18974 (736) Singer No. 20-2
SINGER No. 20-2 ELECTRIC SEWING MACHINE

INSTRUCTION MANUAL

Table of Contents

| Next Page

18974 (736) Singer No. 20-2 Table of Contents Main Parts Accessories To start the Motor To Stop the Motor To Change the Speed Needles and Thread Relative Sizes of Needle and Thread To Insert a Needle To Thread the Needle To Commence Sewng To Remove the Work To Fasten Off the Stitching in the Work To Change the Length of Stitch To Regulate the Tension on the Thread To Use the Cloth Guide To Oil the Machine Further Instructions
THE Singer Electric No. 20-2 machine is a practical, convenient sewing machine made especially for children. It is simple and safe to operate and will do real sewing on dolly's wardrobe, or even on repairs to the grownups' clothing and other household articles. All the electrical parts of the motor are completely inclosed in the bakelite shell, so that there is no danger of electrical shock. The motor requires no attention, except occasional oiling. It must be used only on 110 volt, 60 cycle, alternating current (A.C.).
THE SINGER MANUFACTURING CO. Copyright, U.S.A. S. A, 1922, 1926 and 1936, by The Singer Manufacturing Co. All Rights Reserved for all Countries
Previous Page | Next Page
18974 (736) Singer No. 20-2 Main Parts A. STITCH REGULATOR B. LOOPER C. MOTOR PLUG D. CLOTH PLATE E. PRESSER FOOT F. NEEDLE G. NEEDLE SET SCREW H. CLOTH GUIDE J. CLOTH GUIDE SCREW K. PRESSER BAR LIFTER I. PRESSER BAR M. NEEDLE BAR N. TENSION DISCS 0. TENSION REGULATING THUMB NUT P. SPOOL Q. HAND WHEEL R. SWITCH BUTTON S. FRICTION PULLEY RELEASING LEVER T. CLAMP U. CLAMP SCREW

FIG. 2

Accessories The clamp (T, Fig. 2) is furnished for fastening the machine to a table or desk, if desired. The wooden handle and the two screws also furnished with the machine, are not used unless the motor is removed, for operation by hand.

Table of Contents 3

18974 (736) Singer No. 20-2 To Start the Motor Place the plug on the rubber-covered cord into a wall outlet or light socket. (Be sure that your currently is 110 volt, 60 cycle). Press down on the switch (R, Fig. 2) and the lever (S) at the same time; this will ill start the motor. Release the lever, and the rubber roller back of the hand wheel will touch the rim of the wheel and drive the machine. To Stop the Motor Push down on the switch (R), the same as for starting. When through sewing, pull the plug out of the electric outlet, and wrap the cord around the machine. Do not run the machine or motor except when sewing.
For the slow speed, move the belt to the outside grooves (V) in the two pulleys, as shown in Fig. 3. For the faster speed, move the belt to the inner grooves (W) in the pulleys. Be careful not to stretch the belt. Then replace the guard with it two slots over the metal tabs in the base, press it down and forward into place. If the motor stalls or the rubber roller slips on the hand wheel, it is because the work being sewn is to heavy for the machine. If the rubber roller or the rubber belt should slip when doing ordinary sewing, they may have become hard and brittle with age and should be replaced.

To Change the Speed Two sewing speeds are provided; a slower speed for careful work, or for young children, and a faster speed for ordinary straight work. To change to a different speed, remove the guard (X, Fig. 3) over the motor pulley by pressing it downward and sliding it back. Table of Contents 4
18974 (736) Singer No. 20-2 Needles and Thread Needles for the Machine No. 20-2 are of Class and Variety 21x1 and four of these needles of size 14 are furnished with each machine. Finer or coarser needles, as shown below, can be purchased from any Singer shop or Singer salesman. The size of the needle to be used should be determined by the size of the thread, which must pass freely through the eye of the needle to ensure the successful use of the machine. Relative Sizes of Needles and Thread CLASSES OF WORK Very fine silks, chiffons, lawns, batiste, etc. Fine silk goods, lawns, linens, cambrics, muslins, etc. Shirting, sheetings, muslins, dressmaking, etc. Light woolen goods, flannels, heavy silk, etc. SIZES OF NEEDLES OO O A B C D COTTON 90 SILK OOO OOO OOO OOO OOO

Table of Contents 5

18974 (736) Singer No. 20-2 To Insert a Needle To Thread the Needle

(SEE FIG. 4)

Turn the hand wheel (Q, Fig. 2) over from you until the needle bar (M, Fig. 2) moves up to its highest point. With the screwdriver, loosen the set screw (G Fig. 2) in the lower end of the needle bar, take out the old needle and put the new needle up into the bar as far as it will go, with its flat side toward the right and the long groove in the needle at the left, then firmly tighten the set screw (G).
Turn the hand wheel over away from you until the needle moves up to its highest point. Place the spool of thread on the spool pin, pass the thread from the spool toward you through the two holes (! and 2) in the thread pull-off, to the left and down through the hole (3) in the nipper lever. Draw the thread to the left under the end of the nipper lever and pass it over between the tension discs (4), up though the hole (5) in the machine, from right to left through the hole (6) near the upper end of the needle bar, down and from left to right through the eye of the needle. Draw about two inches of thread through the eye of the needle with which to commence sewing.

To Commence Sewing Raise the presser foot (E Fig 2) by turning up the presser bar lifter (K, Fig 2). Place the material beneath the presser foot and lower the presser foot. Lay the forefinger of the left hand gently on the end of the thread and turn the hand wheel (Q, Fig. 2) over from you in the direction indicated by the two arrows on the hand wheel, until the first stitch is made. Then place the end of the thread back under the presser foot and commence to sew.
FIG. 4 THREADING THE NEEDLE

Table of Contents 6

18974 (736) Singer No. 20-2 To Remove the Work Sew two stitches past the end of the seam and stop the machine with the needle bar at its highest point, then with the left hand draw a finger length of thread to the left through the hole near the upper end of the needle bar as shown in Fig. 5. With the right hand, using a pair of scissors, draw the slack thread FIG. 6. DRAWING THREAD THROUGH THE NEEDLE to the right through the eye of the needle, as shown in Fig. 6, then pull the thread upward from the work, the presser foot being down, and cut the thread close to the goods, as shown in Fig. 7. Raise the presser foot, pull the work from you and the end of the thread will be draw through the loop, as shown in Fig. 8.
FIG. 7. CUTTING THE THREAD CLOSE TO THE GOODS
FIG. 5. DRAWING THE THREAD THROUGH THE NEEDLE BAR

Table of Contents 7

18974 (736) Singer No. 20-2 To Fasten Off the Stitching in the Work When it is necessary to fasten off the last stitch in the work, stop the machine with the needle in the work, place the fingers on the material close to the presser foot to prevent the work from moving, raise the presser foot and take one more stitch in the last hole made and stop the machine with needle bar at its highest point; then with the left hand draw a finger length of thread to the left through the hole near the upper end of the needle bar as shown in Fig. 5. With the right hand, using a pair of scissors, draw the slack thread to the right through the eye of the needle as shown in Fig. 6; then pull the thread upward from the work and cut the thread close to the goods as shown in Fig. 7. Pull the work from you and the end of the thread will be draw through the loop as shown in Fig. 9. To Change the Length of Stitch The length of stitch is regulated by the same lever (A, Fig. 2) under the cloth plate. To make a longer stitch, move this lever away from you. To make a shorter stitch, move the lever toward you. To Regulate the Tension on the Thread The tension on the thread is automatic and seldom requires changing. When the needle is at its highest point, the nipper lever (3, Fig. 4) is raised and the tension on the thread is released so as to allow the needle bar to draw off sufficient thread for the next stitch. If the stitches are so tight that they pucker the material, loosen the tension by turning to the left the thumb nut (O, Fig. 2) at the front of the tension discs. If the stitch is too loose, tighten the tension by turning the thumb nut to the right.

