A typical seismological network contains a number of individual recording stations, controlled from a substantial central office with library, records storage, workshop and laboratories. Very often one of the stations of the network will be located at the headquarters institution, but there is no guarantee that the requirement for quiet operation will permit the installation of the best instrumentation on the headquarters site. For this reason, the station descriptions which follow include only the basic instrumentation, without involving assumptions as to where the necessary support facilities will be provided.
The seismograph is the primary sensing instrument which provides the seismologist with almost all the data of his profession. Until comparatively recently, most seismographs required protection from dust, moisture, and the normal fluctuations of atmospheric temperature, and many of them still use photographic recorders which have to be operated in the dark. In consequence, the conventional vault had to provide a closely controlled laboratory environment which the seismologist entered for the daily servicing of records, and for less frequent but more elaborate exercises in the calibration and adjustment of his instruments.
The provision of this protected environment for the seismographs is quite expensive, and tends to produce a pattern of operation in which the vault is regarded as a long-term investment, and is built on a site chosen for convenience of access from other offices and laboratories of the parent institution. Apart from the direct effect of installation cost on the total scale of the work that any given institution can afford to undertake, the quality of work is depressed if the sites chosen for convenience of operation are unsuitable in other respects and by the fact that the large, fixed investment tends to freeze the entire pattern of organization in its initial form.
The development of seismographs, particularly those intended for the detection of short-period vibrations of the earth, has led to simultaneous improvements in sensitivity and robustness. The increased sensitivity can be utilized only if the instruments are moved away from traffic and other sources of cultural noise. The increased robustness facilitates such dispersal.
In stations designed for the highest performance, the search for isolation and a foundation of hard rock has become more intensive. Some of the best enclosures take the form of a narrow gallery driven into the rock, with lateral branches to house the seismometers. The instruments may be installed directly on flat surfaces of hard rock, or on simple cement plaques if the rock is not hard enough.
In order to supplement the recordings of the first-order installations, many institutions have set up short-period stations away from their main recording base, which are used largely for the short-range observations of local events. Because of their relatively low cost and high performance over a limited range of periods, these 'second order' stations have become established as important elements in the framework of conventional seismology.
The rapid development of new classes of instruments that has taken place largely since 1960 has further extended the range of practical operation. Mobility has increased to the extent that equipment housed in small caravans or even in weatherproof boxes can match the performance of the finest observatory equipment of a decade ago. At the same time sensitivity has been developed so that the instrumental threshold is far below the natural noise level of the quietest sites. The most recent improvements in detection power are being obtained, therefore, either by installing seismometers in deep boreholes, or by laying out extensive arrays of instruments, some of which extend over tens or even hundreds of kilometers of territory. The high cost of these new and very advanced installations is limiting their numbers, but their effectiveness over long distances is giving them an important place in modern seismology.
We now proceed to a more detailed discussion of the main types of station as follows:
An array consists of a number of seismometers spaced out on the ground, and connected by cable or radio links to a central recording system. The overall performance of such a system depends on the shape and size of the array, on the number of elements in it, and on the methods that are used in processing the data. Basically, however, the object is to increase the sensitivity to a particular seismic signal in comparison with other signals or random noise, and thereby, in the most fully developed installations, to produce a 'telescope' which can look into the earth in a specified direction.
In the simplest type of array, the seismometers are distributed over an area which extends over a fairly small fraction of a wavelength of the seismic signal that the system is designed to detect. The outputs of all the seismometers are added together, and the sum is recorded on a single information channel. Under such circumstances, the desired signal arrives approximately in phase at all the detectors, and the output is approximately the arithmetic sum of the individual elements. In contrast to the coherent signals, noise which arises from small incoherent sources near the detectors will produce a sum which is more nearly proportional to SQRT(n), where n is the number of elements. The net result of this difference in summation properties is that the signal-noise ratio increases approximately in proportion to SQRT(n).
If the seismometers are evenly spaced around the circumference of a circle, and we consider the response to a sinusoidal train of seismic waves, it is found that the sum of the individual outputs has the same frequency and phase as that of a single seismometer at the centre point. The amplitude of the summed output depends only on the wavelength of the incoming signal, being independent of azimuth. By subtracting an appropriate fraction of the ring-output sum from the output of a central element, it is possible to obtain a null in the total output for a particular signal wavelength. By using a number of concentric rings it becomes possible to construct a band-rejection wavelength filter, and by passing the output of such a system through a frequency filter we can produce a system which rejects a band of wave velocities.
The commonest example of such a system is an array tuned to accept P and to reject Rayleigh waves. The SQRT(n) discrimination against random noise is maintained, so the system discriminates effectively against much of the noise from both distant and local sources. Arrays that operate on this principle can provide a detection capability down to about m = 4.25 on the unified teleseismic scale, for all sources within a distance of 90° from the station.
The principle of wavelength filtering can be extended to provide directional resolution if the outputs of the different channels are recorded separately, or if appropriate on-line processing facilities are provided. The procedure is to introduce time lags during final processing which compensate for the propagation time of the incoming wave, and then to add the outputs together. In this way, the amplitude of the chosen wave is enhanced in comparison with that of random noise, or with that of other waves for which the wavelength or azimuth of approach differs from that of the selected signal.
Further enhancement of the desired signal is obtained by squaring the total sum, or by summing two subsets of the data and multiplying the outputs together. In the resulting product the signal common to all elements of the array is rectified, and may be displayed in the form of a smoothed envelope, whereas signals not common to all elements produce oscillatory components which tend to disappear on smoothing.
Practical array construction has tended to polarise between the crossed linear patterns, which have been pioneered by the UK Atomic Energy Authority, and filled two-dimensional patterns, developed in the United States. Directional resolution has been pushed below one degree in azimuth and in angle of emergence, and detection level for distant events is below M = 4. An example of signal enhancement from a crossed linear array is shown in Fig. 1.1.3.
In recent years, array processing has been developed to encompass many advanced techniques of signal processing. Modern practice is well reviewed in the report of the NATO Advanced Study Institute on 'Exploitation of seismograph networks' (Beauchamp, 1975).
The first-order conventional stations provide the main coverage of the earth's surface for the observation of long-period and shortperiod earthquake waves over teleseismic distances. As the sources of the seismic waves are distributed over the whole earth, the interest in the data is also world-wide. The first-order station is not primarily a research tool of the institution which operates it. Instead, it is an element in the world network, and should be operated for the benefit of seismology as a whole.
