Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
TRIPARTITE STATIONS AND DIRECTION OF APPROACH OF MICROSEISMS J. E. Ramirez, S.J. Instituto Geotisico at Bogota Seismographs set at the corners of a tri- angle were used as far back as 1884 by Milne and Japanese seismologists in Tokyo for the determination of the direction of approach of certain earthquake waves with various and generally unfavorable results [Imamura, 1902]. This three station distribution, generally consisting of 3 vaults, one at each corner of a triangle, each equipped with seismographs and known today as a tripartite station, began some 45 years ago to be used in an attempt to determine the velocity and direction of micro- seisms. A microseismic wave originating from some distant source and traveling along the earth's surface would reach one of the corners of the triangle first and the other two corners subsequently. By measuring the time interval between the arrival of this microseismic wave at the 3 vaults the direction of travel and the velocity of the wave can be found. The first trial (at least between two sta- tions) was probably made by Omori in Japan in February, 1908, between the station of Kon- go and Hitotsubashi, the mutual distance being 2.29 kms. but he "found it impossible to identi- fy the individual vibrations at the two places" [Omori, 1909, 1913]. Hecker [1915] made a trial at Strasbourg in 1915. He used the NS and EW component of the observatory and placed a NS component first 0.58 km. due north of the central station, and later located it 2.4 km. northeast of the observatory. In the first case, both stations receiving the time signals from the same clock, the waves at the instrument located 0.58 km. south, ar- rived now earlier and now later than at the other station (in one instance there was a maxi- mum change of 1.0 second, during an interval of 11/4 minutes). In the second case, at a dis- tance of 2.4 km. and with two different clocks marking the time signals, there were greater differences in the arrival times. The experi- ment was discontinued and "only the observa- tions of one day were the ones that could be used and even on this particular day micro- seisms were not so strong. Shaw [1922] reported at the Rome meet- ing of the first conference of the section of Seismology of the International Union of Geod- esy and Geophysics upon his work in connec- tion with microseisms and stated that at West Bromwich in 1918 during some experiments with Milne-Shaw seismographs situated 20 me- ters apart and in different building he noticed that the recorded microseisms were identical. During the spring 1919, and 1920, he demon- strated that at a distance of 3 km. each micro- seismic wave was still similar on each record. In the early part of 1921 an attempt was made to compare three stations about 16 km. apart. At this distance the waves were quite different and it was impossible to identify them for comparison. In the first months of the year 1922 two stations were arranged 4 km. apart and on different directions to the stations used in 1919 and 1920. At this distance the waves were again identified without difficulty. In the 1919 and 1920 experiments the differences in time of reception at the two stations were 0.83 seconds. The method gave some evidence that microseisms came from a north-westerly direction to west Bromwich. From January to March 1927, Nasu and Kishinouye temporarily set three horizontal pendulum seismographs near the Seismological Institute of the University of Tokyo to study the phase relation of microseisms at different places. The three stations, with that of the Institute called B, made a set of four, located at the following distances in meters: A B C B 520 920 c 920 430 D 1090 610 520 These investigators concluded, "... the observations are not sufficient to yield definite results, for they were obtained from records of NS component only." Furthermore, "it was very hard to find corresponding minute marks on records. So the comparison of rec- ords of four stations was drawn only in several cases ... The variations of amplitude like beats may be due to oscillations of different period
10 SYMPOSIUM ON MICROSEISMS of a ... land-block . . . bounded by vertical planes of discontinuity" [Kishinouye, 1935]. Krug [1937] undertook during March, May, September and October, 1936, and in January, 1937, the determination of the velocity of prop- agation and direction of microseisms at Gottin- gen. He used four portable horizontal seis- mographs of large magnification (over 6,000) for two movable stations in each of which he had a NS and an EW component. As a third station he used the Wiechert pendulum of the Gottingen Geophysical Institute (mass 1.200 kg.) with a mechanical magnification of 140 but increased by means of an optical system to 4,000. The period of the pendulum was about 11 seconds. The stations were located in the form of an isosceles triangle, with the two equal sides about 1,4000 