FIG. 8. STITCHING FASTENED AT THE END OF A SEAM
FIG 9. STITCHING FASTENED OFF IN THE WORK

Table of Contents 8

18974 (736) Singer No. 20-2 To Use the Cloth Guide To guide the work accurately when sewing close to the edge of the goods, the cloth guide (H, Fig. 2) should be used. Fasten the cloth guide to the cloth plate of the machine by means of the clamping thumb screw (J, Fig. 2), inserting the thumb screw into the screw hole in the cloth plate. The cloth guide can be adjusted to bring the edge of the goods as close to the line of stitching as desired. If desired, the cloth guide can be removed from the machine. To Oil the Machine The machine should be oiled occasionally with the sewing machine oil especially prepare by the Singer Sewing Machine Company. This oil can be purchased from any Singer shop or Singer salesman. Apply one drop of oil at each of the places where a part of the machine moves against another part. About every six months, put a drop or two of CAUTION Do not run the machine or motor except when sewing. When the machine is not in use, always remove the plug from the electric outlet or light socket. Further Instructions for the use of this machine or any other Singer Machine, will be cheerfully given at your local Singer Shop.

Table of Contents 9

Previous Page

occurrence in this catalog. In some cases the location of ancient towns or areas has been interpreted by the authors from descriptions in the literature. BACKGROUND INFORMATION
The Caribbean region, bounded by Honduras, Nicaragua, Costa Rica, Panama, Colombia, Venezuela, the Lesser Antilles, Puerto Rico, Hispaniola, and Jamaica, defines a plate of Earths surface that moves semi-independently of the surrounding plates. The Caribbean plate, flanked by the North American and South American plates, moves eastward, or possibly slightly north of eastward. As the Caribbean plate moves, the American plates are driven under it on its eastern side, a process known as subduction. A vertical offset of the ocean floor can occur in this area. The crust of the Atlantic plates begins to melt as it descends into the hot rocks of the mantle. The molten material, or magma, thus created rises to form volcanoes that become the Lesser Antilles island arc. Along the northern and southern boundaries the Caribbean plate is sliding past the American plates along broken and irregular boundaries that contribute to the complexity of the movement. Finally, on the west, the Cocos plate is being driven northeastward, and is being subducted beneath the Caribbean plate. This movement causes the plate to strain against the surrounding plates, and thus, its boundaries are disclosed by a band of earthquakes that extends around the plates periphery. While the eastern boundary with its typical island arc structure of oceanic trough and volcanic islands would be expected to be the source of tsunamigenic earthquakes, the two major tsunamis affecting Puerto Rico and the Virgin Islands originated on structures transverse to the arc. The 1867 Virgin Islands earthquake and tsunami most probably originated on the Anegada Trough and the 1918 Puerto Rico event occurred along the northeast boundary in the region between Hispaniola and Puerto Rico. Stresses along this northern plate boundary have caused uplift in many of the islands and subsidence in some other areas. Upraised limestone strata (layers) on a fault block create the spectacular cliffs of Mona Island between Puerto Rico and Hispaniola. Upraised limestone strata are also found on Puerto Ricos north coast although they are deeply weathered and eroded. Intensive study of this region by side-scan sonar has revealed an unusual formation on the northern slope of Puerto Rico. A large amphitheater and a smaller one farther to the east apparently were created by slumping that could have been triggered by earthquakes in this area of high seismicity. If these large areas of rock and sediment slid as a single mass, large and destructive sea-surface waves (tsunamis) would have been generated. This catalog was compiled from historical descriptions and primary source material wherever possible. However, in many cases secondary descriptions were the only data available relating to tsunami occurrence and were the primary references used in this compilation. Mitigation of the tsunami hazard in the Caribbean from locally generated tsunamis will be difficult because of the relatively short travel time of waves generated in trench or volcanic areas to nearby inhabited land. In general this is less than 30 minutes to an hour. The local population should be educated to understand that in the event of a strong earthquake or a sudden recession

times (1946 and 1918), along the northeastern section (1867) and in the eastern subduction zone in the Windward and Leeward Islands (e.g. in 1969). E. Tsunami Effects. Tsunamis cause damage in a number of ways. While large, breaking waves are rare, the force of the waves can destroy buildings, piers, bridges and other structures. Even relatively small waves can cause strong currents that in San Francisco and Los Angeles have caused millions of dollars in damage, principally by breaking free fishing boats and yachts which collide with each other and with harbor structures. Damage can also be caused by battering by water carried debris such as logs, boats, autos etc. The retreating waves can scour the support for bridges, piers, breakwaters, etc. and cause failures. Chemical spills and fires caused by ruptured storage tanks are also common. Waves can travel long distances up rivers as bores. It is important to include search and rescue operations in emergency plans. F. Tsunami History. The preparation of a thorough history of tsunami occurrences and effects is important in understanding the local nature of the hazard and designing the most effective plan for mitigation. In Jamaica, for instance, the history shows that most tsunamis are related to landslides. Education regarding protective steps in this country would include warnings to seek higher ground in case of an earthquake. In Puerto Rico and the Virgin Islands, however, a greater danger comes from tectonic tsunamis. People in these areas should be warned to watch ,for a recession of the sea after an earthquake and to seek higher ground should a strongly felt earthquake occur. In the eastern Caribbean, on the other hand, most tsunamis originate from volcanic activity. Since volcanoes erupt over a period of days to weeks, local populations should have sufficient warning from local officials to make appropriate decisions. But it is only through a study of the past causes and effects of local tsunamis that such decisions can be made with intelligence.
Figure 2 shows the localities and years during which tsunamis have affected the various coastlines of the Caribbean. This information can be useful in regard to when the next earthquake might be expected in a specific locale. Throughout this catalog, descriptions of earthquake effects have been avoided in order to emphasize the tsunami danger, however, it must be acknowledged that most tsunamis occur in association with earthquakes and often effects and damages from the two events are difficult to consider separately. Localities shown for tsunamis in Figure 2 are often the sites of the tsunami-generating earthquake, and not the locations of the regions where the tsunami was observed. This catalog contains two separate listings of data. The first is a brief description of possible tsunami effects as noted in the literature for each of the 91 reported instances of tsunamis in the Caribbean area. Details have deliberately been avoided in favor of a short, readable description. For further reading, there is an extensive listing of references with each description of an observed tsunami. The second data listing includes information regarding Caribbean tsunamis in tabular format. The first of these tables (Table 2) lists those tsunamis that the authors have judged to be verified or very likely to have occurred. Table 3 lists tsunamis that have been reported, but in the opinion of the authors are not verifiable from the reports at hand. The validity rating at the end of each entry is based on the following considerations:

