you determine a distance to the fault because of the (known) dispersion -- do you then need one or two other stations' results to pinpoint the location?
Yeah, I was into astronomy in the early 80's -- and that sucked up a bunch of hobby time...

Seismogram analysis etc I will also cover in posts to come. Will give everyone a quick lesson in quakes
but just as a start. if you look at the seismogram on my drum recording above
you can clearly see the arrival of the first 2 sets of waves
the "P" wave firstly it peaks then dies off then the "S" waves arrive and has the needle hitting the stops for a period of time before they too die off.

c heers
Dave

 

[davenn]

Seismology and recording Earthquakes Pt2

OK as promised we will look into some of the finer details.

A couple of posts will deal with quakes themselves, Later ones will look at some of the electronics that get used.

Earthquakes occur on faultlines.. WOW what a revelation They may be very deep in the earth's crust or very shallow, often extending right to the surface

Plate Boundaries
The faultlines are fractures, places of weakness, where the built up stresses have been able to overcome the strength of the rock. These faults can be very small may be only a couple of metres long and deep or they can be many 1000's of km long and extend from the surface and down as deep as 700km. The very large and deep ones are the ones that generally form the boundaries between the tectonic plates
Earthquakes along these plate boundaries outline their locations.....

2 very well known plate boundaries that occur on land are the Alpine Fault in the South Island of New Zealand and the San Andreas Fault that runs 3/4's of the length of California, USA along these 2 plates huge collisions are occurring.
The Alpine Fault -- The Indo-Australian Plate is butting against the Pacific Plate
The San Andreas Fault -- Its the Pacific Plate and the North American Plate

Movement on a fault
There are 3 main types of motion on a fault, as shown below.
but only occassionally are any of these motions pure. That is many times they are a mixture of 2 of those motions in varying amounts....

In pic 3 of the above image, the transform/transcurrent motion is more commonly known as "strike-slip" motion. The San Andreas Fault is almost pure strike-slip. On the other hand the Apline Fault is a mix of Strike-slip and compressional (thrust) faulting.. for every 3 metres of horizontal movement there is 1 metre of vertical movement.
The reason we get quakes is because the rock surfaces on either side of the fault are not contineously and gently slipping past each other. Rather they are held there under compression and over the many years the stresses build up along that boundary. Note that out from the boundary the ground/rock is moving, its just locked together at the boundary. Lets look at a chunk of the South Island of NZ and the Alpine Fault.....

All of the eastern side of the South Island is moving SW at a rate of 38mm/yr. But that motion isnt being taken up along the fault The average return time for major events on the Alpine Fault is ~ 250-300 years. So lets say 275 yrs x .038metres = 10.45 metres of accumulated motion. That motion will be released along some 300km of the faultline.
Its unlikely to rupture for a 10 metre offset along the whole length of the rupture. There may be sections of up to 12 metres of offset, other sections, particularly near the ends, may only offset a couple of metres.

Dave

 

[davenn]

Earthquakes produce 3 main types of seismic waves that propagate out from the rupture area on a faultline

The P (Primary) wave... this is a compressional wave like a sound wave and it travels through the earth's crust at ~ 7 km / sec and up to 13km/sec through the denser mantle. Their freq ranges in the 1 - 10Hz
2) The S (secondary) wave... This is a shear wave, that is ... the particle motion in the ground is perpendicular to the direction of travel of the wave. They have speeds in the crus of ~ 5-6 km / sec and again freq's in the 1 - 10Hz range
3) The Surface waves -- which are a mix of Love and Rayleigh waves. They travel much slower at ~ 2 - 3 km / sec their freq's range for minutes to seconds / cycle
I will let you do some googling / wiki searching if you want to go in depth further

The new seismometer I'm currently building will have a natural period of 10 sec. thats 0.1 Hz

The difference in velocities between the P and S waves is very helpful as it gives a way to determine distance from the sensor to the earthquake....
eg. a time difference of 25 seconds between the arrival of the P and S waves x 8.5 = 212km to the quake. The 8.5 is a "rule of thumb" to give a quick estimate of distance to the epicentre when measuring off the analog paper drum recorder. With a digital system there are incorporated "data tables" that take into account rock type and assoc. velocities of the region between the quake and the sensor and a more accurate distance measurement is possible.
ok lets look at an actual seismogram of a quake I recorded back in the 1990's .....

You can see the clear arrivals of the P and S waves. The surface waves, which are small for this M4.8 quake are lost in the tail of the S waves. Surface waves really stand out of big distant events where they get separated out in time from the S waves.

ohhh speaking of the quake location....
You may well be aware of the term Epicentre. This is the point on the earths' surface directly above the focal point of the quake on the fault. Now for a small quake/rupture this has merit and the event can be pinpointed relatively well if there is no surface rupture.
But talking about a focal point of a major event that has just ruptured 100 or so km's of faultline, the definition of a point source is somewhat blurred. Seismologists would then refer to the region on the fault that has the largest rupture offsets as being the focus or starting point of the event.
A fault can rupture in 2 main ways.
Unilateral Rupture -- the rupture starts at one end of the fault and "unzips" in one direction, and..
Bilateral Rupture -- the rupture starts roughly in the middle of the fault line and unzips in both directions.