The essentially international character of the first-order stations has been emphasised by many ofthe bodies which have considered them, which include the Committee for the International Seismological Summary, the UNESCO Missions, the Committee for the Standardisation of Seismographs and Seismograms, and the Intergovernmental Meeting on Seismology and Earthquake Engineering. The international exchange of data can only be effective if the data themselves are in standard form, and the requirement for standardising both equipment and operation provides the main objective in the production of this Manual.
The consensus of opinion which we are seeking to express may briefly be summarised as follows:
Coverage The stations of the first-order network should be within 1000 km of each other in all parts of the world. Europe, most of Asia and North America are already covered to an appropriate density, although some stations are on poor sites and others need improved instrumentation. The areas which have much less than optimum cover are, most notably, the oceans, followed on land by South America and Africa.
Standardization First-order stations should be equipped to record three components (N-S, E-W, vertical) of short-period and three components of long-period earth motion. Equipment, operation and reporting procedure should conform with world standards as described in this Manual.
Dynamic range First-order stations should provide a continuous record of all seismic signals that are observable above the threshold set by local noise conditions. In highly seismic areas, the traces of the most sensitive instruments are liable to be obliterated by events close to the station, so additional equipment for recording at lower sensitivity should be provided. In stations where destructive earthquakes are to be expected, the range of intensity may require as many as three levels of instrumentation, and the high frequency content of the closest events will require a high recording speed. In such cases, accelerographs and seismoscopes may be appropriate.
The United States Geological Survey is now in the process of building up a network of Seismic Research Observatories (SROs) which use borehole emplacements to reduce the noise level of sophisticated 3-component seismographs to far below that which has been accepted as normal in conventional instrumentation. Digital recording with computer controlled level selection enables a signal-noise ratio of 126 dB to be covered, whilst long-period and short-period analogue outputs are provided by D-A conversion. Extended descriptions of the observatories and their performance are given by Peterson and others (July, 1976) and by Peterson and Orsini (August, 1976). The instrumentation is described in detail by Unitech (1974).
The flexibility offered, at comparatively modest cost, by multi-track analogue or digital recording on magnetic tape has now enabled the total recording requirements (six or more channels of seismic data) of a first-order station to be accommodated on a single tape recorder, and this is having a profound effect on observatory practice. The darkened recording room is replaced by a much smaller allocation of bench or rack space in a more ordinary room, and the photographic darkroom is replaced by an electronic processing laboratory for the production of visible records.
Second-order stations are usually established for short range observation, and for this purpose are frequently grouped into regional networks, reporting to a single interpretation center. The need to obtain optimum performance over a wide range of local conditions, and the preponderance of regional interest over world interest in the data, both militate against the acceptance of the high degree of standardization that is needed for the first-order network. Nevertheless, the importance of international cooperation in the observation of small events is increasing as a result of the steady improvement in the performance of arrays and first-order stations. In consequence, data which would have been considered as purely of local interest a few years ago may now make an important contribution to some world-wide pattern of observations. Record format and operating procedure for second-order stations should therefore follow first-order standards as closely as possible, even if variations in the performance of the sensing instruments must be allowed. As in the case of first-order stations, additional dynamic range is required in the most highly seismic areas.
The erection of temporary or mobile stations to second order standards has been quite common practice for the last fifteen or twenty years, and it is a procedure which has obvious advantages in the investigation of crustal structure and the observation of noise, microearthquakes and aftershock sequences. The progress of the last few years, however, has removed most of the limitations on mobility, so that long-period seismographs and even extensive arrays can now operate using transportable recording equipment.
The advantages of mobility extend beyond the field of strictly short-term operation. With a mobile system, sites can be tested in succession, and equipment left in position as soon as a good one is found. After a sufficient number of individually satisfactory sites has been discovered and occupied, it may take some years to discover whether the chosen pattern of recording sites bears an optimum relationship to the pattern of sources. If the equipment is transportable, the recording pattern can be adjusted at any time.
Mobility of equipment does not exclude an,y of the advantages of permanent establishment. If the extra amenities of a permanent building are shown to be needed on a particular recording site, they can be constructed in the light of operating experience with far fewer uncertainties than if the decision to build had been taken in advance. The important conclusion is that potential operating agencies should consider the relative advantages of portable and fixed-station equipment in any particular role. Manufacturers should consider the enhanced range of potential use which is open to mobile equipment which can be set up to conform with the international standards for first-order stations.
We are now left with the class of stations which occupies the advancing 'frontier region' of seismological technique, such as ocean-bottom stations, borehole installations, and unmanned stations. It is, of course, inappropriate to attempt to standardize new techniques within the framework of old ones, or to expect that research workers entering a new area will necessarily have either the time or resources to conform to the routine of international reporting in addition to making the selective observations which their own projects require. We can therefore do no more than draw the attention of workers in these new branches of seismology to the standards that have been set up to facilitate world-wide interchange of data, and to the services that the international organizations can provide. In this way, it is hoped that as much compatibility as possible will be maintained between new and existing types of equipment.
In choosing the site for a new station we must fulfil several conditions according to the type of station, and the demands are continually increasing with the steady development of seismometry. The following factors are important:
Remoteness from local disturbances (traffic, heavy machinery, wind action on buildings and trees, large lakes, waterfalls) .
Accessibility for personnel and power supply.
Stable underground Hard bedrock, preferably crystalline, within reach of the surface.
Low relief Broken country scatters seismic waves, and passes through mountain ranges can act as channels for high winds. When stations have to be built in rough country, they should be sited as low down in the valleys as other conditions permit.
Obviously, the first two items are not usually compatible, and the selection of the site should therefore be governed by the aim of the station and by the range of options open to the particular institution. Table 2.1, taken from a Vesiac report (Carder, 1963), gives approximate recommended minimum distances between seismograph stations and potential sources of disturbance.
The preliminary assessment of a site in terms of environmental conditions can give only a general indication of expected suitability, and should therefore be followed by the direct measurement of seismic noise level. It is usual to include regular continuous seismic disturbances in the range of period from 0.01 to 10 seconds, although microseisms with longer periods have been observed. Long period tilts, due to moving loads or the pressure of wind on the surface may sometimes interfere seriously with recording.