1767, April 24 [6:00 UT]: Robson reported shocks at Martinique, Barbados and British Guiana. According to reports an agitated sea ebbed and flowed in an unusual way at Martinique and Barbados. Beminghausen, 1968; Mallet, 1854; Robson, 1964, V3 1769: A tsunami reportedly inundated 15 leagues (72 km) along the coast at Port-au-Prince, Haiti. Schubert 1994. V2 1770, June 03 [19:15 LT]: A strong earthquake caused 200 fatalities in Port-au-Prince, Haiti. Waves The sea inundated 7.2 km inland. were noted at Golfe de la Gonave and Arcahaie in Haiti. Beminghausen cites Mallet and gives a similar report dated 1769 (two reports of the same event). Beminghausen, 1968; Heck, 1947; Mallet, 1854; Milne, 1912; Rubio, 1982; Schubert, 1994; Southey, 1827; Taber, 1922a, 1922b. V4 1775, February 11: An earthquake at Hispaniola reportedly 1eveIed several storehouses, and great damage was done by a tsunami, but the exact date and location are unknown. Event may be identical with March 1775 and December 18, 1775. Shepherd and Lynch, 1992; Southey, 1827. V2. 1775, March: Three strong shocks were felt on Hispaniola. Several storehouses were destroyed, and great damage was done by the sea. May be identical with February 11, 1775, and December 18, 1775. Grases, 1971; Rubio, 1982. V2. 1775, December 18: Three earthquakes were reported, and waves reportedly did extensive damage at Hispaniola and Cuba. However, Rubio does not mention any effects in Cuba. Event may be identical with February 11, 1775, and March, 1775. Beminghausen, 1968; Heck, 1947; Rubio, 1982; Southey, 1827; Taber, 1922a, 1922b. V2 1780, October 03 [22:00 LT]: An earthquake was reported to have occurred during a hurricane at Savanna La Mar, Jamaica. The sea rose to 3 m at 0.8 km from the beach and swept away a number of houses. Ten people were killed by the wave, and approximately 300 deaths resulted from the storm. All vessels in the bay were dashed to pieces or driven ashore. It is believed to be a spurious tsunami report, with the effects due to the hurricane storm surge. Heck, Milne, and Beminghausen all quote a date of Oct. 2, as reported by Perry. Millas reports Oct. 3 as the date of the storm. Beminghausen also gives Oct. 22 for this event, incorrectly citing Mallet, who gives the date as Oct. 2. Beminghausen, 1968; Heck, 1947; Mallet, 1854; Milne, 1912, Millas, 1968; Perrey, 1847; Shepherd and Lynch, 1992. Vl 1781, August 01: Grases, citing Hendersons Jamaica Almanac for 1852, reported that a series of waves and disastrous earthquakes that nearly mined the island of Jamaica. No other reports of earthquakes could be found for this day, but a major hurricane is reported. Not reported in Hall. Hall, 1907; Grases, 1971; Henderson, 1852; Millas, 1968. V2 1787, October 27 [14:2O LT]: A small local shock was felt at Montego Bay, Jamaica, and the vessels in the harbor were agitated. Mallet reports earthquakes in Jamaica at Kingston and Port Royal on Oct. 1 and 21. This is a low validity report since no wave was reported, and the agitation may have been due to a seaquake. The event was not reported in Hall, 1907. Beminghausen, 1968; Mallet, 1854; Rubio, 1982; Hall, 1907. Vl 1798, February 22: A local tsunami was reported at Matina, Costa Rica. Eyewitnesses noted unusual sea noises between seven and eight p.m. Molina, 1997. V2

1802, March 19: Earthquakes were reported in February and March at Antigua, St. Christopher, and other West Indies Islands, with the largest (Intensity IV) on this date. It was accompanied by great agitation of the sea. There were no tsunami reports so this was probably due to a sea quake. Beminghausen, 1968; Heck, 1947; Mallet, 1855; Robson, 1964. V2 1802, May 5: Earthquakes at Cumana, Venezuela, reportedly caused the water of the Orinoco River to rise, and left part of the river bed dry. This could describe wave action near the mouth of the river, or bore action up the river. The rudder of a vessel was broken. Mallet, 1855. V3 1812, March 26: A rise of sea level associated with an earthquake reportedly occurred on the Venezuelan coast. Gigantic waves reportedly broke stretches of the sea wall that protected the coast near La Guaira. Singer, et al., 1983. V2 1812, November 11 [lo:50 UT]: The sea was much agitated following an earthquake. At Annotto Bay, Jamaica, anchorage ground sank causing a ship to lose its anchor and 90 fathoms (-180 m) of cable. This may be the description of the effects of a submarine landslide or of subsidence, or could be the description of a tsunami or the action of a seaquake. Hall, 1907; Mallet, 18.55. V2 1822, May 7: At Matins, Costa Rica, earthquake shaking lasted almost 24 hours and caused ground cracking. A local tsunami was reported. The rivers and bays experienced flooding (possible description of a tsunami). Molina, 1997. V2 1823, November 30 [3:10 LT]: At 245 LT a strong earthquake was followed by a tsunami at 3: 10 LT that caused damage in Saint-Pierre Harbor. Beminghausen, 1968; Heck, 1947; Mallet, 1955; Perrey, 1847; Robson, 1964. V4 1824, September 13: Earthquakes were felt at Basse Terre, Guadeloupe, on September 9. On the 13 there was a remarkable rise and fall of the tide at Plymouth, Montserrat. There had been a terrible storm and heavy rain from September 7 to the 9. Mallet, 1855. V2. 1824, November 30: A severe shock was reported at St. Pierre, Martinique. shore. Heavy rain lasting 10 days followed. Mallet, 1855. V2 Ships were thrown on
1825, February: A shock was reported by passengers on a boat near Honduras. A rumbling noise was heard. This is a description of a seaquake. Arce, 1998. Vl 1825, September 20 [1:45 UT]: A local earthquake and oscillations of the sea were noted in Demerara County, British Guiana. An earthquake (MMI=VIB) was also noted at Trinidad, Tobago, St. Vincent, and Barbados. Beminghausen, 1968; Mallet, 1855; Milne, 1912; Perrey, 1847, V2 1831, December 3: At Trinidad and St. Christopher, a violent disturbance at sea was reported, and the shocks were felt on board ship as well as on land. This was a seaquake. An earthquake was also reported at Grenada, Tobago, St. Vincent, and British Guiana. Beminghausen, 1968; Mallet, 1855; Perrey, 1847, Robson, 1964. Vl 1837, July 26: Several shocks accompanied by a large wave occurred during a Martinique hurricane. The wave source is uncertain, Beminghausen, 1968; Grases, 1971; Mallet, 1855; Perrey, 1847. V2 1842, May 7 [17:30 LT]: A strong earthquake caused extreme damage, generated a tsunami, and killed 4,000-5,000 people. At Haiti, the destructive tsunami struck the northern coast. At Mole Saint-Nicolas, and Cap-Haitien, extensive destruction was caused by the earthquake and tsunami. At Port-de-Paix, the