Depending on which happens can have a huge effect on damage caused.
For a given length of faultline , lets say 100km. A bilateral rupture is going to take 1/2 the time to occur than if it unilaterally ruptured. Say 30 seconds of severe shaking compared to 60 sec of severe shaking.
Buildings that may have handled 30 sec of shaking may totally collapse with 60 secs of shaking. Liquifaction ( ground basically turning to quicksand and expelling water) will occur with strong shaking over 40 sec. This occurred extensively in the Christchurch quakes of sept '10, feb and jun '11

Before I finish this posting, some comments on quake magnitude.

Richter Magnitude ML, aka Local Magnitude -- in seismology circles its rarely used these days but the media just cant shake loose and "get with the program"
Named after Charles Richter. The magnitude scale was specifically for quakes in the Californian region for the known rock types there and for quakes within ~ 150km from the recorder

Body Wave Magnitude - Mb This used the maximum amplitude of either the P wave or the S wave from the centre line of the seismogram. This was then plotted on a nonogram to measure the magnitude of the quake.... as was the ML of Richters'

I have drawn an example line on the gram An S - P of 25 sec and a max amplitude of the trace from the centreline of 20mm. This approximates to a ML/Mb magnitude of 5.0

Next we have the Surface Wave Magnitude - MS this was used for the big events where the seismogram was totally overloaded and true amplitudes of the P and S waves could not be determined.

In the last ~ 15 yrs a new magnitude indicator cas come into use and is almost exclusively used these days to describe the magnitudes of all quakes.
It is the Moment Magnitude - Mw The magnitude attributed to a given event is based on the total energy released by the quake. It takes into account the total surface area of the faultline that ruptured.

Regardless of the system used, the magnitude scale is logarithmic ie. a M 5 quake is 10 times larger than a M4, a M6 is 100 x larger than a M4 etc etc

ok that will do for this post

Dave

 

[poor mystic]

I have become interested in tectonics too, having spent a few years in Christchurch in the 90's.
I found an excellent resource:
http://www.google.com/imgres?imgurl...=1&ndsp=6&ved=1t:429,r:0,s:0&biw=1024&bih=490

I have a question - how much (apparent) movement can be expected in a seismograph pendulum for a given quake?

 

[daddles]

I would imagine the propagation of earthquake waves has been intensively studied over the last century or two. Do the elastic properties of the crust pretty much average out so that the earth as a whole can be reasonably modeled as a collection of layers of isotropic material? Or is there enough heterogeneity in the different types of rocks to require sophisticated databases to interpret the seismograms (you implied the latter with respect to digital seismometers)? Do crystal anisotropies enter the picture through e.g. non-diagonal stress-strain tensors that need to be used to accurately locate things? If so, then there must be some sophisticated computing going on behind the scenes to turn the seismograms into accurately-located fault data.

 

[daddles]

Oh, and I wanted to mention that I lived in the California Bay Area for 30 years and never felt an earthquake until the last few years. In the 70's I was living in Mountain View (working in Silicon Valley not far from the San Andreas fault) and was napping one day on my bed. I was awakened by shaking (it felt like someone was shaking me awake) and a sound like someone had hit the building with a sledge hammer. That was an earthquake that was around magnitude 5 and was ho-hum to most people. But it really startled me and woke me up to the earthquake risk -- I moved out of California less than a year later. While my real reason for moving was that I was sick of the congestion and house prices, the earthquake risk also was a factor in my decision to move -- especially when I tried to imagine an event that was 100 to 1000 times more powerful than the one I had experienced.

 

[davenn]

I have become interested in tectonics too, having spent a few years in Christchurch in the 90's.
I found an excellent resource:
http://www.google.com/imgres?imgurl...=1&ndsp=6&ved=1t:429,r:0,s:0&biw=1024&bih=490

I have a question - how much (apparent) movement can be expected in a seismograph pendulum for a given quake?

Thats a nice link, I havent been to before, thanks. Tho I did have the ful globe pic on the puter somewhere. just didnt have the others images that looked at the various regions.


Apparent movement ... as in if you could see it actually move ? not likely unless it was a significant local event. Even for the huge distant events its movement would only be a few micrometres. The system relys on the really hi gain of the pre-amp system, anything up to 1000. My short period sensor pre-amp channel is only running at ~ 500 gain as the local manmade noise is overwelming at ~ 10Hz (1 - 20Hz) that the unit responds to.

Dave

 

[poor mystic]

A seismograph pendulum must remain still while the earth moves - movement in the pendulum is 'apparent' rather than 'real'.
I wanted to know whether light interference might be a practicable means of seismometry.

