Chapter 6 Tsunami Warning Systems Past, Present, and Future Tsunami warning systems of the past and present are based on seismographic and tide-gage information. Recently it has been realized that tsunami detection and warning might be based on the effects created in the ionosphere by a propagat- ing long wave such as tsunami. In the first three sections the instrumentation, warning systems, and sociological problems will be discussed and in the final three sections the concept of a future tsunami warning service based on atmospheric detection of tsunamis will be included. 6.1 Seismic and Tsunami Instrumentation SEISMICINSTRUMENTATION The basic principle of a seismograph and then some recent developments will be considered. In this discussion, the books by Richter (1958) and Hodgson (1964) are followed closely. Although seismic waves lose much of their energy during propagation, they may still be detected by a sensitive instrument called a seis- mometer. When a recording system is added to the seismometer it becomes a seismograph. Thus, a seismograph is an instrument that records the nature of the earth’s movement caused by an earthquake or any other disturbance. The most general movement consists of translation, rotation, and strain. Our interest is in the translation movement due to an earthquake at a distance, and rotational effects that may be important only close to the epicenter (Bullen 1963). The earliest seismograph appears to have been invented in China in the year 132. This instrument consisted of a ring in which several dragon mouths, each holding a ball, were arranged. When an earthquake occurred, depending on the direction of movement, a particular ball would drop into the mouth of a dragon. By mid-19th century, Italy had seismographs that recorded the time of the earth- quake by making a mark on a rotating chronograph drum. Present seismographs, based on the principle of the pendulum, were developed in Japan around 1880; then in Italy, and later in other European countries. The principle of a pendulum-type seismograph is: in an ideal situation a heavy mass, because of its inertia, would not move at all, while the ground would move following the earthquake. As this cannot be realized in practice, the mass is restrained by a spring or some other restoring force, to return the mass to its original position relative to the ground. The spring should not be strong enough to alter the motion significantly. 28 1 FIG. 6.1. Schematic diagram of a horizontal pendulum, the simplest form of seismometer. (Hodgson 1964) The horizontal pendulum is shown in Fig. 6.1. A large mass, M, is held in position by a wire, AM, and a strut, MB, pivoted at B; A and B are points near the top and bottom, respectively, of a vertical support, AB (Hodgson 1964). By em- bedding this vertical support in a concrete pier in the ground a seismometer may be obtained. In the ideal case of zero friction at B, when an earthquake occurs the pier and the upright support will move in response to the incoming seismic waves; however, the inertia of M will tend to keep it from moving. The differential motion between M and the pier is the earthquake signal at this seismometer. In practice, improvements had to be made to this simple system. Once the pendulum starts to swing because of an earthquake, it will continue to do so, and its motion as time goes on will be unrelated to the actual ground movement. Hence, the pendulum must be stopped after it has recorded the initial ground motion. There are several ways of doing this; a paddle attached to the strut, BM, moves in a pool of oil, which will quickly damp the motion, or alternately a copper vane on the strut moves between the pole pieces of a magnet. The motion of the strut in the vane sets up eddy currents that oppose the motion and thereby bring the pendulum to rest. The motion can be recorded in several different ways: a pen attached to the strut would draw a straight line on a moving paper roll, but when an earthquake occurs this line would zigzag. A system of levers could be used to magnify the motion of the strut from io2 to io5 times so that distant earthquakes could be recorded. A mirror and light arrangement for magnification with recording on photographic paper is an alternative. At present it is common to record electromag- 282 netically by attaching a coil to the strut and making this coil move in a magnetic field, and thereby generate a current, operate a galvanometer on this current, and record on a moving paper roll either by pen and ink or photographically. To conserve paper, it is wound round a rotating drum and the paper moves sideways as it rotates, so when it is removed from the drum, the record appears in the form of a number of parallel lines. Knowledge of arrival time of different waves at the station is necessary to determine the time of the earthquake. This can be obtained by providing a time scale, as marks at every minute or so, on the record. Because these times must be accurate, crystal clocks are used at present. The site for a seismograph has to be selected carefully because extraneous factors such as microseisms might mask the true record. It is advisable to locate the seismograph in hardcore rock rather than in soft ground. To describe the translational movement of the ground due to the earthquake three separate components have to be measured; the east-west component, the north-south component, and the vertical component. Thus, at a given seismographic station a complete set of seismographs consists of two horizontal seismographs and a vertical seismograph. Seismic waves have periods ranging from a fraction of a second to several minutes, and no single instrument can give a uniform response over such a range. Thus, different instruments are needed for different ranges: short-period seismo- graphs are used for the period range of 0.2-2 s, and long-period seismographs for the 15-100 s, intermediate seismographs cover the range of 2-15 s, and ultralong- period instruments for periods over a minute. Near earthquakes, with predominant periods of a fraction of a second, could be recorded by short-period seismographs whereas distant earthquakes, usually with surface waves with periods of more than 2 s, could be recorded through long-period seismographs. Bullen (1963) mentioned that if the station has only one instrument, usually a seismograph with good response over the range 6-8 s is used. The important observatories contain three seismographs to cover three ranges. If the observatory is located in a seismically active region, then it has, in addition to these, a low-sensi- tive, strong-motion seismograph that cannot be knocked out by a strong earthquake locally. / The following three parameters specify a simple seismometer (Richter 1958): the free period, the damping constant, and the static magnification. Another para- meter called dynamic magnification is the amount by which the magnitude of the ground motion is magnified. A seismometer of free period much shorter than the periods of the waves it records is called an accelerometer, whereas in the reverse situation, it is called a displacement meter. In the presence of damping, the free period, 7,is modified to r/j where j is a constant that satisfies the relation h2 + j2 = 1, h being an instrumental constant. To determine 7, the damping part of the instrument is removed and the free oscillations are timed, or the period of a damped oscillation is timed and corrected. Although the instrumental constant, h, could be used to specify the damping, in practice another constant, E, related to h through E = e-nh/i, is used. Thus E is the 283 ratio of two successive swings (in opposite directions) of the motion given by h2 + j2 = 1. The static magnification, V, of the instrument can be understood if T is the period and A is the amplitude of the simple harmonic motion in the ground at the seismograph station. If the pendulum performs an oscillation with period, T, and amplitude, B, then the seismogram gives a period, T, and amplitude, VB. Until the beginning of the 1920s, the important consideration in seismograph design was to lengthen the period of the pendulum so that the dynamical magnificat- ion was nearly uniform. According to the formula T = 2n which gives the period, T, of a simple pendulum of length, L, under the action of gravity, g, a pendulum with a period of 10 s needs a cumbersome length of 25 m. In the Milne-Shaw seismograph, a period of 12 s was achieved by designing the instrument so that only a small part of g is used to control a horizontal bracket pendulum. The restoring force for the inverted pendulum is a heavy mass balanced on a knife edge coupled with springs. Recording is done photographically on smoked paper. A seismograph weighing 22,353 kgm was built in Zurich (Richter 1958). Actually, the tiny torsion seismometers developed by Anderson and Wood in 1922 are much more effective than these cumbersome seismographs. Bilham and King (1970) discussed the recent advances in strainmeters and compared their advantages and disadvantages with other geophysical instruments. They considered the frequency response and amplitude response of strainmeters. Laser interferometers were also included in their discussion. For recent advances in seismic instrumentation, see Plesinger (1970), Vogel (1970), Willmore (1970), Berckhemer (1970), Husebye (1970), Bonjer and Mueller (1970), Husebye and Bungum (1 970), Weichert (1970), and Smith (1975).