Lucia. Other islands affected were Martinique, St. Vincent, Dominica, Guadeloupe, Barbados, and Grenada. Lynch and Shepherd, 1995; Robson, 1964; Schubert, 1994; Singer, et al., 1983. V2 1906, January 31: A tsunami was reported at Cumana, at Carupano, at Costas Nueva Esparta, at Rio Caribe, and at Isla de Margarita, Venezuela. Also reported were shaking effects of the waters, inland at Rio Apure, Rio Arauca, Rio Catatumbo, Rio Escalante, Rio Zulia, and Cane Colorado, Maturin. Schubert, 1994, Singer, et al., 1983. V3 1907, January 14: An earthquake (MMI=lX) ruined most of Kingston, Jamaica, and damaged much of the surrounding area, including a suspension bridge at Port Maria. Buff Bay was destroyed. About 1,000 people perished. A large tsunami pounded the northern coast with waves of 2.5 m, at Hope Bay, Orange Bay, Sheerness Bay, and St. Anns Bay, Jamaica, where the sea receded and dropped 3.7-6.2 m. At Annotto Bay, the sea receded 73-93 m, dropping 3-3.7 m below mean sea level three minutes after the shock. The returning wave raised the water level 1.8-2.4 m above normal, sweeping into the lower parts of town and destroying dwellings. On higher land it came up 7.6-9.1 m. At Port Maria, the sea receded 25.6 m 3-4 minutes after the shock and returned 1.8-2.4 m above sea level. At Ocho Rios the sea withdrew 69 m and also receded at Bluff Bay. At Port Antonio, the wave moved a small building near the beach. Waves of lesser significance were repotted along the southern coast of Jamaica. Seiches of 2.5 m were set up in Kingston Harbor. The short time period after the earthquake and recession of the water suggest a local submarine landslide source. Beminghausen, 1968, Hall, 1907; Heck, 1947; Lynch and Shepherd, 1995; Mutty, 1977; Rubio, 1982, Taber, 1920. V4 1911, November 3: A volcano-related tsunami produced extraordinary waves at Trinidad, following explosion of a mud volcano island. Amald and Macready, 1956; Beminghausen, 1968. V3 an
1916, April 24 [8:02 UT]: An earthquake (Ms=7.5) caused considerable damage at Bocas de1 Tore and Almirante, Panama, disrupting electric and water service and cutting the submarine cable linking the two areas. Debris and canoes were carried 198 m inland by knee-deep waves. Storage tanks were destroyed. The pier was damaged, houses were shifted from their supports, small buildings tumbled down, and fresh water flowed from cracks in the ground. Waves flooded Bastimento, Panama, and parts of the city were completely covered by the sea. Witnesses on board a ship reported the event at Bocas de1 Tore. The earthquake was felt as if they were on land. The boat was lifted by the waves and was swept by strange sea currents. A second earthquake (MMI=lX) was listed as having occurred at 4:26 UT on eastern Hispaniola. Beminghausen, 1968; Feldman, 1984; Heck, 1947; Kirkpatrick, 1920; Molina, 1997; Reed, 1917. V4 1916, August: Powerful waves caused the loss of USS Memphis, an 18,000 tonne [sic] cruiser, which in August 1916 was anchored in Santa Domingo harbour. At 1530 the vessel, which drew 8.2 m was anchored 3 /z cables SW of Punta Torrecilla in a light NE breeze. By 1700 she was a total wreck having been carried a distance of over 5 cables by waves estimated to have exceeded 15 m in height. West Indies Pilot, Volume 1 Art 1.149. 1916, November 12: A tsunami reportedly connected with an earthquake occurred at Ocumare de la Costa, Venezuela. Schubert, 1994; Singer, et al., 1983. V2 1918, October 11 [lo:14 LT]: A tectonic event that generated an earthquake (M=7.5) in the Mona Passage, west of Puerto Rico, may have beendue to subduction near the Brownson deep. A tsunami with runup heights reaching 6 m followed the earthquake (MMI=M) causing extensive damage along the western and northern coasts of Puerto Rico, especially to those villages established in a flood plain. At Punta Agujereada, the 5.5-6.1 m amplitude tsunami drowned 8 people, uprooted several hundred palm

1950, August 3: A wave was reported at Puerto Cabello, Venezuela, with an uncertain link to an earthquake, although there were verified reports of an inland earthquake (6.8) at Laguna La Gonzales, Chabesquen, where a mud slide caused flooding northwest of Chabasquen. The above quake also caused landslides at Caserio Providencia, Chabasquen, emptying the Laguna de1 Catire and destroying coffee plantations and three dwellings, and damaging dwellings at Los Bucarer. The earthquake also caused landslides at Puente Saguas, Biscucuy; Barrio El Atlantico, Caracas; La Boca, Anzoategui, Curumato, Guarico; La Adjuntas; and La Aguada, El Tocuyo; and at La Laguna and El Penon, Humocaro Baja; as well as surface ruptures at La Calebrina; Humocaro Bajo; Cementario; Humocaro Alto; San Rafael; Sanare; and Cerros de El Paraiso, Maracaibo. Singer, et al., 1983. VMay 31 [19:58 UT]: A 6 cm wave was recorded on the Puerto Plats, Dominican Republic, tide gauge. It may have been a wave from hurricane Alice that was in the area at this time. Millas, 1968; Murphy and Cloud, 1955. VJanuary 18: A wave caused four ships to be wrecked, and four waterfront buildings to be damaged in La Vela, Venezuela. An earthquake (Mb=5.5), off the coast of Panama, is listed for this time. Beminghausen, 1968; Seismological Notes, 1955. VJune 16: It was reported that a wave caused partial flooding of the towns south Lake Maracaibo in Venezuela. Also mentioned were landslides at Altamira and Calderas, Venezuela. Singer et al., 1983. V2 1968, September 20: An earthquake (Ms=6.2) occurred near the coast of Venezuela, and a tsunami was reported. Singer made no mention of the tsunami, but reported landslides at Chaguama de Loero, Rio Caribe, that destroyed one dwelling and damaged three others. Landslides at La Cumbre Mariano Leon, Tunapuy, reportedly injured two people, and a collapse and settlement occurred at Guiria. Hurricane Edna was passing north of Venezuela at this time. Coffman and Cloud; 1970; Singer, et al., 1983; Lynch and Shepherd, 1995. V2 1969, December 25 [21:32 UT]: A magnitude 7.6 earthquake was felt on Guadeloupe, Dominica, and Martinique, St. Vincent, Antigua and Barbados. A wave was recorded at Barbados, Antigua, and Dominica, with a maximum amplitude of 46 cm at Barbados. Van Hake and Cloud, 1971; PreZiminary Determination of Epicenters (PDE), 1969. VSeptember 13: A wave that may have been associated with a Panamanian earthquake (Mb 5.0) on this date destroyed the pier at Puerto Cumarebo, Venezuela. Schubert, 1994, Singer, et al., 1983. VMarch 16: A moderate earthquake (Mw=-6.3) caused damage and injuries to six people at Guadeloupe and minor damage at Montserrat. It was also felt at Antigua, St. Kit&, and Puerto Rico. A several-centimeter tsunami was recorded at Basse-Terre, Guadeloupe. Lynch and Shepherd, 1995; PDE, 1985. vNovember 1 [l&25 UT]: An earthquake (Ms=4.4) occurred in the Mona Passage off the north coast of Puerto Rico, generating a small wave that was reported in El Nuevo Dia on the 2d. The Puerto Rico Civil Defense reported a notable augmentation of the sea level in the area of Cabo Rojo. El Nueva Dia, 1989; PDE, 1989. VApril 22 [21:56 UT]: A MS = 7.6 created a tsunami that affected the coast of Central America from north of Limon, Costa Rica, to Panama. Less than 10 minutes after the earthquake, the residents at Bocas de1 Tore, Panama, reported that the Las Delicias sand bank, normally covered by 60-90 cm of water, emerged as the sea receded and remained above water for 5-7 minutes. Then several waves entered the bay with great force, flooding the flat northern part of the town 50-100 m from the coast. At Isla de

2.5 4.0-5.0

08 l3:28 UT]

19.5N 69.5W

MS 7.9

?uerto Rico

25 :21:32 UT] 16 14:1101 [lo:25 UT] [21:56 UT]