Seismic noise with periods <=0.1 s is mainly of local origin, and of limited extent. It is caused by wind, traffic, machinery, surf, waterfalls, running water, volcanic activity, etc. Its intensity and periods are influenced by the characteristics of the ground. Often a dominant frequency is observed; it corresponds probably to the minimum group velocity of surface waves. The amplitude ratio between daytime and night sometimes fluctuates in the range 2:1 to 10:1 and over, according to the source. In Japan, it has been found that short-period seismic noise has a higher frequency on granite than on soft ground.
Seismic noise having periods between 0.2 and 2.0 s belongs to a second characteristic group. The most prominent is the component with T = 0.5 s, which has been observed in Europe, the USSR and the USA. The cause of the prevailing occurrence of this maximum has not yet been explained except that the larger amplitudes have been found near towns and industrial centers. No distinct tendency for a daily or seasonal variation has been observed.
The third group of seismic noise, the typical microseisms, are most widely recorded and most discussed. On the station records they appear as groups, in each of which the amplitude increases and then fails off, suggesting some kind of beats. The periods range from 3 to 10 s. During the winter season microseismic storms lasting often 1 to 2 days are strong enough (ground amplitudes on the coast 10-20 µm, inland 1-2 µm) to make some seismograms useless. Microseisms of this sort are caused by cyclonic storms over oceans, and are propagated with gradual loss of energy into the central areas of continents. Sometimes shorter periods (1.5-2 s) of microseisms of similar character are observed near large lakes.
Figure 2.2 shows some typical seismic noise curves. Three dotted lines correspond to the maximum, mean and minimum level published by Brune and Oliver (1959), the dashed lines give two extreme examples observed in the US, the full line curves limit the fluctuations of seismic noise which was found at a European station on bedrock in a populated area, 15 km away from heavy traffic and industry. The level and details of the curve change according to local conditions.
Much higher noise levels occur in towns, particularly at short periods. Fig. 2.2a (after Kanai and others, 1966) shows some typical urban values of various types of foundation.
In order to measure the noise, the records of the most sensitive seismograph with well known frequency characteristics are used, and it is desirable to make special records at an enhanced recording speed (say 60-150 mm/min for long-period noise and 200-300 mm/min for short-period). A number of intervals, each 5 or 10 minutes in duration, are chosen during daytime and night-time, and during different weather conditions and seasons of the year. Parts of the record which contain earthquakes, or exceptional cases of extreme noise amplitude, are excluded.
The operator then measures the period and amplitude of each noise event, taking care to distinguish the components in the case of complex patterns in which waves of differing period are overlapping. The periods and amplitudes, plotted on logarithmic paper, determine the seismic noise curve for the station. The accuracy of this procedure can be much enhanced if facilities for complete harmonic analysis are available.
The quality of the site will be determined not only by noise, but by the amplitude of response to a given earthquake signal. Usually the signal-noise ratio is most favorable on the quietest sites, and the exceptional condition to be avoided is that in which low noise level coincides with generally poor transmission of waves from the earth's interior to the site in question. This point can be investigated by inspecting records of distant earthquakes of known magnitude.
It often happens that the noise curve contains 'windows' of relatively low noise level. Taking into account the variation of permissible trace amplitude as a function of period, we can construct an 'admitted' frequency-response curve by plotting the reciprocal of the seismic noise as a function of period.
If we are setting up a standard station in accordance with an international agreement, it will be possible only to vary the level of magnification of a pre-determined response curve. Normally, this should be done so that the average noise amplitude (measured from the center line) is about 0.2 mm for periods of less than 1 second, and about 0.4 mm for periods between two and six seconds. Such a setting will permit the noise level to increase by a factor of two or three under bad conditions without destroying the legibility of the record.
If we are aiming for maximum performance without regard to the needs of standardization, we may adjust the constants of the seismographs in the station so as to bring the instrumental sensitivity peaks into the noise windows. Often a small shift of the frequency-response curve results in a substantial improvement in the quality of the records (Moskvina and Shebalin, 1961). If the noise spectrum has a sharp peak in it, a rejection filter may be used. Such a filter may consist of an auxiliary galvanometer, connected in series between the seismometer and the registering galvanometer. If the filter galvanometer is lightly damped, a remarkable reduction in magnification will be obtained for seismic frequencies near resonance (Pomeroy and Sutton, 1960).
All frequency filters have the undesirable property of rejecting some of the useful content of the signal. The most fundamental approach is to record a wide band of frequencies (as on magnetic tape) and to apply advanced filtering techniques at the final playback stage.
The two essential functions of the Seismological Observatory are, first, to detect a movement of the ground, and, second, to make a permanent record of' the resulting signal. In the earlier types of seismograph the signal took the form of the deflection of a train of mechanical or optical levers which were connected directly to a suspended mass so the detecting and recording systems were necessarily adjacent to each other. This meant that the operator had to approach the detector each day, in order to change the record, and was thereby liable to cause disturbance.
In sensitive modern instruments, the primary output from the sensor is electrical, and the connection to the recording system can be along a cable, extending for a considerable distance. The seismometer vault can therefore be separated from the recording area, and most network authorities now favor this arrangement. Such separation, coupled with the fact that seismometers are often constructed with protective covers, means that the excavation can be very simple and that the recorder rooms can be located on the surface. These developments have tended to reduce building costs as a proportion of total operating requirements.
Substantial variations in design and construction are possible within these general conditions, and the choice of' a particular type of vault may be conditioned by the availability of funds and materials, by the operating policy of the station and by the number and type of instruments in use. Before giving examples of existing vaults we shall consider the general requirements in greater detail.
Modern seismometers are normally designed with dust-proof covers, but as these have to be removed during setting up and adjustment it is important that the generation of dust in the vault should be kept to a minimum. From the point of view of construction, the implication is that the walls and ceiling should be coated with a dust-free, non-flaking paint, and that linoleum or plastic floor coverings should be used.
In spite of the introduction of improved surface finishes, and the enclosure of seismometers in more or less airtight containers, condensation on the instruments can give rise to corrosion, the generation of electrochemical e.m.f's, and leakage in the electronic circuits. Extreme dryness can also cause ill effects, be cause the generation of static electricity on plastic instrument covers may deflect suspended parts. The vault humidity should therefore be monitored continuously, and held between limits of 20% and 80% of saturation.