meters. It is not clear from the article how the time signals optically marked on the records were received at each station, but it seems that no direct line or radio signals with automatic registration were used. He says: "Taglich wurde zweimal wahrend des Naueners Zeitzeichens um 1 und um 13 Uhr M. E. Z. je 5 minuten gemessen und alle Zeitsignale als Gleichzeitigkeitsmarken op- tisch sufgezeichnet." For waves of 4 to 8 seconds Krug found a velocity of 1100-200 meters per second which seems much too low. "Ob dieser unerwartet niedrige Wert zur Ausbreitung der Energie oder zur Ausbreitung einer bestimmten phase einer kombinierten Welle gehort, konnte nicht entschieden werden." An average of 80 per cent of all the readings gave a direction N 63° E ± 20°. He also found some correlation be- tween the barometric depression on the Nor- wegian Coast and the intensity of movement on the geological conditions of the place of ob- servation. In 1937 Rev. James B. Macelwane, S.J., suggested to the author, then a graduate stu- dent at Saint Louis University, Saint Louis, Mo., an experimental determination of propa- gation and origin of group microseisms by means of a simultaneously timed tripartite sta- tion. This station array consisted of four Macelwane-Sprengnether seismographs: 2 EW components, one under the Saint Louis Univer- sity Gymnasium, and one 6.4 kms. almost due West at Washington University. 2 NS com- ponents, one under the Saint Louis University Gymnasium and one 6.3 kms. almost due South, at Maryville College. Each component was re- cording identical time signals sent over leased wire every few seconds. The accurate and identical timing system, the instrumental ho- mogeneity, and its special design for recording microseisms of periods between 3 and 9 seconds were characteristics of this new tripartite sta- tion. The results were very satisfactory in dem- onstrating beyond doubt that microseismic waves are traveling and not stationary waves, that their direction of propagation can be measured, that the determination of the direc- tion of arrival at Saint Louis of these waves in all observed cases indicated that they came from tropical cyclones over the ocean; and that the bearing followed exactly the movements of the low pressure center and not the location of surf on the rocky coasts. The equations used for calculation of the direction of propagation of the microseisms were particular solutions suited to the case of a right triangle. Macelwane and Gilmore in- troduced equations valid for any tripartite sta- tion and the latter shortened the distances be- tween the stations to about 600 meters using an isosceles triangle as the general shape of the tripartite stations for tracking hurricanes. Simultaneously with the work at Saint Louis University, another tripartite station was being established by F. Trommdorff, at the University of Gottingen in Germany, leading to similar results. According to Macelwane [1946] Tromm- dorff was not sure of the meaning of his re- sults. Macelwane states, "Likewise the direc- tion of propagation determined from the arriv- al times indicates the position of the low pres- sure area. This does not tell us whether it is the storm low itself or the surf caused by it on the coast which is the cause of the micro- seisms." As a result of the experimental investiga- tion at Saint Louis University, the Naval Aero- logical Service under the Guidance of Captain H. T. Orville, U. S. N., became interested in microseismic research because it presented the "possibility of saving lives, money and proper- ty by the ability of seismographs to determine the presence of embryonic tropical storm be- fore there are indications of the geneses of those storms by other means and successfully track these storms without risking lives and property by weather reconnaissance near the eye of an active storm" (Naval Report, 1947a). Accordingly the first naval tripartite sta- tion was installed at the Guantanamo Bay, Cu- ba. In September 1944, this was reported in complete operation and a few months later en- couraging reports were given out with "some substantiation of successful tracking of the sea- sons hurricanes." To direct the establishment of this station and to act as officer in charge of the project, Commander M. H. Gilmore, U. S. NR, was appointed, an officer with many years of experience in geophysics and seismology. A new improvement was made at the Guantanamo tripartite station which was later generalized to other stations. It consisted in connecting the instruments of the outlying vaults with the main vault by a lead shielded cable. Thus the recording of the three instru-