 

[davenn]

I would imagine the propagation of earthquake waves has been intensively studied over the last century or two. Do the elastic properties of the crust pretty much average out so that the earth as a whole can be reasonably modeled as a collection of layers of isotropic material? Or is there enough heterogeneity in the different types of rocks to require sophisticated databases to interpret the seismograms (you implied the latter with respect to digital seismometers)? Do crystal anisotropies enter the picture through e.g. non-diagonal stress-strain tensors that need to be used to accurately locate things? If so, then there must be some sophisticated computing going on behind the scenes to turn the seismograms into accurately-located fault data.

dang lost the long reply, and the short reply... here's the real short reply

there are tables called the Jeffreys and Bullen Seismological Tables these give expected
times of arrivals and time differences of P and S waves for quakes at various depths.
a sample of the tables for S-P time differences is here....

Because of the large number of stations recording events, variations in arrival times can be averaged out and it would be obvious if there is a significant variation from one particular station.
A bit of background on J-B Tables and their use can be found here

cheers
Dave

 

[davenn]

A seismograph pendulum must remain still while the earth moves - movement in the pendulum is 'apparent' rather than 'real'.
I wanted to know whether light interference might be a practicable means of seismometry.

yes essentially the pendulum remains still and its the motion of the frame of the seismometer relative to the pendulum that is being recorded. Of course in practice that doesnt really occur as the pendulum is attached to the frame by the pivot points and not being isolated, results in the pendulum starting to move.
With an undampened pendulum, it will stay swaying after the ground and frame has stopped moving and left to its own devices will sway at its natural period, until the friction of the pivots etc cause it to finally stop oscillating.
Therefore we dampen the pendulum so that its natural oscillation is stopped within several cycles. Otherwise you are just recording the oscillations of the pendulum and not the motion of the ground. This is usually achieved with a copper or aluminium plate between a couple of magnets. Sets up eddy currents and a magnetic field within the copper/aluminium that oppose the other magnetic field..... you get the picture

in days gone bye, both commercial and homebrew seismometers used a bar of metal in an oil "bath" for dampening. This had real problems, apart from being messy, the viscosity of the oil changed with temperature and its age and therefore the dampening factor also changed. Its been pretty much done away with these days.

Elaborate on your light intereference thoughts a bit

cheers
Dave

 

[poor mystic]

There is a technique involving a monochromatic (not laser) source and a pair of Youngs Double Slit interferometers which allows measurement of changes in distance down to half a wavelength or about 300 nm - 350 nm for red light.
The attachment below (which has been submitted to this forum in a previous thread) depicts a closely-related device.
Possibly, interferometric techniques may have a wide range of possible applications in seismometry. For instance high power monochrome sources at the top of tall buildings could be interferometrically assessed and, with many sets of simultaneous measurements,from many buildings, local resonances could be cancelled and a signal obtained.

 

[davenn]

ahhh ok. interesting

it would be interesting to see howthat numeric output could be read as a changing data bit count ... ie. replicate what the current varying voltage is doing. If so the hi gain pre-amp and A to D system could be eliminated

Dave

 

[poor mystic]

Do you see how the idea works? Because the interference detectors are out-of-phase, there is a definite order in which they must be excited. So if A and B represent signals being detected by the 2 detectors, then the sequence ABABABABABAB represents steady movement in 1 direction, while ABABABBABABA contains a reversal.
Of course you can expect missed counts with this sort of thing, so that there must be errors of both magnitude and sense. However the occurrence of missed counts can be reduced by including not 2 but several detectors.
I originally wanted to use this technique for a microphone but a thought experiment showed that even blue light would not be capable of good enough resolution - I wanted my light microphone to be better than the other microphones, not mediochre, nor insensitive. However with seismometry new possibilities appear. The pendulum can swing a much longer lever than the effective pendulum length for instance.
Do you like the aircraft warning lights on the ANZ tower for a top-of-building monochromatic source?

 

[poor mystic]

Quite probably there's some good reason things aren't done this way, but the attachment shows what I have in mind for a pendulum seismometer. The detector is based on the microphone shown in my earlier attachment on this thread.

 

[davenn]

what would you be actually getting out of the interference detector tho ?
a varying voltage? a - + 1024 or what ever bit count ? something else ?
just trying to understand what the software is going to be looking at

its just a little out of my field.

current system is producing a varying voltage and is getting digitized with a 16bit A to D resulting in a > - + 32000 bit count (some 65000 total count)

Dave

 

[daddles]

poor mystic, it sounds like a reasonable approach. However, interferometers have been used for over a century and if there was a good reason to use one in a seismometer, I'd imagine some academic somewhere with unlimited grad student slaves would have already built and published something. From davenn's data, it sounds like the induced voltage method is detecting movement on the order of 1 um; that's pretty sensitive and not a whole lot worse than an interferometric measurement. And I'd bet it could be improved by e.g. using a differential approach like an LVDT.