15.8N 59.lW

MS 7.6
17.ON 62.4W 19.ON 68.8W 4 ?.lN 83.1W

MS 6.8 Mb 5.2

Leeward Is. Puerto Rico

0.46 0.30 0.12 0.1

MS 7.4

Costa Rica

Date 1 Lat. Long. 1 Eq. Mag. 1 Area
Location of Effects 1 Runup I Deaths

09 [ 19:24 UT] 26

10.6N 63.5W

Mw 7.0

16.7N 62.2W

Montserrat

Bastimento Cristobal Portobelo Colon coca Solo Costa Rica Limon Punta CahuitaPuerto Viejo U.S. Virgin Islands St. Croix Limetree Venezuela: Isla de Margarita Tobago Montserrat
Other Possible Tsunami Events 1498-2000

14980802

ORIGIN DATA L I.-I I --e, Ms.9.9N 62.3W Venezuela

EFFECTS DATA I

Venezuela: Boca de la Sierpe

3r 1124 [23:00 LTI 25

16880301 [Gregorian] 1726

[19:OOuTl 12

[4:45uT] 1766 1021
18.3N 72.3W 20.ON 75.5W 7.4N 62.5W MS = 7.5
Haiti Santiago de Cuba and Bayamo, Cuba Venezuela
Haiti Port-au-Prince Jamaica Venezuela: Cumana Orinoco Islands Haiti: Port-au-Prince Hispaniola Hispaniola I

[9:00 UT]

18.5N 72.3W 19.ON 72.4W 19.ON 72.3W or I
Haiti Hispaniola and Cuba Hispaniola I

117503

[ 1:45 UT.] 03 [23:40 UT] 2011 n? ,?L
MMI=VIll 12.4N 61.5W 10 0x7 L <TTl
Guiana Trinidad and St. Christopher Ll^.:-:^__^
Demerara County Trinidad St. Christopher \fl^.L.:^.^

Charlotte Amalie

Ocumare de la Costa
Texas: Galveston Venezuela:

Wave height(m)

canlpano
Cuba: Playa Pauchita, Ranch0 Veloz, Las Villas Cuba: Santiago de Cuba Venezuela: Cumana Venezuela: Puerto Cabello Dominican Republic: Puerto Plata Venezuela: La Vela Venezuela: Lago de Maracaibo, Venezuela Venezuela: Puerto Cumarebo
REFERENCES Capt., 1809. On the agitation of the sea at Antigua, Nov. 1, 1755, Royal Society of London, Philosophical Transactions, Abridged, Vol. 11, pp 9-10. Arce, M.F., Molina, E. Havskov, J., and Atakan, K., 1998. Tsunamis in Central America, Tech. Rep. I1 I-12, Reduction of Natural Disasters in Central America, Earthquake Preparedness and Hazard Mitigation, Inst. Solid Earth Physics, Univ. Bergen, Norway, 34 PP. Amald, R. and Macready, GA., 1956. Island forming mud volcano in Trinidad, British West Indies, Bull. Amer. Assoc. Petroleum Geologists, Vol. 40, No. 11 p. 2756. Beminghausen, W.H., 1964. Tsunamis and seismic seiches reported from the eastern Atlantic south of the Bay of Biscay, Bull. Seismological Sot. Amer., Vol. 54, No. 1, pp 439-442. Beminghausen, W.H., 1968. Tsunamis and Seismic Seiches Reportedfrom Western North and South Atlantic and the Coastal Waters of Northwestern Europe. Naval Oceanographic Office, Informal Rep. No. 68-85, Washington, D.C. 20396, p 41. Bodle, R.R., and L.M. Murphy, 1948. Tidal Disturbances of Seismic Origin, United States Earthquakes, 1946, U.S. Dept. of Commerce, Coast and Geodetic Survey, Government Printing Office, Washington, D.C., 23 pp. Calder, Eliza S., Simon R. Young, R. Steve, J. Sparks, Jenni Barclay, Barry Voight, Richard A. Herd, Richard Luckett, Gill E. Norton, Lutchman Pollard, Lucy Ritchie, Richard D. A.Robertson, and the Montserrat Volcano Observatory Staff, 1998. Special Report 06, The Boxing Day Collapse, Dec. 26, 1997, Montserrat Volcano Observatory, West Indies, Montserrat, http://www.~eo.mtu.edu/volcanoes/west,indies/souf~ere/govt/specrep/specrep06.html. Camacho, Eduardo, 1994. The tsunami of April 22, 1991 in Central America, Tsunami Newsletter, Vol. 25, No. 1, pp 6-7. Campbell, J.B., 1991. Earthquake History of Puerto Rico, Appendix I, Part A, Aguirre Nuclear Power Plant PSAR, Weston Geophysical Research, Inc., Weston, MA, 75 pp. Centeno-Grau, M., 1969. Estudios Sismologicos, Bibloteca Academia Ciencias Frisicas, Matematicas y Naturales, VIE, Caracas, Venezuela, Segunda Edition. Coffman, J.L., and W.K. Cloud, 1970. Miscellaneous activities, United States Earthquakes, 1968, U.S. Dept. of Commerce, Coast and Geodetic Survey, Government Printing Office, Washington, D.C., p 73. Davidson, C., 1936. Great Earthquakes, Thomas Murty and Co, London, 286 pp. Deville, Ch. Sainte-Claire, 1867. Sur le tremblement de terre du 18 Nobembre, 1867 aux Antilles, Extrait des Comptes Rendus des Seances de 1Academie des Sciences, LXV, Seance du 30, Decembre, 1867, Institute Imperial de France, Academic des Sciences, Paris, p 1110. Dunbar, Paula K., and Lowell S. Whiteside, 1994. Earthquake coupling with other natural disasters, Fall Meeting, American Geophysical Union, poster session, San Francisco, CA. El Nuevo Dia, 1989. Puerto Rico, newspaper article, Nov. 2, p 11. Feldman, Lawrence, H., 1984. A catalogue of historical documents pertaining to the earthquake damage in Panama and Costa Rica, unpublished manuscript. Feldman, Lawrence H., 1993. Mountains of Fire, Lands that Shake: Earthquakes and Volcanic Eruptions in the Historic Past of Central America, (1505-1899) Labyrinthos Press, Culver City, CA. Affleck,

