Sensor Workshop at ITP | Reports / Tsunameter

TSUNAMETER SYSTEM

Possible Causes of Tsunamis

Tsunami waves are most often caused by seismic disturbances in the ocean, which abruptly push huge volumes of water out from the fault's center. Underwater landslides and even meteor impacts can set off tsunamis as well. Vulcano eruptions, if in a coastal location, can also contribute to the generation of tsunami waves.

The DART System (Deep Ocean Assessment and Reporting of Tsunamis)

This complicated network of sensors elaborated and applied also by the NOAA (National Oceanic and Atmosferic Administration) includes:

Bottom Pressure Recorder (BPR) including Digiquartz Broadband Depth Sensor, a computer, a data logger. This very well calibrated device records the pressure in deep water using Paroscientific's Digiquartz technology that revolves around crystal quartz resonators, http://www.quartzdyne.com/technology/quartz.htm , a type of material that apparently very well responds to the changes in pressure at depths between 1000 and 8000 meters.
Acoustic link to communicate to floating buoy
Satellite telecommunication (GPS Iridium) http://www.iridium.com/corp/iri_corp-understand.asp
Optional ADCP or CTD (Acoustic Doppler Current Profiler or Conductivity, Temperature and Density Sensor). They can help give a better subsea picture of the current flow and the general state of the water, and this data can help in trying to detect the tsunami when still in deep ocean. And possibly help prevent false alarms which are a very recurrent issue ; many locations at risk of tsunamis were, at least in the past, very shy not only to the technology, but aso to letting out Tsunami warnings and alarms for fear of economical loss of tourism and business. http://www.whoi.edu/instruments/viewInstrument.do?id=819

For a similar system here is a detailed website from an Italian manufacturer, Envirtech http://www.envirtech.org/seafloor_observatory.htm

Tsunami detection algorythm

The tsunami detection algorithm in the software works first by estimating the amplitudes of the pressure fluctuations and then testing these amplitudes against a threshold value. This algorythm is at the heart of the submersible system and basically alarms the buoy any time the threshold is being breeched. The formula shown is only part of the algorythm which we will look at later. Basically the rule of thumb i based my mathematical study on is that the wavelenght of a tsunami in deep ocean is also greater that its depth, while the height of tsunami waves when still at the origin can even be in the range of only a couple of cms.

lambda is the wavelenght (in relation to the depth) and it equals the Period of the wave multiplied by the squaqre root of gravitiational acceleration multiplied by the depth.

Goals

The engineer's design goal before 2004 included an expense goal of under 250.000$. Therefore i wanted to see how far i could go with only a 250$ budget and see until when i was able to retrieve scientifically valuable information from what i had compiled.

My design goals

The original question i faced while reading and researching extensively on the Tsunami analysis and esoteric attempts to detect it when still in deep ocean, was: "Could I, if not replicate, at least mimik the 250.000$ system with the little knowledge i have on oceanography?" I would like to think that, although i am still not an expert, i have gained a great deal of knowledge on the subject now and definitely a good understanding of the device itself.

ITP Alumn and Oceanographer Toshitaka and Engineer William were not only the perfect place to start this research but also the most helpful resources:

My Tsunameter Device

My components of choice:

I chose the components based on 2 major factors, what William had suggested and also my budget restrictions. I spent a lot of time at the beginning of the project in trying to find a submersible pressure sensor, but whether for the lack of datasheets on ebay or the price ($400 to $20.000) or for the size of this type of equipment (1m ca.) I had to opt for another network of sensors. Also, ADCPs, which taking advatange of the doppler effect basically "ping" the surface of the water and based on the refraction calculate the current movement, can go up to 30.000$ or more and have issues in clear water because of the fact that the suface might be at times too transparent to reflect and can therefore return useless data. http://en.wikipedia.org/wiki/Acoustic_Doppler_Current_Profiler

So this is is what i used and what i plan to use in the future:

ADXL330 accelerometer 25$-35$ http://www.sparkfun.com/commerce/product_info.php?products_id=692 To calculate the X,Y,Z movement and especially to watch for the Z axis and identify wave crests.
GPS module from Sparkfun 50$-60$ http://www.sparkfun.com/commerce/product_info.php?products_id=465 To obtain real time readings of UCT time, so that i can calculate the wavelenght and wave interval in time. Plus watch for location, once the device will be anchored somewhere in deep water, and to watch for speed over ground.
Microcontroller 4$-40$ I am using an Arduino, but possibly an ATMEL chip with AD converters might do the same job.
Fish finder/Depth sounder 60$-100$ http://www.humminbird.com/products.asp?id=611 There are all kids of depth sounders and fish finders. The best and cheapest manufacturers are Hummingbird and Garmin. I intend to purchase an interfaceable depth sounder that takes NMEA formats, so that i can enter that data into my algorythm, but so far i have only used the fish finder from Hummingbird, as part of the prototype. Using SONAR, the fish finder, not only finds fish, but also gives me a constant reading of the depth and displays wirelessly the image of the floor bottom.
Bluetooth wireless module (BlueSmirf) 60$ http://www.sparkfun.com/commerce/product_info.php?products_id=582 Used to transfer the data coming from the sensors to my computer using BlueRadios Technology.
Hydrophone 40$-250$ http://www.dolphinear.com/ (here needed for purposes other than scientific measurement, for performance) Originally, I wanted to confront the readings from the accelerometer with the readings from the hydrophone. But not only most things that happen in the water happen at non-audible frequencies, but amplifying the signal we also esponentially amplify noise. So although the hydrophone can add a layer of information to the system, it revealed to be fairly problematic. I at first tried with this hydrophone http://cgi.ebay.com/10-Heavy-Duty-Piezo-Stereo-Hydrophone-Contact-Mic_W0QQitemZ330073801758QQihZ014QQcategoryZ15198QQrdZ1QQssPageNameZWD1VQQcmdZViewItem and an amplifier, but then opted for the DolphinEar which i later used in a performace to control video. More details on the purpose of the hydrophone visit http://www.pmel.noaa.gov/vents/acoustics.html

