breakwater stability under tsunami attack-for a site in nicaragua.pdf

8/9/2019 breakwater stability under tsunami attack-for a site in nicaragua.pdf

1/220

Breakwater stability under tsunami attack for a site in Nicaragua

ing. K. Cuypers


2/220

The background picture of the front page is a well known colour woodcut byKatsushika Hokusai, a famous late eighteenth century Japanese artist. It is No 20 fromthe series Thirty-Six Views on the Mount Fuji . Many textbooks and web sites depictthis wave as a tsunami wave. Indeed it resembles the shape of a tsunami wave when

breaking occurs. But in fact it is a wind generated wave. Nevertheless it has becomean international symbol for tsunamis.The small pictures in front are:

1. Severe damaged caisson breakwater at the Port of Okushiri after the 1993Hokkaido tsunami

2. A snapshot of a computer model of the 12 July 1993 tsunami in the EastJapanese Sea

3. Damage due to the 1964 Alaska tsunami4. The 1946 Aleutian tsunami arrives on the Hawaiian coast as a bore5. An unbroken tsunami wave arriving at a steep coast


3/220

reakwater stability under tsunami attack for a site in Nicaragua

Final Report

05-01-2004 ing.K.Cuypers


4/220

Master Thesis: Faculty of Civil Engineering, TU Delft

Cooperating company: HAECON n.v.Harbour and Engineering ConsultantsDeinsesteenweg 110B-9031 DrongenBelgium

Supervising committee:TU Delft Prof. M.J.F. Stive

Prof. G.S. Stellingir. H.J. Verhagen

HAECON Dr.ir. M. Huygensir. N. Gunst

Student: ing. K. [email protected] st.nr. 1 055 739

Date: December 2003

Colophon


5/220

PrefaceThis report is the final thesis for the masters degree in Civil Engineering at theDelft University of Technology. The study was conducted in cooperation withthe Belgian Harbor and Engineering Consultants company HAECON n.v.

I would like to thank the following people who attributed to the realization ofthis thesis. First of all, I would like to thank my graduation committee at TUDelft: Prof. M.J.F. Stive, who as chairman of the committee inspired me withhis enthusiasm, Prof. G.S. Stelling, for his help in setting up the mathematicalmodel and especially ir. H.J. Verhagen, who was always available for givingadvice. Secondly, I would like to thank the people of HAECON for theircontinuous support, in particular ir. N. Gunst who was always available toanswer my questions and to provide me with the necessary information

concerning the Nicaragua project. Also, I would like to thank Dr. ir. M.Huygens and ir. P. de Pooter for their feedback on the conducted research.Special thanks to Prof. J. De Rouck and Dr.ir. P. Troch, of the University ofGhent, for their critical reflections on the breakwater stability, and to Dr.ir.P.H.A.J.M. van Gelder, of TU Delft, for his kind advice regarding somestatistical questions.Finally, I would like to thank my parents for their support during all theseyears.

Kim Cuypers

The Hague, 5 January 2003


6/220


7/220

5 DERIVING THE DESIGN TSUNAMI FOR THE SITES OF INTEREST

5.1 INTRODUCTION................................................................................................................................15.2 MODEL SET-UP .................................................................................................................................2

5.2.1 Model description.......................................................................................................................25.2.2 Model input .................................................................................................................................35.2.3 Available calibration data .........................................................................................................7

5.3 PIE DEL GIGANTE ..........................................................................................................................105.3.1 First Calculation ......................................................................................................................105.3.2 Possible reasons.......................................................................................................................145.3.3 Calibration of the model: the 1992 Nicaraguan tsunami......................................................315.3.4 Design wave..............................................................................................................................35

5.4 BAHIA DEL SALINAS ....................................................................................................................385.4.1 Bathymetry at Bahia Del Salinas ............................................................................................395.4.2 Design wave at Bahia Del Salinas..........................................................................................405.4.3 Comparison between the design wave at Bahia del Salinas and Pie del Gigante..............41

5.5 DESIGN TSUNAMI WAVE .....................................................................................................................42

6 CONCLUSIONS

PART 2: Breakwater Stability

7 LITERATURE STUDY ON THE CAISSON BREAKWATER STABILITY 7.1 ON THE HYDRAULIC ASPECTS OF TSUNAMI BREAKWATERS IN JAPAN ....................1