Grases, J., 1971. La sismicidad historica de1 Caribe, Documentos de Trabajo, Informe que presenta a la Junta Directiva de1 Comite Conjunto de1 Concrete Armado, Oct., Caracas, Venezuela. Hall, M., 1907. The Great Earthquake of January 14zh 1907, Third Report of Earthquakes in Jamaica, Special Issue of the Weather Report, No. 337, Government Printing Office, Kingston, Jamaica. Heck, N.H., 1947. List of seismic sea waves, Bull. Seismological Sot. Amer., Vol. 37, No. 34, p 269. Heilprin, Angelo, 1903. Mont Pelee and the Tragedy of Martinique, J.B. Lippincott Co., Philadelphia and London, 333 pp. Henderson, 1852. Jamaica Almanackfor 1852, p 36. Herridge de Guerrero, Christine M., 1998. email communication, July 21. Hess, H.H., 1932. Interpretation of gravity anomalies and sounding profiles obtained in the West Indies by international expedition in 1932, EOS, Transactions, American Geophysical Union, Vol. 13, pp 26-33. Kirkpatrick, R.Z., 1920. Earthquakes in Panama up to January 1, 1920, Bull. Seismological Sot. Amer., Vol. 10, No. 2, pp 121-128. Lander, James F., and Patricia A. Lo&ridge, 1989. United States Tsunamis (Including United States Possessions) 1690-1988, Publication 41-2, National Oceanic and Atmospheric Administration, National Geophysical Data Center, Boulder, CO, 265 pp. Lyell, Sir Ch., 1875. Principles of Geology 2, 12 Edition, 2 Vol., Murray, London, 146-154. Lynch, J.J., and R.R. Bodle, 1948. The Dominican earthquakes of August 1946, Bull. Seismological Sot. Amer., Vol. 38, No. 1, pp 1-18. Lynch, Lloyd L., and John B.Shepherd, 1995. An earthquake catalogue for the Caribbean Part II. The macroseismic listing for the instrumental period 1900-1991, presentation at the Caribbean and Latin American Seismic Hazard Project Workshop, Melbourne, FL, 45 PP. Mader, Charles, 1997. 1775 Lisbon Tsunami, CD video of modeling of travel times, shown at the Caribbean Tsunami Workshop, Mayaguez, Puerto Rico. Mallet, R., 1853. Catalogue of Recorded Earthquakes from 1606 B.C. to A.D. 1850, Part I, 1606 B.C. to 1755 A.D. Report of the 22nd Meeting of the British Association for the Advancement of Science, held in Belfast, Sept. 1852, John Murray, London, 177 pp. Mallet R., 1854. Catalogue of Recorded Earthquakes from 1606 B.C. to A.D. 1850, Part II, 1755 A.D. to 1784 A.D., Report of the 23Tdmeeting of the British Association for the Advancement of Science, held in Hull, Sept. 1853, John Murray, London, pp 118-212. Mallet, R., 1855. Catalogue of Recorded Earthquakes from 1606 B.C. to A.D. 1850, Part III, 1784 A.D. to 1842 A.D., Report of the 24 Meeting of the British Association for the Advancement of Science, John Murray, London, 326 pp. Mangeney, Anne, P. Heinrich, R. Roth, G. Boudon, and J.L. Cheminee, 1998. Numerical simulation of the December 1997 tsunami generated by a debris avalanche in Montserrat, Lesser Antillels, Abstracts, International Conference on Tsunamis, UNESCO, May, Paris, p 88. Mercado, Aurelio, 1997a. 1987 Venezuela earthquake and tsunami, personal communication. Mercado, Aurelio, 1997b. 1918 tsunami fatalities, e-mail communication. Mercado, A., and W. McCann, 1998. Numerical simulation of the 1918 Puerto Rico Tsunami, Natural Hazards, Vol. 18, No. 1, pp 57-76. Millas, Jose Carlos, 1968. Hurricanes of the Caribbean and Adjacent Regions, 1492-1800, Academy of Arts and Sciences of the Americas, Miami, FL, 328 pp. 91

Von Hake, CA., and W.K. Cloud, 1971. Miscellaneous Activities, United States Earthquakes, 1969, National Oceanic and Atmospheric Administration, Rockville, MD, 53 pp. Watlington, Roy A., and Shirley H. Lincoln, 1997. Disaster and Disruption in 1867: Hurricane, Earthquake and Tsunami in the Danish West Indies, Eastern Caribbean Center, Univ. of the Virgin Islands, 134 pp. Weissert, T.P., 1990. Tsunami travel time charts for the Caribbean, Science of Tsunami Hazards, Vol. 8, No. 2, pp 67-78. Whiteside, L.S., D. Dater, and P. Dunbar, 1996. Seismicity Catalog CD-ROM, National Oceanic and Atmospheric Administration, National Geophysical Data Center, Boulder, available at and U.S. Geological survey, Golden, CO, also co, httu://www.nedc.noaa.gov/see/hazard/tsu.shtml.

THE. NEED FOR UNDERWATER

HAZARDS PREDICTION
Philip Watts Applied Fluids Engineering, Inc., PMB #237,5710 E. 7* Street, Long Beach, CA 90803
ABSTRACT As of early 2000, scientists wemunable to assessmany underwater landslide hazards, to predict their occurrence following a nearby earthquake, to evaluate their tsunamigenic potential, and to warn coastal cmmmnities of imminent danger. Underwater landslides pose a continuous threat to US coastal economic activity, including valuable offshore structures, communication cables, and port facilities. Underwater landslides can generate tsunamis reaching at least 30 m above sea level, surpassing bounds of tsunamis generated by earthquakes. In the 199Os, more than 2400 people perished from landslide tsunamis as villages were swept clean by walls of water moving faster than residents could run, notably during the 1992 Flores Island, Indonesia and 1998 Papua New Guinea events. Local tsunamis also threaten lives and property along most US wadal waters, including Southern California. This fact calls into question the preparedness of US coastal communities for such events and fuels the need for underwater landslide prediction. This report summarks the motivation for a workshop funded by the US National Science Foundation and reports on the consensusfinding of workshop participants.
Science of Tsunami Hazards, Volume 20, Number 2, page 95 (2002)
INTRODUCTION Underwater landslides or submarine mass movements are generic terms encompassing all sires and shapes of sediment, rock, and reef failures. Can scientists predict the occurrence, location, and dimensions of underwater landslides for a given continental margin and earthquake trigger? This is the central question that the Workshop on the Prediction of Underwater Landslide Occurrence and Tsunami Hazards off of Southern California attempted to answer from March 10-11, 2000 at the University of Southern California, Los Angeles, California. The basic answer is yes: several methods have already been devised and several were described in presentations at the workshop. However, underwater landslide hazard assessment remains difficult because the accuracy of prediction techniques remains largely unknown, so there are no clear confidence limits. There is also a dearth of sensitivity analyses of existing predictive models, so key physical quantities remain to be identified. The number of case studies applying or comparing predictive models is quite small. The 1998 Papua New Guinea event provides one of the first complete tsunami case studies with modern seismic records, exhaustive onland investigation, several post-event marine smvcys, and successful numerical simulations. Predicted probability distributions have rarely been compared with distributions of documented or historic events. A lot of fundamental research remains. Tsunamis, a Japanese word meaning harbor waves or tidal waves, have been traditionally associated with nearshore earthquakes. The largest tsunamis readily propagate across an entire ocean to intlict significant damage and loss of life. From this perspective, either an earthquake generates a tsunami that threatens the entire Pacific Basin, or a credible tsunami threat only exists where the earthquake is felt. Locally, the earthquake is the only tsunami warning needed the larger the earthquake, the larger the expected tsunami. The Pacific Tsunami Warning Center was created in the mid 1900s following several large transoceanic tsunamis to warn distant places, especially Hawaii, of pending tsunami arrival and potential tsunami amplitude. In contrast, the decade of the 1990s saw numerous modest earthquakes that generated devastating tsunamis without any significant transoceanic tsunamis. The term local tsunami was coined to distinguish these potentially surprising events from their transoceanic brethren. Recent case studies of local tsunamis suggest that underwater landslides can be responsible for most of the devastating impact of local tsunamis. As if to underscore this point, remote tsmrarni sensors in the open ocean occasionally detect tsunamis following earthquakes where none were expected. Researchers now consider tsunamigenic landslides triggered by the earthquake. Consequently, the term landslide tsunami came into use to describe those events where underwater landslides generate the most hazardous local tsunami. The word tsunami can now encompass several tsunami sources generated by dil?erent geological events, e.g., earthquakes and landslides. The tsunami amplitude is no longer predictable from earthquake magoitude alone. On the one hand, few underwater landslides are tsunamigmric as they are either too small or too deep to generate an appreciable water wave. On the other hand, some of the largest tsunamis ever produced on earth were landslide tsunamis. Scientific observations and case studies are driving a paradigm shift in our understanding of underwater landslide and tsunami hazards. Effective hazards assessment and local tsunami warning demand that underwater landslide hazards, including tsunamis, be predicted. WORKSHOP OBJECTIVES Some invited scientists, both before and after the workshop, perceived that landslide tsunamis constitute a scientific discipline at the juncture of seismology, soil mechanics, marine geology, and fluid dynamics. The juncture is clearly more interdisciplinary and more complex than a simple boundary between two scientjfic disciplines. However, the perception of a distinct scientific discipline can only be validated by the response of fellow scientists to study natural hazards such as landslides and tsunamis. Is underwater landslide hazards an appropriate and desirable label for the collective research effort? A workshop is one mechanism whereby the enthusiasm of the scientific conimunity can be assessed. Therefore, an informal workshop objective was to assemble a group of scientific headers who could potentially form an 96