TRIPARTITE STATIONS AND DIRECTION OF APPROACH OF MICROSEISMS 11 ments could be made accurately on the tripple drum at the main vault, instead of recording at each vault, with a resultant saving of consid- erable time. In December 1944 it was decided that ad- ditional tripartite stations should be estab- lished, one at the Naval Air Station of Rich- mond, Florida, (later moved to Opalocka) and one at the Naval Operating Base of Roosevelt Roads, Puerto Rico. The two new stations were in complete operating condition by Au- gust 1945. Thus a new technique was intro- duced as a check by triangulation on the origin of microseismic waves by means of the three stations of the tripartite type. The bearings and crossbearings obtained in 1945 with these stations seemed convincing, and the results were published in detail by Gilmore, Macelwane and others. In 1946 a tripartite station was built at Corpus Christi, Texas, (First report came out 17 August), and a single station at Antigua (was ready for operation in October). In 1947 a single station was built at Trinidad and a tripartite station at Swan Island (Feb- ruary). At present tripartite or single sta- tions include Bermuda, Whiting Field Jackson- ville and Cherry Point. A tripartite station was established in the Pacific area, on Guam in the summer 1947, and single stations at Okinawa and at other points. The station in- stalled at Corpus Christi on loose sand was dis- continued due to the unsuitable ground founda- tion. Trinidad and Antigua were also discon- tinued; Roosevelt Roads and Guantanamo Bay are now single stations. Finally in 1949 a special tripartite station was installed on the grounds of the Florissant seismograph station, for the purpose of study- ing the nature and origin of the 0.2-0.5 second period microseisms by means of special capac- ity seismographs developed at Saint Louis Uni- versity under Dr. Joseph Volk and Dr. Florence Robertson. One corner of the triangle was es- tablished at the Florissant seismic vault, the second corner was approximately 600 feet due northeast and the third corner was about 800 feet due northwest of the Florissant vault. Regarding the direction of approach of microseisms to a tripartite station, this was the view point of Naval Aerology in 1947: "The direction given by a tripartite station seems to be accurate frequently within 10 or 20 de- grees, but sometimes greater deviations are to be expected due to geological idiosyncrasies, error of the observers and the fact that the source of the microseisms is not necessarily in the center of the hurricane and even may be an extended area with a different starting point of the longest wave for each station at given moment" (Navy report 1947, b). It was also remarked that for a particular storm "all the bearings obtained are through some por- tion of the storm. Some of them perhaps lead the storm center, while most of them lag be- hind." It has been my experience also that in a particular storm the bearings obtained may differ widely, but averaging several readings leads to a truer indication of the direction of approach. According to various authors this varia- tion of the intervals of arrival of waves at the corners of a tripartite station may be accounted for as due to the possibility that the energy source of microseisms waves may not be at a point but that it is rather a wide source vary- ing its maximum around the center of the storm, or even that it is due to interference from various other sources such as a new sys- tem of microseismic disturbances coming from a different direction. More recently Gilmore has found evidence to state that microseisms "may not always be propagated outward through the earth's crust from the center of the storm in straight lines because of refraction and reflection," and that in order to track storms with microseismic cross bearings from tripartite stations accur- ately, charts are needed showing all refraction around each station. Hence, a new method has been developed by the Navy Microseismic Re- search, called the micro-ratio technique of storm tracking, which according to its author Gilmore, may permit very accurate tracking of storms that are far from land. Summing up, the tripartite station sys- tem, either in the form of a tripartite station for determining the bearing of the source or in the form of a set or sets of tripartite sta- tions for locating and following the origin of the source of microseisms has been used and still can be used successfully, provided the sta- tions have a suitable ground foundation, a proper distance between themselves, and as much as possible, instrumental homogeneity, accurate and identical time system, and simul- taneous recording on a triple drum. The direction of approach of microseisms can be rather accurately determined by each tripartite station by averaging of several read- ings. The crossbearings from tripartite sta- tions can locate energy sources and track them continuously from hour to hour and detect them days before they can be detected by any other method. REFERENCES HECKER, O., Versuche zur Bestimmung der Fortpflan- zungsgeschwindigkeit der Bodenbewegung der mik- roseismisehen Unrhue., MitteUurtgen der Zentral- buros, Gerlands Beitr. z. Geoph., v. 14, pp. 28-33, 1915.