7.1.1 Design formulas ..............................................................................................................................1 7.1.2 Comparing to the Goda design formulas for unbroken short period waves...............................3 7.1.3 Comments ........................................................................................................................................6

7.2 OTHER REFERENCES...................................................................................................................... 7 7.3 CONCLUSION ....................................................................................................................................7

8 DESIGN OF THE CAISSON BREAKWATER 8.1 THE CAISSON BREAKWATER FOR PIE DEL GIGANTE.........................................................1 8.2 THE DESIGN TSUNAMI WAVE .....................................................................................................1 8.3 CAISSON TSUNAMI STABILITY...................................................................................................2 8.4 CONCLUSIONS..................................................................................................................................3

9 LITEARTURE STUDY ON THE RUBBLE MOUND BREAKWATER STABILITY 9.1 LABORATORY STUDY FOR THE DESIGN OF TSUNAMI BARRIER....................................1 9.2 DYNAMIC WAVE FORCE OF TSUNAMIS ACTING ON A STRUCTURE .............................7 9.3 EXPERIMENTAL STUDIES ON TSUNAMI FLOW AND ARMOR BLOCK STABILITY

FOR THE DESIGN OF A TSUNAMI PROTECTION BREAKWATER IN KAMAISHI BAY.9 9.4 CONCLUSIONS................................................................................................................................14

10 THEORETICAL STABILITY CONCEPT FOR THE RUBBLE MOUND BREAKWATER 10.1 STONE STABILITY UNDER FLOW...............................................................................................2

10.1.1 Izbash ..........................................................................................................................................2 10.1.2 Dutch Formula...........................................................................................................................3 10.1.3 CERC formula ............................................................................................................................4

10.2 STONE STABILITY GOVERNED BY THE VELOCITY ON TOP OF THEBREAKWATER CREST ....................................................................................................................5

10.2.1 A comparison between the calculated values...........................................................................7


8/220


9/220

reakwater stability under tsunami attack for a site in Nicaragua

IntroductionThis thesis is related to the technical feasibility study of the Nicaragua DryCanal. The Nicaragua Dry Canal is a container traffic between the Atlantic andthe Pacific coast of Nicaragua. The containers will be transported by train froma container port at the Pacific coast to a container port at the Atlantic coast. Thisthesis deals with the feasibility of the container port at the Pacific side of

Nicaragua. As the region is sensitive to tsunami attack, it is essential to take thetsunami risk into consideration.

The two proposed harbor locations at the Nicaraguan coast.

This study will focuses on the stability of the breakwater under tsunami attack,for the two proposed harbor locations, Pie del Gigante and Bahia del Salinas.

The study consists of two parts: the derivation of a design tsunami and the breakwater stability.

Part 1: Design TsunamiThe aim of part 1 is to formulate a realistic design tsunami for the two harborlocations at the Pacific Coast of Nicaragua.At first the basics of tsunami science will be presented. Next out of theavailable tsunami data for Nicaragua it will be tried to derive a design tsunamiwave. A statistical approach for deriving the design tsunami is preferred, but itis uncertain if this will be possible. The design tsunami has to be derived fordeeper water as to minimize local disturbance. And from deeper water thetsunami has to be derived to the two possible harbor locations. For these wavecalculations a simple set-up is preferred with a 1D hydraulic model.


10/220

In chapter 1, a general introduction to tsunamis is given. Chapter 2 gives anoverview of the tsunamis that have occurred in Central America, and of theavailable data concerning these events. In chapter 3, the necessary backgroundknowledge is given to judge the possibility for linking the available tsunamidata to a tsunami wave. Also insight is given into the relevant tsunami science.

Chapter 4 analyses the dataset from chapter 2 and it is decided if a designtsunami will be derived by a statistical approach or based on a large historicalevent. In chapter 5, the design tsunami event is derived for the two locations in

Nicaragua. In chapter 6, the conclusions concerning part 1 are brieflysummarized.