Most vaults are below ground level, have very little ventilation and, if constructed out of concrete, have had several tons of water incorporated in the structure. Often, there is some seepage from outside, and it may be found that several liters of water have to be ejected each day. The methods available for getting rid of this water are to use chemical or mechanical dehumidifiers, or to use a combination of heat and ventilation to evaporate it.
The use of chemical dehumidifiers inside the vault has a number of disadvantages, particularly if the volume of water that has to be removed is large. Calcium chloride, which used to be common, has the disadvantage of being highly corrosive. Absorbent, non-corrosive materials like 'silica gel' can now be used repeatedly after drying out, but this can be time-consuming and costly if a lot of water is involved.
Mechanical dehumidifiers work on the principle of condensing the moisture on a refrigerated surface, and involve the inclusion in the vault of a mechanical device which is a potential source of noise and which may break down. When they switch on, the latent heat extracted from the water vapor is added to the power consumed in the refrigeration unit, and can produce a considerable disturbance in the vault temperature. It is therefore important that any such machines which are used in vaults should be conservatively rated, and that they should operate continuously.
The use of controlled ventilation depends on the ability of the outgoing air to carry more water vapor than the input, and this will depend on the degrees of saturation and the temperatures in the inward and outward air streams. The method must be applied with caution if thermal stability is to be maintained. Quantitative figures for the moisture which can be carried out under various conditions may be derived from Table 3.1 below.
Moisture carried in per cubic meter of air = 11.4 grams Moisture carried out per cubic meter of air = 22.5 grams Net transport = 11.1 g/m3 Air flow required to eject 5 liters of water = 5000/11.1 = 450 cubic meters
Thus if the vault had a capacity of 100 cubic meters, it would be necessary to change the air in it 4-1/2 times per day. It would, in fact, be quite hard to maintain adequate stability of temperature with such a rate of ventilation, and one would hope that the circulation could be reduced when the vault had dried out.
We now see that the ventilation is economical and effective if the air outside the vault is cool and dry, but that in hot and humid climates an excessively high vault temperature may be needed. In the most difficult circumstances the best procedure would be to use a mechanical dehumidifier to dry the air going into the vault, and then to keep the vault interior warm enough to provide an adequate moisture transfer in the outgoing stream.
Excessive dryness in the vault may result in the accumulation of static electricity or in the photographic paper becoming fragile. In such cases, the air in the vault should be artificially moistened. The effect of later drying of the paper may be removed by using a 10% solution of glycerine or ethylene glycol in the final washing water.
The importance of temperature control in the vault arises from the fact that a seismograph is capable of detecting extremely small movements of' the ground, and does so by detecting relative movements between elastically connected elements of its own structure. If the temperature changes, the pier and the rigid parts of the seismograph may expand or contract slightly, but such direct effects are much less important than changes in the elastic constants of springs or strip hinges, or the effects of slight tilts due to unequal expansion. These latter effects may produce relatively large displacements between the sensitive detecting elements of the seismometer especially for vertical seismometers.
Seismometers designed to detect short-period waves are much less sensitive to temperature disturbance than those designed for long-period work. The reasons for this are as follows:
The mechanical systems of most seismometers are designed to simulate the behavior of a mass hanging on a spring, or of a long vertical pendulum, the extension of the spring or the length of the pendulum being proportional to the square of the seismometer period. The equivalent mathematical length is about 25 cm for a 1-second seismometer and more than 200 m for a 30-second instrument. Tilts of the base or expansions of the spring displace the seismometer mass as though they were applied to a seismometer whose actual dimensions were as large as these mathematical equivalents, and are therefore much more significant at the longer periods.
The disturbances that appear on the record are the total effects of the temperature changes which occur within time intervals comparable to the range of earth periods for which the seismometer is sensitive. If this period is a second or so, only very slight changes have time to take place within the structure of the instrument. Long-period seismometers, however, are attached to detecting systems designed to respond to disturbances lasting several minutes, and this gives time for a given environmental fluctuation to produce much greater internal effects.
Even if some violent temperature disturbance does affect a short-period instrument, the performance will return to normal soon after the disturbance is removed. Long-period instruments may, on the other hand, take hours or even days to recover, and during this time valuable records may be lost.
In addition to the disturbances that can appear on the record, we must consider long-term effects which can carry the mass against the stops. In electromagnetic instruments this is the most serious effect, for a drift rate of say 10-4 cm/s might produce an entirely tolerable deflection of the light spot, but would move the boom out of its operating range within a few hours.
The consequence of these factors is that modern, well designed seismometers with periods of a second or so can often be exposed to the full temperature range of an extreme climate. Long-period seismometers, however, need protection against even the body heat of an operator entering the vault. They should always be covered with insulating material, and all practical methods of stabilizing the surrounding temperature must be adopted. Usually, two covers are used to protect long-period seismometers, the outer one of styrofoam or other light isolating material. A small heating cell or bulb inside the top of the outer cover minimizes turbulent air currents. Another possibility which has been tried experimentally is to maintain a vacuum within the inner cover.
The temperature disturbances can originate either inside or outside the vault. The most important ones, and the methods of minimizing their effects, are listed below:
Conduction of external variations through the walls Two to three meters of earth cover is usually sufficient for this purpose, except in regions where heavy rain storms or melting snow can cause water to percolate through the cover. One solution which has been suggested for difficult cases is to bury the vault to an adequate depth in dry fill, then to cover the area with a waterproof membrane of sufficient extent to deflect water away from the walls of the vault, and finally to add sufficient additional cover to hold the membrane in place and to prevent accidental damage. Often, of course, it is possible to use an underground gallery or cave with so much cover that even a substantial water flow takes up the average temperature of the surroundings before penetrating to the level of the instruments.
Effects of ventilation Even if the walls of the vault are perfectly insulated, some disturbance of temperature will take place whenever the air inside is replaced with some from outside. The amount of air exchanged in the course of routine visits of the operator can be kept quite small but, as we have seen in an earlier section, much larger air flows may be required to remove moisture.
The chief principle to be observed is that the air which is allowed to enter the vault should be at about the same temperature as that which is already inside, so that we are led to the concept of using a thermostatically controlled ante-room to act as a reservoir for the areas which house the most sensitive instruments.