12 SYMPOSIUM ON MICROSEISMS IMAMURA, A., Seismic triangulation in Tokyo, Publica- tions of the Earthquake Investigation Committee in Foreign Languages, No. 7, pp. 5-24, 1902. KlSHINOUYE, F., Microseisms of four second period ob- served with horizontal seismographs, Bulletin of the Earthquake Research Institute, v. 13, pp. 146- 154, and 608-615, 1935. KRUG, H. D., Ausbreiting der naturlichen Bodenunrule (Mikrosimik) nach Aufzeichnungen mit transpor- tablen Horizontal-Seismographen, Zeitschrift f. Geophys., v. 13, Heft 7-8, pp. 328-348, 1937. MACELWANE, J. B., Storms and the origin of micro- seisms, Annales de Geophysique, v. 2, fasc. 4, p. 6, 1946. SHAW, E., Communication de M. J. J. Shaw sur les mouvements microseismiques, Comptes Rendus des Seances de la Premiere Conference Reunie a Rome du 2 au 10 Mai 1922, Union Geodesique et Geophys- ique Internationale, pp. 52-53, 1922. OMORI, F., Report on the observation of pulsatory oscil- lations in Japan, Bulletin of the Imperial Earth- quake Investigation Committee of Tokyo, III, No. 1, pp. 1-35, v. 5, No. 3, pp. 109-147, 1909, 1913. Supplement to U. S. Navy hurricane microseismic re- search. Navaer 50-ir-189, May, 1947a. New developments in Naval aerology for reserve aero- logists, v. 1, p. 2, June 1947b. Discussion J. E. DlNGER Naval Research Laboratory The tripartite station is a useful tool for studying the direction of approach of micro- seisms. However, in any discussion of a tri- partite station it is essential to point out the limitations which must be taken into considera- tion if one is to obtain the greatest usefulness from the tripartite network. First, one must recognize the tolerance im- posed on the accuracy of computed bearings when taking into account the maximum accur- acy of the measurements obtained from a given set of tripartite instruments. The size and shape of the tripartite network, the speed of the paper, and the sharpness of the trace all affect the maximum accuracy that can be achieved, assuming that well-formed micro- seisms are being propagated across the tri- partite network. In fact, the accuracy varies with the direction of approach to a given net- work. A triangle having one large oblique angle will have a considerably greater accur- acy if the microseisms approach along a direc- tion parallel to the long side than will be the case if the approach is at right angles to this long side. An equilateral triangle will come the nearest to giving equal accuracies in all directions. Let us take one typical network to illustrate the instrumentation errors one might encounter. Assume an equilateral tri- angle having sides of 4,000 feet, a chart re- cording speed of 150.00 cm/min, and a micro- seismic wave traveling across the network at 8,000 ft/sec in a direction which bisects one of the angles. If one can superimpose the traces with an accuracy of ± 1 mm, the maxi- mum errors that can enter the bearing compu- tation will be ± 11°; a spread of 22°. This example I believe approaches the ultimate in instrumental accuracy; in practice the errors may be considerably larger. In view of the fact that in the computa- tion of a bearing one is in effect measuring the relative phase differences between the three recordings, it is highly essential that the three seismographs do not introduce any phase shifts into the record; or at least that the three seismographs introduce identical phase shifts. This factor demands special attention if any component of the system, such as the pende- lum or galvanometer, has a natural period in the range of the periods of microseisms be- ing recorded. Strict attention must be paid to proper damping of all such components. So far as phase shift is concerned a seismometer working into an electronic amplifier which in turn actuates the recording mechanism is to be preferred over a seismometer working directly into a recording galvanometer. In the former case there is no reaction of recording element on the seismometer to complicate phase rela- tions. A serious limitation of the tripartite sta- tion arises from the very nature of the micro- seisms. This limitation has been pointed out in the literature by a number of writers, among them being Trommsdorff [1939], Bungers [1939], Leet [1949], Donn & Blaik [1952], and Kammer & Dinger [1951]. This limita- tion arises from the observation that, in gen- eral, microseisms crossing a tripartite station do not consist of a single coherent wave train but rather are the composite of several wave trains which may differ in direction, period, and wave type. One obtains evidence that the microseisms do not consist of a coherent wave when the separation of the three seismometers is large (several miles), for in this case it is often difficult to identify the corresponding por- tions of the three records. The lack of co- herency is also illustrated by Leet [1949] in a five-minute sample record made by a three- component seismograph. This sample record shows a mixture of Love and Rayleigh waves. Leet suggests using a three-component regis- tration at each corner of the triangle so that the type of wave motion in a given interval of the record can be determined. A complete rec- ord of this sort will possibly permit one to se- lect wisely the portion of the record to be used for bearing computation.