Part 2: Breakwater StabilityThe aim of part 2 is to formulate a breakwater design which can withstand thedesign tsunami derived in part 1 .Two different breakwater types will be studied: the caisson breakwater and therubble mound breakwater. For both structures, first the available literature will

be reviewed. If the available literature turns out to be insufficient for designinga breakwater, a theoretical stability concept will be set up. Afterwards, the

breakwater design for short period waves will be tested on a tsunami load andif necessary the design will be adapted.In chapter 7, the available literature on caisson breakwater stability under atsunami attack is presented. In chapter 8, the caisson breakwater is designed towithstand the design tsunami. The literature study concerning the stability ofrubble mound breakwaters is presented in chapter 9. Because of the limitedspecific references concerning the stability of rubble mound breakwaters, atheoretical stability concept is set up in chapter 10. In chapter 11, the rubblemound breakwater is designed to withstand the design tsunami load. In chapter12 the conclusions concerning part 2 are briefly summarized.

Chapter 13 gives a summary of the entire study, together with an overview ofthe conclusions. Recommendations for further research are formulated inchapter 14.


11/220

Part I: Design Tsunami

The Waiakea area of Hilo, Hawaii, after the 1960 Chilean tsunami struck the Hawaiianislands. The force of the debris filled waves bent the parking meters.


12/220

05-01-2004 Part 1 - Chapter 1 - 1

1. GENERAL INTRODUCTION TO TSUNAMIS

A tsunami is a series of ocean waves generated by any rapid large-scaledisturbance of the seawater . Most tsunamis are generated by earthquakes(Satake, 2002). Other occasional causes are, volcanic eruptions, landslides,undersea landslides, underwater explosions or meteor impacts. All these eventstrigger a series of fast moving, long waves of initial low amplitude that radiateoutward in a manner resembling the waves radiating when a pebble is dropped ina pond.

Figure 1.1 Tsunami waves radiating outward like when a pebble is dropped in a pond.

The term tsunami comes from the Japanese term meaning harbor wave. Becausetsunamis often cause large standing waves in harbors and bays, which can persistfor many hours, tsunamis in ancient Japan became known as great harbor waves.Other terms for tsunamis are tidal wave or seismic sea wave. The term tidalwave is an exact translation from the ancient Greek name for a tsunami. It refersto the initial manifestation of a tsunami at the shore, where it often resembles afast ebbing or flooding tide. The term tidal wave is less commonly used, toavoid the association with tides. It is not only incorrect with regard to its origin asa tsunami has nothing to do with the tides, but also inappropriate in its descriptivecharacter. Seismic sea wave is the most descriptive term, as most tsunamis aregenerated by large underwater earthquakes.

As stated above far out the most common cause for tsunami generation areearthquakes. Earthquakes are commonly associated with ground shaking that is aresult of elastic waves traveling through the solid earth. However, near the sourceof submarine earthquakes, the seafloor is permanently uplifted and down-dropped,

pushing the entire water column up and down (Figure 1.2).


13/220

Part 1 - Chapter 1 - 2 05-01-2004

Figure 1.2 Underwater earthquake resulting in an uplift and down-drop of the ocean surface.

The potential energy that results from pushing water above mean sea level is thentransferred to horizontal propagation of the tsunami wave (kinetic energy) (Figure1.3). The initial water displacement is split into a tsunami that travels into thedeep ocean (distant tsunami) and another tsunami that travels towards the nearbycoast (local tsunami). In deeper water a tsunami is barely noticeable and will onlycause a small and slow rising and falling of the sea surface as it passes.

Figure 1.3 Wave generation and propagation as a result of the initial surface disturbance

Near their generation point, tsunamis have a great wave length, often exceeding200 km. The maximum ocean depth lies between 8 to 10 km in the oceantrenches. This makes a tsunami a typical long wave, as the wave length is muchlarger than the water depth. Therefore the propagation speed of tsunamis doesonly depends on the water depth ( .c g h= ). As the local tsunami travels over thecontinental slope, shoaling begins: the tsunami waves slow down and becomecompressed, causing them to grow in height. This results in steepening of thewaves (Figure 1.4).


