The Seismic Sonata

Seismic Sonata:

A Musical Replay of the 1994 Northridge, California Earthquake Heard From Albuquerque, New Mexico

A Demonstration of Seismic Signal Perception Using Music

Algorithm Composition By Marty Quinn

The Seismic Sonata Audio Display

Musical Legend and Description

The Seismic Sonata is a musical translation of the 6.7 magnitude earthquake that struck Northridge, California on January 17th, 1994. The music is generated from data recorded at the seismic listening station hundreds of miles away in Albuquerque, NM. Since this station is quite a distance from the epicenter of the quake, and since the different kinds of earthquake waves travel through the earth at different speeds, it allows the listener to hear the various waves hit Albuquerque over the course of less than 10 minutes. The data includes approximately 2 minutes and 23 seconds of background 'noise' before the initial P-waves strike. A few minutes later the more powerful S-waves arrive. The data was converted into music using software developed by Marty Quinn of Design Rhythmics Sonification Research Lab and was funded by the IRIS Consortium for use in their earthquake museum displays.

The data for this project consisted of a text file that contained header information followed by over 36000 numbers that represented the vertical movement of the earth over 1 hour recorded at Albuquerque, NM. The numbers are collected at the rate of twenty per second. The header information identifies the station, the time, and the number of data samples, along with a few other items. The data numbers are arranged as a list of positive and negative numbers, the larger the values either positive or negative, the larger the vertical movement either upward or downward from the original 0 position. Since there is only one station involved, there is no need to represent the header information in the music. Instead, we concentrate solely on representing the vertical movement numbers.

In creating a musical representation of these seismic signals, we held the following goals in mind. We wanted the listener to be able to:

1. Accurately assess the shape of the seismic waves.
2. Perceive the intensity of the seismic signals.
3. Emotionally sense when the earthquake waves are more intense.

Accurate seismic signal representation

The entire report contains numbers that span a wide range of values containing 1 to 6 digits. Long stretches of the report tend to stay within a certain range of numbers. Using a traditional graph of the data, as shown in figure 2, we can see that at the beginning, there is hardly any movement and hence, the numbers are very small. It is difficult, using such a display, to see any shape to the waves. After a few minutes, larger numbers become prevalent and the shape of the waves can begin to be discerned. After a few more minutes, the largest numbers in the data show up and the resulting wave shapes are very clear. After awhile, the numbers subside again into medium intensity values.

The Long View or Coarse Audio Display

Using sonification techniques, we map this view of the data onto a scale of 45 pitches based on a C major scale. Lower notes represent lower values and higher notes represent higher values. In the current sonification these notes are played with an oboe sound. Therefore, at the beginning of the music we would hear a middle note played on oboe that would not change in pitch for quite a while. To avoid sounding the same note over and over again, we simply play the note once and then sustain it. In most cases, we also lower the volume of the note with each successive data value until we reach a predetermined minimum value. This technique helps the listener to focus only on change in the data and can ignore redundant values. Since our brain naturally ignores redundant or superfluous events in our environment, such as the back and forth movement of wind shield wipers, it seemed natural to provide a musical analog to mimic this tendency.

The Fine or Zoomed Audio Display

But what if we also want to see the fine detail in the waveforms even in lower intensity parts of the record? In order to do this, we could provide a visual zoom of a certain region of the data. Figure 3, for instance, zooms in on the beginning of the file. As you can see, there are definite wave shapes even in these low numbers. Using sonification techniques, we can map this view of the data, about 60 numbers or 3 seconds worth, to the sound of a piano. The range of this smaller view of the data is mapped to the same 45 note C major scale. Since we can hear two sounds at once with no problem, we can now hear the long view of the data played by the oboe, and the short or zoom view of the data played by the piano.

The Audio Intensity Legend

There is another problem. Since the range of every 60 numbers in the zoom changes, how does the listener know how to interpret the piano sounds? Are the pitches large or small numbers? In order to solve this problem, we created a musical intensity legend that accompanies the zoom piano part. The intensity is composed of three categories of sounds.

1. Plucked strings and bell pitches descend the scale one note for every digit contained in the highest number for that section of the data. For instance, if the highest number is 345, three notes would be played in sequence, down one step, down one step, down one step, one per second, and repeating until a change in the largest number per zoom section. Each repeat starts on the same note. If the highest number is 232,456 then the listener will hear a series of six plucked sounds. Also, when the numbers are smaller, the plucks will be quieter. When the numbers are larger, the plucks will be louder up to a point.
2. In addition, a low string sound sustains the lowest pitch reached during the plucked sequence. This helps to indicate the heaviness of the numbers.
3. Additionally, when the zoom range changes, a timpani drum is struck to signal to the listener to pay attention: the numbers just got bigger or smaller. The louder the timpani sound, the bigger the numbers.

So, combining these three audio displays, including the coarse, the zoomed and the intensity legend, creates a powerful audio display of seismic data. The zoom adjusts over time to allow the listener to discern the shape of the waveforms during all sections, the legend communicates the concept of intensity, and the long view suggests how the current set of numbers compares overall. The listener can choose at various times to focus on one or the other sounds using common listening filtering skills, such as we use when we listen to the singer in a recording, or focusing on the drum, bass or instrumental solo in a song.