Smart seismology

June 1997

RMIT's latest digital seismograph is better, smaller, cheaper and smarter than the average earthquake.

A black drum revolves slowly, a delicate needle tracing lines across its surface... the lines wobble, then a flurry of peaks and dips screams 'earthquake!' Somewhere out there, the earth is moving.

Tucked away in a Melbourne basement, an industrial bench holds a row of these familiar drum seismographs. But the RMIT Seismology Research Centre has moved far beyond the classic instrument, and its network of monitoring equipment is going digital in a big way.

The centre's latest product has "every new bell and whistle we think is essential for a seismograph now or for the next few years," says the centre's director, Gary Gibson. The Kelunji D-series is the fourth generation of seismograph developed and built at the centre, and is designed to cope with every type of seismic monitoring from the occasional natural tremor to the rigorous requirements of mining or exploration seismology.

The centre's monitoring network currently covers Victoria and New South Wales (several overseas countries also use Kelunji equipment for monitoring their own seismology). Seismographs scattered through urban and country areas feed their readings back to the Melbourne basement, where a Macintosh II computer co-ordinates the show. Although it's a few years old, the Mac copes easily with the demands of the network.

The main component of the Kelunji D-series gear is housed in a die-cast aluminium box and weighs around 12 kg. At 34 x 25 x 18 cm it is about half the size of previous models. Everything about it has been improved and streamlined, from its ability to absorb information, through the physical layout of its circuits, to its power management system.

In the field, a Kelunji unit will have a number of remote digitisers connected to it. These sensors take the readings, and the seismograph records them. Older systems used shielded cabling to make the connections, and because of noise problems were limited to distances of a few hundred metres. The D-series components are connected by clear, digital serial links instead, allowing larger arrays of sensors to make more detailed recordings of a region. The serial interface is capable of transferring information at up to 115,200 baud.

"In an array of seismographs you have them in a line," says Mr Gibson. "Looking at a seismic wave crossing that line tells you a lot more than looking at any one unit can. If you're looking at a building you would have one digitiser in the basement, another half-way up, one at the top, and maybe another one out on bedrock nearby as a reference. Each digitiser provides three channels of information and by looking at those twelve channels you can see how the building is moving." Measuring motion along three axes--east-west, north-south and vertical motion--requires triaxial transducers. The D-series uses multiple triaxial transducers for extra oomph. "With the D-series you only need one recorder to get the whole story, whereas in the past you'd need four separate recorders. It's all done using the same computer, the same storage, the same everything else, so it's a negligible extra cost to make it multi-channel."

Previous Kelunji models used in Australia have run on 1 Mb of memory per month for recording seismic activity. The new model has "virtually unlimited" memory capacity for storing multi-channel recordings of vibrations. Data is usually stored on up to four PC cards, either SRAM cards (currently up to 6 Mb each), flash memory cards (up to 64 Mb each) or on PC card hard disk drives (up to 500 Mb).

"If that's not enough, you can plug in a SCSI card and mount a magneto-optic or DAT drive inside the recorder case. And if that's not enough, we can have a PC card connecting to a big external hard disk," says Mr Gibson. "Although we're Macintosh fanatics, to make life easy we made the D-series devices all DOS-compatible. You can plug them into your Macintosh or your DOS computer and just read them as files. There's no translation. It's all menu-driven, and we've got a real-time operating system and multi-tasking, meaning it can do lots of things at the same time."

The D-series has a full PC card protocol that operates modems, ethernet connections and Internet access. "We have FTP and Telnet servers so that you can copy from your recorder in Laos or the Philippines or wherever. If you can't get a direct ethernet connection with the plug-in dongle--if you're out in the bush with a lap-top--you can plug in a telephone to the seismograph and make your connection that way," says Mr Gibson. "By the end of this year each D-series will have an internal World Wide Web server. Running sensor networks anywhere in the world from the RMIT Seismology Research Centre would be no problem."

Telemetry is currently a major expense in running the seismology network. In the future, long-distance communication between seismographs in the field and the Macintosh II at RMIT may be liberated from the current system of leased land-lines. For short ranges, spread spectrum digital radio modems are providing a viable alternative. For long-range communication the centre will make greater use of triggered events relayed over the Internet. Mr Gibson says "we'll be able to set up networks in very remote places, like an active volcano, and have the data continuously coming to a safe distance away. You would just have a digitiser and radio transducer, drop them from a helicopter--quickly--and get out of the place."

Each circuit board in the seismograph has its own 68000 series computer, all connected with serial links rather than with a bus. "You often have problems with a bus: one board has a problem and it drags the others down," says Mr Gibson. "Here there's no bus, all communication is serial, and if you want to test any part of a machine you can plug in a serial line to the relevant circuit board and talk to its computer in the background" without disturbing the seismograph's other functions.

Timing is vital to seismology: an accurately timed and dated series of recordings gives a precise picture of how and where the earth is moving. "In the past we used short-wave radio receiver to get time signals," says Mr Gibson. "It was very tedious making sure your clocks were accurate. The Global Positioning System now is wonderful."

Most seismographs achieve an accuracy of one-tenth to one-half of a second in twenty-four hours. The Kelunji D-series seismograph has a built-in GPS receiver and a socket for a satellite aerial. The aerial itself is about half the size of a credit card, and about 1 cm thick.

"The GPS satellite tells you not only where you are but what time it is. So the GPS receiver gives us the time accurate to 0.00001 second easily--ten microseconds. You just have to plug in the aerial," says Mr Gibson. The GPS circuit board uses more power than the rest of the recorder combined. Unless the unit is connected to mains power or a big solar panel, the GPS would normally be switched off; timing is then done with a temperature compensated crystal correctable to one-thousandth of a second (or better), and calibrated with the GPS once a day or once an hour.

The new Kelunji seismograph features intelligent power management. Mr Gibson says "this sounds very dull, but it's very important because one of the most common problems with seismographs is running out of power. The solar panels can't keep up, or whatever."

The D-series has a 12 V rechargeable battery, and overall uses less power than previous models. An internal regulator for the charger keeps track of voltage and current, and every module in the local system--the digitisers, the PC cards, the GPS--is also monitored.

"When something goes wrong with anything electronic, almost invariably the first hint of a problem is in the power supply: it either uses more current or less," says Mr Gibson. As soon as a fault is detected, the power management system can switch off any module and keep the rest of the set-up running.

The first batch of twenty-five D-series seismographs are "working quite nicely," says Mr Gibson. "Everybody seems to be very happy with them. There will be a few changes, such as locks to hold the circuit boards into place--especially useful when you travel by air."

A D-series seismograph can be built by hand in about one week. "If you compare the D-series with most other equivalent seismographs--not many people build digital seismographs anyway--we have by far the most versatile and probably the simplest at the same time," says Mr Gibson. "We have three different functions on one circuit board, where some other seismographs would have separate boards for each of those functions. And no one can come anywhere near our power supply sophistication," he says. When system designer Vaughan Wesson gave papers in the United States about the power management and PC card systems, the RMIT advances were written up in an electronics trade magazine.

"There are about twenty people working in seismology in Australia," says Gary Gibson. "I would daresay the total number has decreased over the last ten years rather than increased." He suggests two reasons: decreased money and improved equipment. "It's just not economical to run a small seismology group. What we're doing over time is gradually modernising the network, putting these new fancy machines in and automating it. In fact, roughly five people are running a hundred machines, whereas if you go back ten years we would have been running twenty-five machines with five people--and doing less with the data we collected."

Photo of Kelunji system designer Vaughan Wesson by MLRuwoldt