14/220

05-01-2004 Part 1 - Chapter 1 - 3

Figure 1.4

Shoaling process of tsunami wave, clearly showing the steepening of the wave

Contrary to many artistic images of tsunamis, most tsunamis arriving at the coastdo not result in giant breaking waves. Rather, they come in much like very strongand very fast tides (i.e., a rapid, local rise in sea level) (Bryant, 2001). Much ofthe damage inflicted by tsunamis is caused by strong currents and floating debris.The small number of tsunamis that do break often form vertical walls of turbulentwater called bores.Because of their great length, arriving tsunamis at the coast travel much fartherinland than normal wind waves. The extend of the area that is affected by the

onshore tsunami is expressed by the run-up height. The run-up height is the heightabove a reference level, which the water reaches as the tsunami waves run out onthe coast (Figure 1.5).

Figure 1.5 Tsunami arriving at a coast as fast incoming tide, causing a large run-up.

Tsunamis rank high on the scale of natural disasters. It has been estimated thattsunamis cause between 5 and 15% of the earthquake damage worldwide(Bernard, 2002). Since 1850, tsunamis have been responsible for the loss of over120,000 lives and billions of U.S. dollars damage to coastal structures andhabitats. Over 99% of these casualties were caused by local tsunamis that occurabout once a year somewhere in the world. The last decade 12 major tsunamishave struck coastlines around the Pacific Rim, causing more than 4,000 casualties

and an estimated 1 billion U.S. dollars in damage. This makes tsunamis the mostdestructive ocean waves of all.

Hrun-up


15/220

05-01-2004 Part 1 - Chapter 2 - 1

2. TSUNAMIS ON THE PACIFIC SIDE OF CENTRALAMERICA

As in the literature no ready-to-use methods are available for a statistical approachfor determining a design tsunami, it will be tried to set up a statistical approach

based on available tsunami data for Nicaragua.In this chapter the necessary dataset of tsunami events will be derived. On the

basis of the available data it will be decided which extra information is necessary.Although the sites of interest are located in Nicaragua, due to the exceptionaloccurrence one is forced to put the statistical investigation into a broader view of

the region. As a first assumption the Pacific side of Central America (Mexico toPanama) is regarded as one region.

Tsunamis are classified into three categories, distant (> 750km from the source),regional (100-750km from the source) and local (< 100km from the source)(Fernandez, 2000). In principle the Pacific coast of Central America can beexposed to the three types of tsunamis.Looking at the entire Pacific Region Ida (1981) sets up a distribution for tsunamigenerating earthquakes for different regions in the Pacific (Figure 2.1). As can beseen Central America is not the most active tsunamigenic

region in the Pacific, but a significant part of all generated tsunamis are generated in Central America:

6.5% during 1900-1980. This makes local and regional tsunamis a significanthazard for the Central American coast.Due to geographical conditions, the impact from distant tsunamis is relativelysmall for the Central American region. For the total energy radiating from thecircum Pacific zone, the percentage of the received energy was only 6.4% inCentral America during 1900-1992 (Hatori, 1995).Therefore we can focus in this study on the local and regional tsunamis in CentralAmerica, meaning that the source region can be limited to the Central Americanregion.

tsumanigenic: tsunami generating (e.g. tsumanigenic earthquakes)


16/220

Part 1 - Chapter 2 - 2 05-01-2004

Figure 2.1 Local and regional tsunami activities in the Pacific Ocean, from 1900 to1980

expressed in percent

A dataset for the tsunamis generated at the Pacific side of the Central Americanregion for the period 1900-2002, was derived out of the Historical Tsunami

Database for the Pacific, 47 B.C. to present. This is the most up to date tsunamicatalogue available. In appendix 1 it is described how the data are retrieved.The results of this query are presented in Table 2.1. The epicentral locations of thetsumanigenic earthquakes are plotted on a map of the Central American region(Figure 2.1).