Thermal disturbances arising inside the vault Disturbances can arise inside the vault from any fluctuation of the input power, including the entry of the operator, and the switching on or off of lights. The use of ordinary domestic thermostats in the heating system is highly undesirable, as the abrupt transients introduced by such devices are much more disturbing than the slow drifts which may be associated with a steady heat supply. Most observatories accept the much smaller disturbances which are directly associated with the routine visits of the operator, but there is room for experiment in techniques which would involve switching off heaters, the power of which was equivalent to that of the lights which were switched on, and to the body heat of the operator himself.
Electric heating is almost the only one in common use, and the normal control is by a variable auto-transformer on an alternating current supply. When it is used, it is important to avoid the danger of introducing excessive hum into short-period galvanometers or electric circuits, as the coils of many seismometers can act as both capacitive and inductive collectors. The chief precaution is to avoid the use of power circuits with open loops in them, such as those which can arise when heaters are wired in series, or when a long run of resistive cable is used to provide an extended heat source. Heating elements should also be high up in the vault, to minimize convection currents.
In much the same way as we have classified the sources of thermal disturbance, we can divide sources of mechanical disturbance into those arising outside the structure of the vault and any connected buildings, those arising in parts of the building other than the actual seismograph enclosure, and those generated inside the vault by movements of the operator.
External disturbances which can be detected at the surface of the ground have been discussed above, and the only point which requires attention at this stage is that many types of disturbance attenuate strongly with increasing depth. Some attenuation would occur if the ground had the same properties at all depths, but it is much more pronounced when the surface materials are less rigid than those below. Thus it is desirable to excavate as deeply as conditions and finances permit. The use of explosives should be minimized, to avoid fracturing the rock.
It is most undesirable for the vault to be covered by any superstructure above ground, as wind pressure can generate considerable disturbance. The practice of constructing vaults in the cellars of occupied buildings is even worse.
It is sometimes believed that the construction of a pier, surrounded by a vertical slot, offers a method of insulation against such sources of disturbances. If, however, the slot is deep enough to have much effect, it leads to a pier standing high above its point of attachment to the ground, and such a structure is, in itself, unsatisfactory. In fact, the best that can be done in these circumstances is to stand the seismometers on a solid floor, or at most to construct a low concrete plinth, as far away from the bearing walls as possible.
The vertical slot is also unsatisfactory as a means of isolating the instruments from the weight of an operator standing nearby, but substantial improvement can be achieved by using a floor held clear of the ground, and so designed that the load is carried as far away from the seismometers as possible. Examples of the various types of floor will be found amongst the vaults described below.
Air pressure variations can cause serious disturbances of long period seismometers. These disturbances are sometimes recorded as long-period waves ( T = 20 s to several minutes) which may mask the ordinary surface waves of comparable period and low amplitude. The influence of pressure variations can be minimized by using rigid covers, and by sealing the edges of the cover to the pier outside the base plate, and by keeping the volume of the seismometer parts balanced on opposite sides of the suspension hinge.
Many types of disturbance can be caused by electric induction, by leakage currents and by thermal e.m.f 's. It is important to avoid loops in power circuits and to minimize switching. The wiring between seismometers and galvanometers should be of high-quality shielded cable. Connections should be as short as possible and contacts of different alloys should be avoided.
Drainage Movement of water round the vault can disturb the temperature and cause movements of the ground. If the flow is checked, the pressure may build up and cause seepage through existing cracks and may be sufficient to create new ones. It is therefore good practice to construct a drainage ditch on the uphill side of any vault which is built on sloping ground, and also to provide drainage around the footing. In Arctic Canada, vaults in shattered and quite soft rock are operated at about freezing point, so that 'permafrost' extends through the whole medium, and provides a very solid and waterproof environment.
The essential characteristic of a first-order routine station is that it should be spacious enough to accommodate a full set of seismographs (at least three long-period and three short-period instruments) for operation continuously. Solid bedrock, freedom from interference, access in all weathers and internal convenience therefore determine the design.
Two sets of plans for the US World Network of standardized seismographs are included (Figs. 4.1 and 4.1a) of which one shows the seismometers in a vault separated from the recording and control area, and the other shows equivalent accommodation combined into a single building. The latter arrangement shortens the cable run from the seismometers to the galvanometers, but exposes the seismometers to some disturbances during service calls.
The installations are specialized in the sense that they provide all the essential operating facilities for the standard set of equipment, without leaving any pier or recording space for other purposes. Sometimes, improved thermal stability is provided by double-shelled construction, either with or without ventilation between the walls.
The practice of putting the dark room adjacent to the recording room provides maximum convenience for processing the records, but it requires that an adequate supply of water should be available. In the case of vaults attached to an organization which has office and darkroom accommodation elsewhere, some duplication of accommodation and equipment may arise from this practice.
In the Canadian Standard Vault (Fig. 4.2), the recorders and seismometers are placed in the same room, and a vestibule is provided for control equipment and storage. The internal layout of the two sections is almost identical at all stations in the network, but the relative levels of the vestibule and the main vault can be adjusted to suit local site conditions. If necessary, the two rooms could be spaced at opposite ends of a connecting tunnel.
Each of the large piers can accommodate three seismometers and their associated galvanometers, and a 3-component recorder will stand across the corresponding pair of small piers. This arrangement allows straightforward connection with short wires between the seismometers and galvanometers, and keeps the entire system within easy reach of the operator for testing and adjustment. As the floor is bridged to the main foundations of the building, the movements of the operator are no more likely to disturb the seismometers than they would if the seismometers were in an adjacent room, but the Canadian operators' instruction manual does indicate that a temperature disturbance affects the long-period seismometers during the routine service calls.
Under the extreme climatic conditions which are typical of Canada, it is more than usually important to control the exchange of outside air with that which is inside the vault. Entry to the vestibule is therefore through double doors, which provide an air lock, and the temperature of the vestibule itself is controlled by a thermostat. The main vault is provided with a steady supply of heat from electric elements controlled by a variable ratio transformer. There is no specific provision for ventilation.
The lighting arrangements are worthy of discussion. On passing through the outer door, the operator gains access to S1, which controls the porch light, and S2, which will normally turn on white lights in the air lock and vestibule. When he is ready to enter the vault, he throws S3, which changes the vestibule lights from white to red, and turns on red safelights in the vault. Inside the vault he can, if necessary, turn on white lights by means of switches S5 and S6. The switch S4 in the vestibule, and the associated connector R, are to enable the recorders to be driven either from the mains, or from a controlled-frequency battery-operated alternating-current supply.