Currently being added:

Arduino mini instead of regular 30$-40$
Gyro-Tilt Sensor 65$-100$ http://www.sparkfun.com/commerce/product_info.php?products_id=395 To stabilize the Z axis against any change of direction of the vessel.
ZigBee Radios and 100$-200$ To substitute the BlueSmirf.
Pressure Sensor coming soon...200$-400$

Final Device:

for more details on the entire process starting from the first experiments i made with the GPS up until now (almost) check the blog at http://itp.nyu.edu/~bp432/blug/archives/sensors/

Preliminary tests

The first tests were focused on making sure that the container was water-tight, especially when in the prototyping phase where i was still using a serial communication and therefore a USB cable. I also played with stabilizing the device in the water, recently i started using marbles and mostly rice to counterbalance the container.

Test in Greenport, Long Island

HOBO Pendant G Logger

The first few tests, because of the fact that the device had no wires, I used a HOBO Pendant G Logger which is a very well packaged and interfaced accelerometer that needs to be retrieved and connected to a computer. http://www.onsetcomp.com/solutions/products/kits/gkit.php5

It is a waterproof package and comes with a nice graphic software for both Mac and PC that looks like this:

And this is the last time I saw the little toy boat, since it was taken by the undercuccrent and must have logged for about 13 hours as it was programmed to do.

Tsunameter tests

CODES IN ARDUINO:

I had two pieces of code for this project. The first that just serially transmitted the GPS string parsed in order to only send the RMC NMEA setting that includes the time, longitude, latitude, speed over ground, course over ground and date and it also serially communicated the unfiltered readings of the accelerometer. I used the regular serail pins for the Bluetooth while i used software serial for the GPS communication.

The second piece of Arduino code calculated a Simple Average on 5 samples and the Standard Deviation on 5 samples for each X, Y, Z pin.

CODES IN MAX-MSP:

I also played with a couple of patches in Max MSP, one that would simply visualize the data coming in from the accelerometer and the parsed GPS data:

The other Max patch used a OpenGL object to map out X, Y, and Z and showed the movement of the wave:

Problems, Challenges and Discoveries

My computer battery doesn�t last longer than 30 minutes. It was a huge issue when i tested out on the beach and i wasnt aware of that.
One problem was given by the X and Y confusion. I think with a gyro, or ignoring X and Y altogether will fix the issue.
The buoy/container won�t be sensing anything smaller than its diameter. Basic physics law, but underestimated.
The wavelength of a tsunami wave is always greater than the water depth.
I also discovered that there isn't a direct relationship between the magnitude of the earthquake and the tsunami being generated and therefore false alarms can happen when a high magnitude earthquake is detected.
I made some useless tests.I really wanted to make something useful out of the data coming from the accelerometer so I searched for conversion formulas with no luck. Then i decided to run a test (probably obvious to most) to see if by controlling the water depth the accelerometer readings would change accordingly. The answer is no, at least from this specific test. And I am still looking for a formula that can help complete the following math, by giving me some kind of conversion from the voltage value to a unit of measurement that i can actually use and understand.
Math

I calculated th distance using this formula, since it was the most accessible to me and i had the data.

Phase velocity, as all other following formulas, is a good example of how the situation is mathematically different if in deep water versus shallow water. In fact, at shore waves, are mostly wind generated and non-dsipersive, while deep water waves take into consideration the dispersion relation. http://oceanworld.tamu.edu/students/waves/waves2.htm

This formula was necessary to derive these other formulas from:

And this is the dispersion relation formula that seems to be at the heart of Pierson and Moskowitz experiments including their calculations made with accelerometers mounted inside of buoys.

The formula reads that the wave frequency in radians per second squared equals gravity multiplied by the wave number multiplied by the tangent of the wave number multiplied by the depth.

The formula changes according to the situation as in deep water the depth is always greater than one fourth of the wavelenght. And therefore the tangent of (kd) equals 1. The formula for shallow water situations changes because of the depth being smaller than 1/11th of the wavelenght and therefore the tangent of (kd) equals kd.

Next Steps

Construct a structure similar to the DART system with a buoy or using multiple floating buoys.
Remote control boat to drive further at coast and drive itself back.
Submerge the device to see if this way it ignores small wind generated waves.
William suggested in his most recent email a very different approach that i would like to try:

Accelerometers are mems devices, and as such are capable of sensing very very fine movements, like sound waves in water. You can immerse the accels in a substance like baby oil, or ISOPAR (very toxic), and immerse them in water.� You will need lead, b/c the oils are lighter than water. Then experiment with what freqs the accels are keyed up to; they will have a sweet spot.� Your data will not look great b/c the ADXL does quite a bit of filtering, BUT, you may be able to squeeze some meaningful wave pressure and velocity info.� Especially if you have more than one in the water, and know which way they are pointing.� If you have that, you can do a waveform modeling on the cheap.� Might not be all that useful, but then might.