17/220

05-01-2004 Part 1 - Chapter 2 - 3

date time generation point earthquake data tsunami data generation area

Year Mo Da Hr Mn Sec Lat Long Dep Ms Mw Mt Int Hrmax D F C TR Source Region

1 1906 1 31 15 36 1 -81.5 25 8.6 8.7 8.7 3 5 M 1000 T SAM Columbia-Ecu ador:TUMACO

2 1907 4 15 6 8 16.7 -99.2 25 8.2 7.8 1 2 S T CAM Mexico

3 1928 3 22 4 17 9.6 16.84 -96.02 15 7.5 7.4 T CAM MEXICO

4 1928 6 17 3 19 31.8 16.65 -96.5 2 7.8 7.9 8.1 1 2.5 S T CAM MEXICO: NEAR COAST OFGUERRER

5 1930 8 31 0 40 36.6 33.9 -118.6 0 5.2 -0.5 6.1 M L CAM S.CALIFORNIA

6 193 2 6 3 10 36 54.2 19.46 -104.17 25 8.1 7.6 8.2 -0.5 0.7 N T CAM Central Mexico,Jalisco

7 1932 6 18 10 12 17.8 19.53 -103.72 71 7.8 7.5 7.8 -2 0.1 N T CAM Central Mexico, Jalisco

8 1932 6 22 12 59 27.5 19.09 -104.41 1 6.9 1.5 10 M T CAM Central Mexico

9 1934 7 18 1 36 27.8 8.01 -82.56 25 7.7 7.6 1 2 T CAM Costa Rica-Panama

10 1941 12 5 20 47 3.3 8.61 -83.18 32 7.5 7.3 -3 0.1 N T CAM Central America

11 1941 12 6 8.5 -84 33 7 -3 0.1 N T CAM Costa Rica-Panama

12 1948 12 4 0 22 21.6 -106.7 33 7 1.5 S 4 T CAM Mexico:MARIA MADRE ISLAND

13 1950 10 5 16 9 10.4 -85.2 60 7.7 7.7 7.8 -2 0.1 N T CAM Nicaragua,Costa Rica-Panama

14 1950 10 23 16 13 14.3 -91.8 30 7.1 7.5 7.5 -1 0.2 N T CAM Guatemala-Nicaragu a

15 1950 12 14 14 15 16.3 -98.2 50 7.5 7.1 -1 0.3 N T CAM Acapulco, S.Mexico

16 1951 8 3 0 24 13 -87.5 100 7 9 0 10 T CAM Potosi, Honduras

17 1957 7 28 8 40 9.8 16.88 -99.29 36 7.9 1.5 2.6 S 68 T CAM S.Mexico:GUERRERO

18 1958 1 19 14 7 25 0.99 -79.48 20 7.3 7.7 1.5 S 20 T SAM Colombia-Ec uador

19 1962 3 12 11 40 15.3 8.09 -82 .68 19 6.8 -3 0.1 N T CAM Costa Rica-Panama

20 1962 5 11 14 11 56.8 17.14 -99.7 46 7.2 7 0 0.8 N 4 T CAM S.Mexico

21 1962 5 19 14 58 13.1 16.99 -99.69 16 7.1 S 30 T CAM S.Mexico

22 1965 8 23 19 46 1.5 16.17 -95.85 10 7.8 7.5 -1 0.4 N T CAM MEXICO

23 1973 1 30 21 1 13.5 18.45 -102.96 37 7.5 7.6 8 -3 0.1 N 56 T CAM S.Mexico:FARIAS,TECOMAN

24 1979 3 14 11 7 15 17 .77 -101.23 24 7.6 7.6 -1.1 0.4 N 5 T CAM S.Mexico:GUERRERO

25 1979 12 12 7 59 4.6 1.6 -79.36 24 7.7 8.1 2.5 5 M 600 T SAM Colombia-Ec uador

26 1981 10 25 3 22 15.8 18.12 -102 21 7.3 7.2 -3 0.1 N T CAM MEXICO

27 1985 9 19 13 17 49.7 18.46 -102.37 21 8.1 8 1.5 3 M 0 T CAM MEXICO: MICHOACAN: MEXICO CI

28 1985 9 21 1 37 13.5 17.83 -101.62 18 7.5 7.5 0 1.2 S T CAM MEXIC0: SW COAST: MEXICO CIT

29 1988 4 30 15 42 32.6 -117.3 5 N 0 M CAM San Diego, S. California

30 1992 9 2 0 16 2.7 11.72 -87.39 45 7.2 7 .7 7.9 2.8 10.7 M 170 T CAM Nicaragua

31 1994 1 17 12 30 55.5 34.16 -118.57 14 6.8 6.7 -3 0.1 N 0 T CAM Southern California

32 1995 9 14 14 4 33.9 16.84 -98.61 29 7.2 7.3 -1 0.42 N 0 T SAM Oaxaca, Mexico

33 1995 10 9 15 35 54.1 19.05 -104.21 26 7.3 8 2 5 M 1 T CAM Jalisco, Mexico

34 1996 2 25 3 8 16.6 15.93 -98.1 17 6.9 7.1 -3 0.12 N 0 T CAM Oaxaca, Mexico

35 1999 9 12 18 17.62 -101.6 2.5 8 S 0 U CAM Guerero, Mexico

Table 2.1 Databank for the tsunamis at the Pacific side of Central America from 1900 to 2002.