The vault at Eskdalemuir (Fig. 4.3) is designed to combine the ease of access of the Canadian vault with some thermal isolation between the operator and the seismometers. To do this, the recorder piers have been combined into a wall which runs almost the whole length of the vault. Storage cupboards above the recorders, and provision for roller blinds which can be drawn down into vacant spaces, make an almost complete barrier between the two halves of the vault. During normal service visits, the operator can remain on the side away from the seismometers, and his weight on the H-shaped bridged walkway will bear mainly on the foundations on his own side of the vault. Two additional piers, and some solid floor space are provided on the left hand side of the entrance walkway, in order to provide accommodation for types of equipment which do not fit conveniently on to the other pier arrangement.
The Royal Observatory, Edinburgh, Scotland, has a research vault (Fig. 4.4) which incorporates a number of unusual features. In contrast to a routine vault, it is designed for the development and testing of new types of seismograph, rather than for the day-to-day operation of fully engineered instruments. It is therefore provided with piers of two different sizes, and with an area of solid floor for the accommodation of tall racks of electronic equipment.
As the research program may require the vault to be occupied by scientists for a substantial part of the recording time, every effort has been made to protect the instruments from mechanical and thermal disturbance. To this end, the center gangway is suspended from the ceiling, and by a fulcrum 2-1/2 meters outside the vault. Thus, the main loads are communicated to the walls (which are concrete poured against solid rock) at a level about 3 meters above the pier foundation, giving isolation equivalent to a 3-meter vertical slot. The transverse walkways are supported between the wall footings and the center gangway, giving less perfect, but still considerable, isolation.
Heat is supplied by means of resistive elements laid on top of the ceiling slab, and these, in turn, are covered with insulating material and a waterproof membrane. The top heat sets up a temperature inversion which inhibits convection in the vault, and the concrete slab between the heaters and the room attenuates any short-term fluctuation in power supply. Temperature-sensing elements are set at the level of the heaters, so that fluctuations can be detected and corrected before they have time to produce appreciable effects inside the vault.
Ventilation is provided by means of a small blower, injecting air through a thin-walled sheet metal duct, which is mounted along the ceiling of the entrance corridor. Near the intake, the air is warmed to a pre-set temperature by means of a proportional thermostat, which controls the power supply to a heater by means of a motor-driven 'Variac' transformer. In the course of its passage down the duct, this air comes partially into equilibrium with the outgoing air stream, so there is some further smoothing of any fluctuation of input temperature. Inside the vault, the air is injected near the warm ceiling, and then proceeds slowly downwards, against the natural direction of convection, and escapes into the corridor through the slots at the edges of the center gangway. The advantages of the system are that the outward air flow along the entrance corridor minimizes the disturbance produced when the doors are opened, and that it permits the fairly rapid circulation which is desirable when the vault is occupied for research.
For such purposes, it is sufficient to construct smaller buildings which provide reasonably satisfactory performance with conventional photographic recording equipment, although perhaps at some loss of' temperature stability and convenience of operation. The basic conditions for a reliable recording remain, however, the same as for first-order stations.
A semi-permanent building designed for rockburst studies in Canada (Fig. 4.5) illustrates the possibilities. The recording hut measures 2.1x2.1 meters, and has a small entrance vestibule which serves as a light trap. The internal accommodation is sufficient to house a conventional 3-component recorder, 3 galvanometers, and auxiliary equipment. The seismometers are housed in a small concrete shelter (Fig. 4.5a), which is constructed a short distance away from the recorder shelter. Experience has shown that the concrete work can be prepared by two men in about three days. When the foundations are complete, the building can be taken down, transported to a new site and set up again within a day. Although originally designed for a short-term project, the buildings were eventually set up to yield supporting data to a first-class station in a seismic area, and remained serviceable for a considerable number of years. Figs. 4.5b and 4.5c illustrate small underground vaults of the French networks.
The installations as described are adequate for short-period seismographs of almost any type, but the unheated seismometer shelter often lacks a sufficiently stable temperature for long-period work. Seismometers of about 15 seconds period have, however, been successfully operated in such shelters by enclosing them in a thermally insulated case, and by installing a thermostatically controlled heater in the shelter. A further improvement might be obtained by placing an outer shelter over the other one, and transferring the thermostat and the heater to the outer air-space.
Other types of shelter which have been used for both long-period and short-period seismometers are shown in Figs. 4.5d and 4.5e. These small installations can provide operating conditions which are substantially more stable than those in the majority of permanent vaults, and their only major disadvantage is the difficulty of gaining access to the equipment.
In recent years there has been a steady development of equipment which requires no special housing, including seismometers in waterproof cases and recorders that can be operated in a lighted room. Short-period waterproof seismometers are now fully competitive in price with unsealed instruments of equivalent performance. Photographic recording equipment for daylight operation is more expensive than a conventional photographic system, because the photographic types require fairly elaborate light-shielding design and the direct-writing systems require the use of fairly high-powered amplifiers to drive the stylus.
To a large extent, the extra cost of recorders for daylight operation is offset by the ease of finding suitable accommodation, and by convenience of operation. Even in a large permanent station, the use of hot-stylus or ink recorders may involve a net saving in operating cost, because the recording room can be used as a general office or laboratory. In field operations, daylight recorders can be set up in private houses, in caravans or even in tents, and in some installations completely weatherproof boxes are being used, which require no protection whatever.
In the case of waterproof seismometers, the minimum requirement is to provide protection against wind and the impact of raindrops, and in some cases it is sufficient to bury the seismometer in open ground. A better procedure is to dig a small pit down to bedrock or undisturbed hardpan, to pour in enough cement to provide a level base and, after setting up the seismometer, to cover the pit with boards and a layer of earth. One advantage of a pit of this type, in comparison with all other shallow installations, is that mechanical coupling between the overburden and the foundation is reduced to the absolute minimum, so that excellent conditions for short-period detection are often obtained.
For the very highest performance, such as that which is demanded in sensitive array stations, installation of sealed seismometers in boreholes up to 100 meters deep is coming into favor. Fig. 4.6 shows an installation of this type, as used in the Large Aperture Seismic Array in Montana.
The regional and world-wide objectives of station seismology have already been discussed, and the requirements of the world-wide system are adequately described by the statement that this network should evolve towards uniform coverage by first-order stations on a 1000 km grid, with sites chosen for foundation quality and noise level.