Recommendations for the US Natjonal Science Foundation Underwater landslide hazards present research opportunities within multiple directorates and divisions of the National Science Foundation. As of now, underwater landslide hazards do not fall neatly into any one directorate. In order to facilitate funding opportunities within the current institutional structure, workshop participants recommended merging support from different divisions to fund underwamr hmdshde.hazards research l The US government already possessesa wealth of existing marine geology data, much of which can be made or already is publicly available. These data are often an untapped or underused source of information for underwater landslide research and hazard mitigation purposes because of the difficulties involved in fmding and requesting the data. In order to facilitate the productive use of this data, workshop participants recommended establishing institutional links to locate and distribute archives from the US Navy, Mineral Management Service, US Geological Survey, etc. to researchers. l The workshop assembled a new composite of landslide triggering theories. Yet, almost no sites of underwater landshde research either receive or are amenable to a thorough examination of the causes of and potential for underwater landslides. In order to perform a thorough landslide case study and site specific hazard assessment, workshop participants recommcndcd choosing an intensive research site.such as Santa Barbara California. At this site, a thorough suite of tectonic and sedimentary measurements could yield invaluable insight into underwater landslide hazards, improve existing engineering models, validate underwater landslide stability analyses, and enable prediction of future landslide events. Underwater landslides form a complex and interdisciplinary research subject that could benefit from further synthesis of disparate modeling efforts. In order to facilitate. such syntheses and promote sensitivity analyses of landslide hazards, workshop participants recommended developing a landslide failure community model in order to model 3D failure surface formation, to study early time landslide motion and deformation, and to examine the role of tectonic structures such as faults in failure. l Landslide tsunami generation remains a poorly understood phenomenon for which there has recently been a proliferation of different numerical models with widely differing assumptions. In order to guarantee and promote tsunami hazard assessment, workshop participants recommended developing a tsunami generation community model including landslide tsunami sources and earthquake tsunami sources. l Researchers present at the workshop perceived that underwater landslide hazards was a relatively young and rapidly changing scientific discipline. One workshop would not suffice to detine the interests and needs of participating researchers. In order to further interdisciplinary collaboration as well as the development of the research community, workshop participants recommended funding another underwater landslide hazards prediction workshop. l Tsunami warning centers are currently set up to mitigate the impact of distant tsunamis. A felt earthquake was considered su&ient warning for local tsunamis. Devastating landslide tsunamis can appear with little to no felt earthquake, and can possess an amplitude far in excess of any concurrent earthquake tsunami. In order to help save lives endangered by landslide tsunamis, workshop participants recommended developing a prototype local tsunami warning system, Among other goals, such a system would identify and characterize underwater landshdes by seismic and acoustic techniques. l Post-event tsunami surveys during the 1990s have revealed a wealth of information regarding landslide tsunami hazards. Nevertheless, significant events are sufficiently rare that there remains much to confirm and even more to learn. In order to further understand the onhmd impact of landslide tsunamis, workshop participants recommended continuing support of International Tsunami Survey Teams. l Marine surveys are proving v&table tools for understanding and modeling landslide tsunami generation. However, only a handful of such surveys have been carried out and the inherent complexity of geological systemswill require many more before patterns emerge. In order to tiuther understand the offshore generation of landslide tsunamis, workshop participants recommended continuing support for marine surveys of tsunami source regions.

Recommendations for Other Research Institutions and Activities
l The private sector has significant Snancial concems exposed to underwater landslide hazards. In order to further prediction of underwater landslide hazards, workshop participants reammended seeking private research support, perhaps from oil and gas producers, insurance companies, or port facilities. l There are a sign&ant diEerences between the needs of researchers and the needs of disaster managers. In order to promote underwater landslide hazards mitigation, workshop participants recommended producing consumable tsunami hazard products such as underwater landslide hazards maps, probability distributions of landslide and tsunami events, observed landslide and tsunami recurrence rates, mulerwater landslide hazards risk analyses, hazard mitigation and preparation measures, cost/benefit analyses, and port survivability studies. Researchers need regular contact to keep their research up to date and to expand interest in their field. In order to promote common research interests and share the latest research results, workshop participants recommended organizing Special Sessions at AGU Meetings and other scientific events. l Researchers need printed venues in which to publish their latest work. For a relatively new research discipline, this can be especially difficult. In order to promote common research interests and share the latest research results, workshop participants also recommended organizing special issues of recognized journals. l Hazard mitigation in general often involves public education. In the case of tsunami hazards, public education has proven particularly effective at saving lives. In order to promote tsunami hazard mitigation, workshop participants recommended increasing public awareness of tsunami hazards through press releases, news conferences, television programs, web sites, tsunami animations, etc.
CONCLUSIONS The workshop considered the state of the art in seismology, soil mechanics, marine geology, and tsunami generation as a starting point in underwater landslide hazards research. During the workshop, it became clear that,new synergies are indeed providing opportunities to predict underwater landslide hazards. Landslide tsunamis motivate the urgent need for prediction, although other underwater landslide hazards are also of serious concern. Given the sparse temporal and spatial distribution of large underwater landslides, prediction is a crucial aspect of hazard assessmentand hazard mitigation. On the one band, relatively new marine geology tools enable a broader assessmentof ocean floor stability, while on the other band engineering models merge previously distinct aspects of landslide failure into predictive models. These interdependent opportunities feed the growth of a what some workshop participants termed a scientific discipline unto itself. The objectives of this discipline will include the prediction of the probabilities, locations, dimensions, motions, deformations, and hazards of prospective underwater landslides. Landslide tsunamis pose the greatest local tsunami threat according to a consensus opinion of the 67 scientists attending the workshop. Tsunamis are one of the most important natural hazards facing the five Pacific US states, occasionally inflicting more damage and casualties than large earthquakes -- viz., the 1964 Alaskan earthquake. Local tsunamis have reached 15 m above sea level during the 1998 Papua New Guinea tsunami and 26 m above sea level during the 1992 Flares Island, Indonesia tsunami, both due to nearby underwater landshdes. More than 2400 people perished from these tsunamis as villages were swept by churning walls of water moving faster than residents could run. The 1998 Papua New Guinea event has proven to be and will likely continue to be a valuable case study with which to validate models of underwater landslide hazards. To be sure, more case studies are needed, some of which should be based on the data and expertise acquired by oil and gas producers as well as the US federal govermncnt. Workshop participants have chosen the Santa Barbara, California continental slope as an ideal case study that can involve most interested scientists, agencies and institntions. 101