The data are retrieved out of the Historical Tsunami Database for the Pacific,47 B.C. to present.

Legend Earthquake data

Dep : depth of the source earthquakeMs : surface-wave magnitudeMw : moment magnitude


18/220

Part 1 - Chapter 2 - 4 05-01-2004

Tsunami dataMt : tsunami magnitudeInt : tsunami intensityHrmax : maximum observed or measured wave height in metersN : total number of available run-up and tide-gauge observationsD : damage code

N : non damaging eventS : slight damageM : moderate damageL : large (severe) damage

F : number of reported fatalities due to the eventC : cause of the tsunami

T : tectonicL : landslide

M : meteorologicalU : unknown

Generation areaTR : tsunamigenic region codeCAM : Central America (7 N - 35 N, 125 W - 75 W)SAM : South America (58 S - 7 N, 100 W - 60 W)Source Region : source region of the tsunami. Descriptive indication of the tsunami source area

Figure 2.2 The epicentral locations of tsunamigenic earthquakes in Central America for the

period 1900 to 2002


19/220

05-01-2004 Part 1 - Chapter 2 - 5

To be able to work with this data, and to make correct interpretations one has tohave the necessary background knowledge concerning tsunamis. In the following

chapter an overview of tsunami science will be given from a hydraulicengineering point of view.Looking at the cause of a tsunami it is clear that earthquakes are the dominantcause : 32 tsunamis on 35 are generated by earthquakes and these events areresponsible for all casualties in Central America. This conforms the general ideathat tsunamis are predominantly generated by earthquakes. Therefore the rest ofthe study will be emphasized on earthquake generated tsunamis. In the followingchapter an overview of the most relevant subjects of tsunami science will begiven. In appendix 2 a short introduction to seismology is given, which is a mustwhen one is not acquainted with seismology.With the background knowledge out of chapter 3 and appendix 2 it will be tried,in chapter 4, to decide how a design wave can be coupled to the available dataset;is a statistical approach possible or does the largest tsunami event has to be takenas a base for a design tsunami.


20/220

05-01-2004 Part 1 - Chapter 3 - 1

3 BASICS OF TSUNAMI SCIENCE

In this chapter an overview is given of the basics of tsunami science. When onestarts to study tsunamis one gets a very fragmented and incomplete notion of thetsunami subject. Because on one hand a lot of information is easily available inthe so called popular sciences. On the other hand very specific information can befound in publications. In between, very little information is available. Informationout of popular sciences gives a very superficial insight and they have the tendencyof focusing on exceptional events. Publications are usually focused on one

particular aspect of tsunamis. Therefore compiling a correct and clearrepresentation about what a tsunami actually is, how it behaves itself, how it can

be described, how it can be modeled is a difficult and time consuming job.This chapter is a compilation of tsunami science emphasized on the considered

problem in this study: deriving a design tsunami to a coastal location in order touse it for breakwater stability analysis. The purpose of this chapter is not to give atotal overview of tsunami science, but to give a correct, summarizedrepresentation. It is therefore the key to the understanding of this study.The chapter is split up into a paragraph about tsunami generation by earthquakes,a paragraph concerning tsunami waves, a paragraph describing the modeling of atsunami and a paragraph giving an overview of the common tsunami scales.

3.1 TSUNAMI GENERATION BY EARTHQUAKES

3.1.1 Generation mechanism

Although landslides, volcanoes and asteroid impacts can all trigger tsunamis, byfar the most common cause are submarine earthquakes. Even if the seafloor itselfdoes not trigger tsunamis, the shaking may trigger coseismic landslides, whichcan cause tsunamis.