Regional objectives are, however, far more complex, because the overall reduction of scale limits the choice of station location while generally increasing demand for high sensitivity in short-period operation. The requirements can be specified in terms of network density, of frequency response and of dynamic range, all of which are related to amplitudes and periods of' seismic waves, noise level and seismicity.
The number of stations needed in a given area is determined by relating the concept of earthquake magnitude (Section PAR 3) to the magnification with which a station can be operated without producing an excessively noisy trace. For this purpose we have used Richter's table of the trace deflections of a 'standard' Wood-Anderson seismograph for a standard event at various distances. (Similar calculations can be made for different types of seismic waves using the amplitude factors in section PAR 3.1.) We assume that regional short-period equipment can be set up in such a way as to permit accurate readings of seismic events which deflect the trace by 1 mm, and we choose a period of 0.8 s as the one at which the magnification of the equipment must be specified. We can now construct a table connecting the magnitude ML of a detectable event on the Richter local scale with range of detection and station magnification.
The table shows how increasing the sensitivity of a station will increase the area over which it can detect events of a given magnitude, and will simultaneously reduce the magnitude of the smallest event which can be detected within a given area. Both effects increase the statistical yield of the more sensitive station, as compared with the less sensitive one.
Next, we must note that readings from at least 3 stations are required to locate a seismic source in a region of known structure, so that new information about crustal structure can be deduced only from the seismic data which remain after the location requirements are satisfied. A minimum network suitable for crustal structure work would, therefore, provide for observation of each event by, perhaps, 6 to 8 stations. These statements, coupled with the figures in Table 5.1, define the network density required.
The density needed for a minimum network can in principle be reduced by combining teleseismic readings with local ones. The best first-order stations now provide long-range observations of P waves for events of magnitude as low as 3 on the local scale, and provide precision in epicentral location which compares favorably with the standards of regional operation. At least one local observation is, however, required for the determination of origin time, so that the utilization of long-range readings on any event of magnitude 3, can increase the total weight of the data by an amount similar to the contribution of two local stations.
The frequency-response characteristics of the classes of instruments that are favored for teleseismic work are listed in section INST 1, but we shall now consider the operational requirements which are needed for particular programs of work. The discussion requires the use of the concepts of wave type and of earthquake magnitude, which are described in sections REC 2 and PAR 3 respectively, so that we can take into account the prevailing amplitudes and periods of the seismic waves concerned.
Maximum sensitivity to the compressional, or 'P' waves of distant earthquakes is obtained by using seismographs of Class A (for definitions of Classes A-D, see section INST 1.1). For the study of small tremors at short distances, superior high-frequency response is required, and background noise may often be reduced by reducing the acceptance of waves having periods near 1 second. Thus, an installation for the observation of the smallest observable tremors at distances of less than 30 km might well consist of a galvanometer and pendulum having periods of 0.01 and 0.6 s respectively. Short-period response down to 0.05 s should be maintained for the observation of events out to 300 km.
The high-frequency components of transverse waves are strongly absorbed in the earth. Thus Class A seismographs will show the S waves of small earthquakes at distances out to a few hundred kilometers, but larger earthquakes, and those observed at longer distances, require the superior low-frequency acceptance of the other types of instrument. For general use over all ranges, seismographs of Classes B and C give the best results. Some seismologists, who like to see both P and S waves on the same record, favor the wide-band response of Class D.
Surface waves are characterized by the fact that components of different frequency appear at different stages of the development of the wave train, which spreads over an increasing length of time as the waves travel round the earth. Strong response to low frequencies is essential for this type of work, Class B being the most effective for displaying the waves of weak earthquakes, or waves which have been several times round the world. Classes C, D, and E suffice for observations of relatively short surface waves ( f = 5-25 s), such as those used in the determination of magnitude, or for the study of crustal structure.
The dynamic range of an ordinary seismograph will cover an amplitude ratio of about 1:100, whereas the ratio in signal level between an earthquake of magnitude 5 at a distance of 10 000 km and one of magnitude 8 at 100 km is about 1:106. It follows that stations that are required to record large, nearby earthquakes as well as small distant ones must be equipped with seismographs of both high and low magnification, at least three recording levels being required under the most extreme conditions. The table which follows lists the maximum magnification for which the trace will remain within 50 mm of the center line under various conditions, and thereby determines the choice of the least sensitive instruments required in the station. The highest sensitivity which can be used will be determined by the noise level. The ratio between the highest and lowest magnifications thereby specified will enable the station authority to decide how many levels of sensitivity are required to cover the range.
The extension of seismic networks into relatively inaccessible regions is making the staffing problem increasingly important. There is some trend towards the maintenance of remote stations by part-time staff who have little or no responsibility for interpretation, but who are expected to keep the station in good order and to process the records.
We can therefore break down the minimum duties for such staff operators as follows:
To exchange the recording paper and to carry out time control and descriptive marking on the records.
To carry out photographic processing.
To notice and report not only the most obvious breaks in recording, but also such faults as changes in period or damping, sticking of coils, zero shifts, contact faults or irregularities of drum speed.
To make preliminary or emergency readings of records and to report them as necessary.
General experience is that the operation of a standard first-order station requires 2 to 4 man-hours per day throughout the year, and that preliminary reading and reporting take somewhat longer. Provisions for sickness, holidays, etc. lead to the conclusion that the basic routine of such a station can be regarded as work for two men. A general training in the maintenance of mechanical and electronic equipments, plus competence over the area covered by the operative sections of this Manual are the minimum qualifications required.
A second-order station will require almost as much effort if daily processing and reporting is required, but the work load is much reduced if the operator is allowed to accumulate records for, perhaps, a week before processing a batch. If this more relaxed schedule is permitted, maintenance of a second-order station becomes a suitable spare-time occupation for an operator whose normal employment is in a different field.
In modern tape-recording stations, photographic processing is eliminated, and system checking is done by visual inspection. If the operator's test routine is designed to monitor station performance without requiring records to be played back, the routine load can be even lower than that of a second-order photographic station. Playback of magnetic tape, including end-to-end audio or visual checking is, however, more demanding than the visual check of a set of drum records, and if this is required one should estimate up to half a man-day monitoring time per day (24 hours) of recording. The qualifications required of the station operators within the limited terms of reference which have been described will not cover the full requirements for utilizing station output for research purposes, or even for preparing the final list of routine readings. This interpretative function is, however, often carried out by headquarters or regional offices, where a larger and more highly qualified staff will be needed.