TRIPARTITE STATIONS AND DIRECTION OF APPROACH OF MICROSEISMS 13 0 6 8~ 10 12 14 16 18 20 22 24 26 28 30 0 2 4 6 8 K) 12 14 16 « 18 20 22 24 26 28 M Figure 1. Illustration of two wave trains crossing a station at right angle Perhaps the most serious factor which in- troduces incoherency in the recorded micro- seisms is the possibility of two or more wave trains simultaneously crossing the station in different directions. In this case the trains will add together in different phases at each seismometer with the net result that a false direction and velocity will be computed. Figure 1 illustrates a very simple case of two wave trains crossing a station at right angles. For this example the apparent velocity is 11,200 ft/sec as compared to 8,000 ft/sec for the com- ponent waves; and the direction is intermedi- ate between the direction of the component waves. Figure 2 shows a graph of the ratio of apparent velocity to the real velocity as a function of the angle between the direction of propagation of two similar wave trains. This indication that the apparent velocity of the recorded microseisms is increased if two or more wave trains simultaneously cross a network has suggested a method of selecting the bearings computed from a tripartite sta- tion. This method is reported by Kammer and Dinger [1951] and later studied by Bonn and >t=4(jin«ft*8j) 45 90 135 180 225 270 315 360 Angle * Figure 2. Ratio of apparent velocity to real velocity as a junction of angle between direction of propagation of two similar wave trains. Blaik [1952]. Figure 3 shows a plot on a po- lar graph of a typical series of bearings ob- tained by the Naval Research Laboratory tri- partite station on 21 November 1950. The distance of each dot from the origin is a
14 SYMPOSIUM ON MICROSEISMS N 0° 10° 20° 30° 40° 120° 180° 170° 160° 150° 140° S Figure 3. Typical bearings and velocities observed on 21 November 1950 measure of the reciprocal of the apparent ve- locity and the orientation of the point with respect to the origin gives the computed direc- tion for the bearing. A wide scatter in direc- tion is apparent when all points are consid- ered, however, the spread in direction is nar- rowed, in this case to about 12 degrees, if only those bearings having an apparent velocity of 10,000 ft/sec or less are considered. Figure 4 shows the weather map existing at the time the bearings of Figure 3 were taken. The in- terpretation placed on these and similar re- sults for various storms is, that at those in- stances when the computed velocity is a rea- sonable value, the recorded microseisms con- sists of a coherent wave train coming from a single source. The question of refraction and reflection was raised in the preceding paper as a limita- tion on the usefulness of the tripartite sta- tion. Bonn and Blaik [1952] have also re- ferred to refraction as a possible source of error in pointing to the area of generation of microseisms. It may well be that refraction is important, but it is believed that the exis- tence of refraction will be very difficult to identify as long as so much uncertainty exists in knowing where the true area of generation really is. The assumption that refraction is the reason why tripartite bearings do not point to the center of a hurricane or low-pressure area does not seem justified until it is proved that these centers are the area of generation. However, the use of earthquake records to study refraction of seismic waves is a valid approach since in this instance the location of the source is well known. To summarize, it can be stated that the tripartite station has definite limitations. To obtain the greatest accuracy from a tripartite station one must (1) use the most advanced technique in instrumentation and (2) some method must be applied which selects the por- tions of the records to be used to compute the bearings so as to insure the use of the most nearly coherent wave trains that exist during a given microseismic storm.