Figure 6. Azimuth angle dependenceof the most predominant period in the noise-eliminated spectra at Ayukawa (top) and Tosashimizu (bottom). The azimuth angle of the epicenter is defined in Figure 1. Dotted and chain lines in the figure -pond peninsulaand water channelto the tide station. in a sea of constant depth. Slopes of the shelf cause a displacement from the coincidence. AMPIXITJDE OF PREDOMINANT PERIOD COMPONENTS with extension directions of
Some typical examples of the noise-eliminated spectra are shown in Figure 7. In the figure it is indicated that spectral peaks are common to all the tsunamis at each tide station. Periods of 42, 20, 15 and 8.6 min arc identified as the common predominant periods for Ayukawa. On the other hand periods of 56,33 and 18 min for Tosashimizu. Some of them correspond with the most predominant periods of the noise eliited spectra. The predominant periods correspond to troughs of the background spectra and the amplitude is proportional to amplitude of the raw spectra. At Ayukawa the longest period component in the four predominant periods is predominant at the north tsunami and the shortest one is relativeli predominant at the southeast tsunami. At Tosashimizu the shortest predominant period of 18 min predominates for south sauces. A peak of 33 min is not observed in the 1995 KikaZma tsunami and a peak of 18 min is notable in
Figure 7. Some typical examples of the background-noise eliminated spectra for Ayukawa (left) and Tosashimim (right). Dotted lines indicate predominant periods commonly observed, which ate calculated as averages of the assumed groups for all the tsmximi
amplifude rafio of componenf.i $y--yq 0. i i. i. ; * :.

[20 min] :e

50 li0

200 250

azlmumot~mmwlcl:de4ll
Figme 8. Amplitude ratio of spectral component of 8.6 min to one of 42 min (top) and another one of 20 min to one of 42 min (bottom) of Ayukawa. The same for dotted and chains lines as

Figure 7.

amplitude ratio of component

7.0 8.0 5.0 4.0 3.0 -

[I5 min]

I - 50

7.0 8.0 5.0 4.0 3.0 2.0 -

[33 min]

. rn; H.s=: I 50

I H. yrn: I I

1.0 Oo

$2: deg)

azimuth of epicentar(
Figure 9. Amplitude ratio of spectral componentof 18 ruin to one of 56 miu (tep) aud another one of 33 ruin tc one of 56 ruin (bottom) of Tosssbimizu. The samefor dotted and chains lines ss Figure 7. the 1972 Mindanao tsunami. In the next step we take relative amplitude of the predominsnt periods to that of the longest one to cancel the effect of earthquake magnitude. In this operation we assumed that a magnitude dependence of the spectral amplitude on the period is small. The results are shown in Figure 8 for Ayukawa and in Figure 9 for Tosashimixu. In the figures directions of peninsula and channels are also indicated as shown in Figure 6. It is approximately mentioned that the shortest period components, 8.6 min for Ayukawa and 18 min for Tosashimixu, are most amplified at approximate directions of the channels. It is suggested that peninsulas prevent the second longest ones to propagate from northeast directions in both the cases. This facts

the most predominant period. The main reason is in less tsunamis of south origin in comparison with northeast tsunamis. But it is possibly explained from relative location between the bay and Axishima Island. The latter is located in front of the former and the former receives tsunami from outer sea through a narrow channel at east of the latter. The narrow channel makes the excitation difficult because of a small chance to a normal incidence to the bay. The normal incidence has an advantage of excitation of the natural oscillation (Nakamura and Watanabe, 1961). The space distribution of the most predominant period in the noise-eliminated spectra at Ayukawa (Figure 5) shows a group of predominant period shorter than 10 min existing at northeast of the tide station. One of possible explanations is trapping and leaking of the short period component by the Kinkaxan Island. It is known that there is a focusing effect of island to tsunami (e.g. Abe, 1996a,b). The effect is explained with a refraction of tsunami around the shallow slope. In this case the trapping is explained from a kind of resonance of tsunami to the trapped wave around the island. In a rough approximation the Kinkamn Island is a circular island of 4 km in diameter and has a shallow sea of 50 m in depth around it. At that time the natural period, wavelength (circumference of the circle) divided by long wave velocity at the shallow sea, is about 9 min, which is almost equal to smallest one of the predominant periods. The wave trapped at the island was radiated to the energy toward direction opposite the sources. The frequent receiving at Ayukawa is attributed to the frequent radiation of wave. This mechanism is effective for some limited region in the sources. As for the derivation of source mechanism of tsunami using a spectral superposition (e.g. Rabinovich, 1997) we only emphasize an importance of propagation effect in this stage. The fact that the same predominant period is observed at a liited and observation point instead of the generation effect. CONCLUSION We conducted spectral analysis of 35 and 27 Pacific tsunamis observed at Ayukawa and Tosashimixu, respectively. The most predominaut periods obtained were 22f 3(51 %), 8+ 3(31 %) for Aytikawa and 21+ 5(96 %) for Tosashimixu. We eliminated the background noises from the raw spectra using spectra of time histories recently observed in quiet-sea conditions. As the result the most predominant periods of

region of the

sources leads us to consider the propagation effect as a relative location between source
the tsunamis dispersed into 2-3 groups. It is shown that these periods were locahxed in the space distribution of the epicenters and depended on the azimuth angle. Period component of periods, 42, 20 and 8.6 min for Ayukawa and 56, 33 and 18 min for Tomshimku, predominated. In the azimuth angle dependence it is observed that the shortest ones were much ampliied in the azimuth angles same as those of bay axes and second shortest ones were prevented from propagating by peninsulas. Thus selective amplifications of tsunamis are verified for tsunamis which were observed in bays. It is suggested that the Kinkaxan Island contributed the period component of 8.6 min to the propagation to Ayukawa. Noise-elimination of tsunamis observed in bays clarified a selective amplification to the incident angle. The azimuth angle dependence leads us to conclude that propagation effect is important in an analysis of tsunami. ACKNOWL.EDGEMENT Author thanks to staff of Meteorological Observatories at Ayukawa and Tosashimixu, Japan Meteorological Agency for offering the tide gauge records. REFERENCES Abe, K. (1986): An explanation of characteristic distribution of the tsunami maximum inundation heights observed at the small islands, Sci.Tsunami Hazards, 4,153164. Abe, K. (1990): Spectral characteristics of the 1983 Japan Sea Tsunami observed in Japan, Tohoku Geophysical Journ.33,97-106. Abe, K. (1993): Tsunami spectra as a synthesis of source spectrum and shelf response, Proceed.luGGnOC International Tsunami Sympo.,151-163. Abe, K. (1996a): Focusing effect of islands on a tsunami, Part 1. Acase of the 1993 Hokkaido Nansei-oki earthquake tsunami (in Japanese), Zisin, 2,49,1-9. Abe, K. (1996b): Focusing effect of islands on a tsunami, Part 2. A case of the 1983 Nihonkai-chubu earthquake tsunami (in Japanese), Zisin, 2,49,11-17. Abe, K. (2000): Predominance of long periods in large Pacific Tsunamis, Sci.Tsunami Hazards, 18,15-34. Aida, I. (1982): Role of seiche in tsunamis (in Japanese): Marine Sci. Monthly, Kaiyo Publishing, KK, 149,419-427. Honda, K., T.Terada, Y, Yoshida and D.Isitani (1908): An investigation on the