ARANOVICH, Z. I. (Editor) 1966. Parameters, Amplitude Frequency and Phase Characteristics for Central Seismic Stations in the Soviet Union. (Moscow: Inst. Fiziki Zemli, Akad. Nauk, SSSR.) [In Russian.]
ARANOVICH, Z. I., VVEDENSKAYA, N. A., GUBIN, I. E., KONDORSKAYA, N. V., FEDOROV, S. A., SHEBAEIN, N.V. 1966. Instraktsiya o poryyodke proizvodstva i obrabatik nablyndeniy na seismichekikh stantsiakh yedinoy systemy seismicheskikh nablyndeniy SSSR. [Instructions on operation and interpretation for the seismological stations of the unified seismological network system of the USSR.] (Moscow: Institut Fiziki Zemli Akad. Nauk, SSSR.)
BEAUCHAMP, K. G. (Editor) 1975. Exploitation of seismograph networks. (Leiden: Noordhof) 647 pp. [Proceedings of the NATO Advanced Study Institute held at Sandeford, Norway, 1974.]
BIRTILL, J. W., and WHITEWAY, F. E. 1965. The application of phased arrays to the analysis of seismic body waves. Phil. Trans. R. Soc. Ser. A., Vol. 258, pp. 421-493.
BRISCOE, H. W. et al. 1965. Large aperture seismic array. IEE Proc., Vol. 53, No. 12, pp. 1821-1859. [Related papers contributed by H. W. Briscoe, R. G. Enticknap, P. L. Fleck, C. B. Forbes, R. A. Frosch, P. E. Green Jr., C. S. Lin, R. M. Martinson, R. Obenchain, C. F. Romney, R. J. Swain, R. V. Wood Jr.]
KIRNOS, D. P. (Editor). 1962. Apparatura i metodika nablyudenii na seismicheskikh stantsiakh SSSR. [Apparatus and methods of observation in the seismic stations of the Soviet Union.] (Moscow: SSSR, Izd. AN SSSR.) 167 pp.
MATHEY, R., and ROCARD, Y. 1958. Performances de seismographes a forte amplification. [Performance of high-magnification seismographs.] C. R. de l' Ac. Sc., Vol. 247, pp. 1764-1976.
MOHAMMADIOUN, B. 1967. Structure du manteau et du noyau terrestres, d'apres les spectres d'energie de ondes longitudinales. [Structure of the earth's mantle and core from the energy spectrum of longitudinal waves.] In Papers presented at the Ninth Assembly of the European Seismological Commission, 1-7 August 1966 in Copenhagen, pp. 175-187 (Copenhagen: Akademisk Forlag.) or Ann. Geophys., Vol. 23, pp. 61-107.
PETERSON, J., BUTLER, H. M., HOLCOMB, L. G., and MUTT, C. R. 1976. The seismic research observatory. Bull. Seism. Soc. Am., Vol. 66, pp. 2049-2068.
PETERSON, J., and ORSINI, N. A. 1976. Seismic research observatories: upgrading the worldwide seismic data network. EOS Trans. AGU., Vol. 57, pp. 548-556.
UNITECH. 1974. Operation and maintenance manual, Seismic Research Observatory Data Recording System. (Austin, Texas: Unitech Inc.)
UNITED STATES COAST AND GEODETIC SURVEY. Handbook of the World- Wide Standard Seismograph Network. (Ann Arbor: Institute of Science and Technology, University of Michigan.)
AKAMUTU, K. 1961. On microseisms in the frequency range 1 c/s to 200 c/s. Bull. Earthq. Res. Inst., Vol. 39, pp. 23-75.
BERNARD, P. 1941. Etude sur l'agitation microseismique et ses variations. [Study of microseismic motion and its variations.] Doctoral thesis and Ann. I.P.G. Paris, Vol. 19, pp. 1ff.
BERNARD, P. 1964. An observational guide for microseisms. IUGG Chronicle, No. 54, pp. 102-105.
BRUNE, J. N., and OLIVER, J. 1959. The seismic noise of the earth's surface. Bull. Seism. Soc. Am., Vol. 49, pp. 349-353.
CARDER, D. S. 1963. The requirements of a high sensitivity seismograph station. VESIAC Report, No. 4410 63 X, Contr. SD-78. (Ann Arbor: Institute of Science and Technology, University of Michigan.)
COULOMB, J. 1956. L'agitation microseismique [Microseismic agitation.] Handbuch der Physik., Vol. 47, Geophysik I, p. 140.
FRANTII, G. E., WILLIS, D. E., and WILSON, J. T. 1962. The spectrum of seismic noise. Bull. Seism. Soc. Am., Vol. 52, pp. 113-121.
FRANTII, G. E., WILLIS, D. E., and WILSON, J. T. 1963. The nature of high-frequency earth noise spectra. Geophysics. Vol. 28, pp. 547-565.
KANAI, K., TANAKA, T., MORISHITA, T., and OSADA, K. 1966. On microtremors XI. Bull. Earthq. Res. Inst., Vol. 44, pp. 1297-1333.
KARNIK, V., and TOBYAS, V. 1961. Underground measurement of the seismic noise level. Studia Geoph. et Geod., Vol. 5, pp. 231-236.
LABROUSTE, Y. G., KUMETZ, and RICHARD, H. 1961. Reduction des distortions instrumentales sur des seismogrammes a courte period. (Reduction of instrumental distortions on short-period seismograms.) Ann. Geophys., Vol. 20, pp. 226-234.
MOSKVINA, A. G., and SHEBALIN, N. V. 1961. A study of seismic noise and calculation of the optimum seismograph constants. Studia Geoph. et Geod., Vol. 5, pp. 227-230.
MURPHY, A. J., and SAVINO, J. M. 1975. A comprehensive study of long-period (20-200 sec) earth noise at the High-Gain World-Wide seismograph stations. Bull. Seism. Soc. Am., Vol. 65, pp. 1827-1862.
POMEROY, P. W., and SUTTON, G. H. 1960. The use of galvanometers as band-rejection filters in electromagnetic seismographs. Bull. Seism. Soc. Am., Vol. 50, pp. 135-151.
Date created: 8/25/97 Last modified: 9/9/97 Copyright © 1997, Global Seismological Services Maintained by: Eric Bergman email@example.com