TRIPARTITE STATIONS AND DIRECTION OF APPROACH OF MICROSEISMS 15 WEATHER MAP FOR inert w««th«r map ahowa the frontal and ptttoia that nxlated 12 hours previous time ,t the targe weather map thown The .irn-i* vwheie precipitation wai fall- ; v-iterday ate covered with .â¢: A companion of the two . .â¢<! bow the weather pattern hat the pa^t \2 houii. -A/T>r *I Figure 4
16 SYMPOSIUM ON MICROSEISMS REFERENCES HUNGERS, R., Die Uberlagerung zweier Wellen ver- schiedener Herkunftsrichtung, Zeitschrift fur Geo- physik, v. 15, pp. 321-332, 1939. DONN, W. L., and BLAIK, M., A study and Evaluation of the Tripartite Seismic Method of Locating Hur- ricanes. Technical Report on Seismology No. 19 of the Lamont Geological Observatory, 1952. LEET, L. DON., Discussion of Tripartite Microseismic Measurements. Bull. Seism. Soc., v. 39, pp. 249- 255, 1949. KAMMER, E. W., and DINGER, J. E., Hurricane Well as a Generator of Microseisms, J. of Meteor., v. 8, pp. 347-353, 1951. TROMMSDORFF, F., Untersuchungen uber die naturliche Bondenruhe (Mikroseismik) mit transportablen Dreikomponentenstationen. Zeitschrift fur Geo- physik, v. 15, pp. 304-320, 1939. Discussion MARION H. GILMORE U. S. Naval Air Station at Miami Father Ramirez has given an excellent 65- year history on tripartite stations used in the study of earth motions connected with earth- quakes and microseisms. He has described the systems used and mentioned some of the results obtained by investigators in many countries,- including a few comments on his own experi- ments at St. Louis University in 1939. His summary of the view point of Naval Aerology in 1947 and again in 1952 is essentially cor- rect. Before one is able to discuss adequately the reliability of bearings and cross-bearings from microseismic storms it is first necessary to show where they originate. Father Ramirez dismissed this important point in these words, "The results were very satisfactory in demon- strating beyond doubt that microseismic waves are traveling and not stationary waves, that their direction of propagation can be measured, that the determination of the direction of ar- rival at St. Louis of these waves in all ob- served cases indicated that they come from tropical cyclones over the ocean, and that the bearing followed exactly the movements of the low pressure center and not the location of surf on the rocky coasts." In spite of these statements there are still a few doubts concern- ing the actual source of regular typhoon-hurri- cane microseisms with periods of 3.5 to 6.0 seconds. The pounding of large ocean swells from a storm at sea upon a land mass or a con- tinental shelf cannot be the direct cause of large storm microseisms unless an abundance of observational data are disregarded. Micro- seisms have been repeatedly recorded several days before the energy front from newly formed storm swells could reach a land mass on which the seismograph was located. There- fore, in order to establish again the fact that this type of storm microseism is generated when the energy from a tropical storm is trans- mitted by some coupling mechanism directly to the ocean floor around the area of the storm, the following observational data are submitted: 1. Microseisms travel approximately 100 miles per minute while the energy front of storm produced swells seldom exceed 18 miles per hour. In other words microseisms could trav- el 740 miles in seven minutes or less but it would take at least 40 hours for newly formed ocean swells to travel the same distance. The data presented in Figure 7 of the following paper shows three typhoons passing over an area of the Pacific that is almost equal dis- tance from Guam, Okinawa and Manila, or ap- proximately 740 miles from each station. A critical analysis of the data will show that each storm quickly intensified into a typhoon with greater wind force and that the micro- seisms, in all three cases, immediately regis- tered a sharp increase in amplitude at Guam, Okinawa and Manila. It is physically impos- sible for swells to have had anything to do with the simultaneous increase in microseisms be- cause there would have been a delay of at least 40 hours for the intensified swell crest to reach the nearest land mass. Nor could the sudden intensification, which caused the increased microseismic activity, have occurred two or three days earlier for, had such been the case, the swells would have reached Guam several days before reaching the other two sta- tions. The microseismic amplitude curves, Figure 1, show no such delay in either of the three storms. The simultaneous arrival of the larger microseisms at the three stations can be explained only by the theory that energy from severe tropical storms is transmitted to the bottom of the ocean where it immediately generates microseisms that are propagated out- ward in all directions at a speed slightly great- er than 100 miles per minute. It will be noted, also, that the three typhoons were going away from Guam and approaching the other two stations. It is common knowledge that a swell traveling ahead of a storm will attain great height and period, while those traveling in the opposite direction never become prominent. But the per cent of increase in microseisms at Guam was as large or larger than at the other two stations. 2. Another argument against the surf theory of generation of microseisms is that the amplitude of microseisms recorded at Guam, Swan Island and Bermuda, islands surrounded for great distances by uniformally deep water, show no correlation with the state of the sea surrounding them. Heavy swells may pound
TRIPARTITE STATIONS AND DIRECTION OF APPROACH OF MICROSEISMS 17 against a steep coast within one hundred feet of a seismograph or approach a continental shelf without causing the least increase in hurricane type microseisms. At other times each of the above stations have recorded micro- seisms more than ten times normal amplitude while the state of the sea around the islands ranged from a dead calm to swells of only one to three feet. Many other investigators also have reported a complete lack of correlation between surf and microseismic amplitude. 3. There is still another obvious reason why storm microseisms used in tracking severe tropical disturbances could not be caused by ocean swells striking a coast. The energy front of ocean swells, racing ahead of similar storms at any average velocity should travel equal distances in water of uniform depth, re- gardless of the direction from which they come. In other words, if storm swells gener- ate storm microseisms, a seismograph located on an island surrounded by several hundred miles of uniformly deep water should start re- cording large microseisms as soon as the hurri- cane is a fixed distance from the station re- gardless of the direction of approach. If this were true lines drawn around such a station, for example Bermuda, showing the location of a 90 knot hurricane when microseisms are first recorded should be nearly circular. But this is far from true. The Bermuda seismograph has recorded many large microseismic storms when hurricanes of similar intensity moved directly toward the station from the east, the south, or the west and these data were used to prepare the chart shown in Figure 6 of the following paper. The microseismic range for hurricanes located south or west of the station is double the range to the east. It is therefore obvious that swell activity reaching the Bermuda coast cannot account for the generation of such storm microseisms because the limiting distance is not the same in all directions. The only theory that agrees with all the available data is that the microseisms are generated in the ocean bed in the vicinity of the disturbance. The microseisms generated in the vicinity of a storm are transmitted outward in all direc- tions through the earth's crust according to established laws of physics pertaining to wave motion in an elastic medium. Wave motion through a perfectly homogeneous substance is transmitted in straight lines from the source and decreases in amplitude, in direct propor- tion to the square of the area covered. How- ever, such conditions of complete homogenity exist over very limited areas of the earth. Thus, it is only natural to expect that most micro- seisms generated by storms do not arrive at a station from the exact direction of the storms. "Microseismic Barriers," major discontinuities in the earth's crust or gradual changes in the density and elasticity of a portion of the earth's crust, are very logical phenomena that appear to reflect, to refract, and to absorb microseisms. The following conclusions are therefore apparent: (a) The true origin of storm micro- seisms appears to be in the area of strong winds of a hurricane or typhoon, (b) It is possible to calculate accurately the direction microseisms are traveling when they pass over a tripartite station. If such microseisms from tropical storms were always propagated out- ward in concentric circles it would be very simple to determine their exact origin and the location of the storm by means of cross bearings from two or more tripartite stations. How- ever, microseisms, as explained above, do not always travel in straight lines, and this results in large errors unless maps of the characteristic refraction patterns are constructed around each tripartite station. Until such charts are made and the proper corrections applied micro- seismic bearings will continue to look like those in Figure 1. Discussion from the Floor Melton. I would like to comment at this point that many of our observations constitute a sta- tistical problem, and it should be valuable to examine some of our observations in that light. This term "coherence" which we have been using is much discussed in the literature, and, in particular, the subject of cross correlation as we move these two seismographs apart. It is obvious that if we place them on the same pier their correlation should be unity or 100 per cent, provided the instruments are operat- ing properly. We should like to see the curve of this correlation factor as the separation is increased to the order of distances we have been discussing. Byerly. (Commenting on tripartite measure- ments) It makes a difference how you set these seismographs down, too. You can set them down side by side and they won't show the same thing at all if you don't set them down properly. Melton. Yes. But this business of properly planting a seismograph is a controllable thing. Working in the marshes of Louisiana which literally float, I have observed that when re- flection instruments were simply set down in the mud, some of the reflections came out ly- ing on their backs. However, the proper way to plant such instruments is on about 20 feet of pipe pushed firmly into the marsh, and if one plants them this way the reflections come out properly. (Byerly asked if there were any cases of mi- croseisms and no storms. Gilmore replied that there are a few cases. Peoples asked if the reverse was true in any case. Gilmore's reply was yes, but then it turns out the storm has been â¢over-rated in intensity. Melton asked about the reliability of the intensity of the storms. Van Straten replied that as the storms slow down they usually intensify. Macelwane
18 SYMPOSIUM ON MICROSEISMS US. NAVY RCH PROJECT Figure 1. Microseism bearings from Guam asked for the difference between tropical and extra-tropical storms. Van Straten replied that there are not the sharp fronts in the tropical storms. Deacon raised the point that it is the energy in the storm that is important. Bath asked if the waves differed greatly. Deacon said no.) Deacon mentioned attempts recently made by Mr. Darbyshire in his laboratory to find the direction of the source of microseisms by com- paring phases and amplitudes of the east-west and north-south components using recordings from the Galitzin seismographs at Kew made available to him by the Superintendent of the
TRIPARTITE STATIONS AND DIRECTION OF APPROACH OF MICROSEISMS 19 Observatory. The comparisons were made for selected, simple, meteorological situations when there was little doubt of the actual source of the microseisms. The experiments were most disappointing and the main conclusion was that the ground movements at Kew were so com- plicated, presumably by the interference of waves which have suffered multiple refractions and reflections, that comparison of the two horizontal components gave little information about the direction of the source. During the experiments Fourier analyses were made of simultaneous recordings of waves and the three microseism components. The most striking features of these analyses was the striking similarity of the period spectra of the north- south, east-west, and vertical components. Though it was fairly certain that the direction of the storm changed over a time of 30 hours from west through northwest to north, the relative amplitudes of the east-west and north- south records and spectra remained the same. The horizontal components always had the same range of periods as the verticals; this might indicate that there was no appreciable movement due to Love waves which would be likely to widen the period range of the hori- zontal components. The amplitudes of the vertical movements were roughly twice those of the horizontal com- ponents. The use of tripartite stations is based on the assumption that the microseisms approxi- mate to a regular, simple wave system, with the wave crests travelling as straight lines. The fact that tripartite stations in some places, and at some times, give excellent results shows that the microseisms sometimes do travel as simple waves, but experience in Great Britain indicates that such behavior is exceptional there.