Not all earthquakes generate tsunamis. The pattern and extent of vertical grounddeformation from an earthquake uniquely determines whether or not a tsunami isformed. The ground deformation is determined by the fault geometry. Asexplained in appendix 2, earthquake fault geometries basically fall apart in threefundamental end members: strike-slip, thrust-slip and dip-slip faults (Figure 3.1).Strike-slip faults involve horizontal motion of the earths crust, while thrust-dipand dip-slip faults induce vertical motion.


21/220


22/220

05-01-2004 Part 1 - Chapter 3 - 3

earthquakes, are potential tsunami generators (unless of course they trigger amassive landslide into the sea).

General the larger the moment magnitude of an earthquake, the larger the area thatis deformed and the greater amount of slip, thus producing disproportional largertsunamis than smaller events.In addition to an earthquakes magnitude, the deeper the hypocenter or focus of anearthquake, the smaller the vertical deformation of the earths surface. A deeperhypocenter allows the seismic energy to spread over a larger volume. Earthquakesdeeper than about 30km rarely cause sufficient deformation to generate tsunamis.But truly great megathrust earthquakes that occur deeper than 30km, such as the1960 Chilean event (M w=9.5), can occasionally trigger tsunamis.

However often the earthquake magnitude suggests no or only a small tsunami to be generated, as in fact a considerable tsunami is generated, these earthquakes arecalled tsunami earthquakes.

3.1.2 Tsunami earthquakes

Tsunami earthquakes are earthquakes which generate disproportional largetsunamis for there surface wave magnitude M s. For most seismic earthquakes inthe Pacific, the size of the tsunami increases as the surface wave magnitude M s increases. However it is known that many earthquakes with small and moderatemagnitude M s can produce large and devastating tsunamis. A well knownexample is the Meiji Sanriku earthquake of 1896 (M s = 7.2) which was barely feltalong the adjacent coastline, yet the tsunami that arrived 30 minutes afterwards

produced run-ups exceeding 30 m at some locations and killing 22,000 people.The 1992 Nicaraguan earthquake (M s = 7.2) is also a well know example of atsunami earthquake, barely felt by coastal residents, yet producing a largedestructive tsunami.The generation mechanism of a tsunami earthquake is not well understood, at thismoment two possible causes are put forward; seismic generated submarinelandslides and slow rupturing along fault lines.

3.1.2.1 Seismic generated submarine landslides

Earthquakes can induce submarine landslides, which can amplify the initial waterdisplacement caused by the seismic bottom movement. This explanation has notyet been proven conclusively, although recent research of Synolaksis (2003)suggests that up to one third of the tsunamis is cogenerated by submarinelandslides. The reason that this co mechanism is hard to prove is the difficulty ofdetermining whether an underwater landslide has occurred.Figure 3.3 shows the generation mechanism of a tsunami by a landslide.


23/220

Part 1 - Chapter 3 - 4 05-01-2004

Figure 3.3 Generation of a tsunami by a landslide

3.1.2.2 Slow rupturing along fault lines

A distinct characteristic for tsunami earthquakes is a slow rupturing process,causing a deficiency in short period energy. Only broadband seismometers,sensitive to low frequency waves can detect these slow earthquakes. Thereforeconcealing the possibility of tsunami generation for ordinary seismometers.Figure 3.4 illustrates the difference between a tsunami earthquake and an ordinaryone. The Hokkaido 1993 tsunami was an ordinary event. The earthquake thatgenerated it lasted for about 80 seconds and consisted of five large and two minorshock waves. The earthquake was regionally felt along the Northwest coast ofJapan and produced a deadly tsunami on this coast. In contrast, the Nicaraguantsunami of 1992 had no distinct peak in seismic wave activity. The earthquakewas hardly felt along the nearby coast, yet it produced a killer tsunami.

Figure 3.4 Comparison of the rate of seismic force between a normal tsunamigenicearthquake (Hokkaido) and a slow tsunamigenic earthquake (Nicaragua),

a so called tsunami earthquake

The cause of the slow rupturing can be found in the presence of an accretionary prism developed at the interface of two crustal plates or in a plate interface filledwith soft subducted sediments (Figure 3.5).


24/220

05-01-2004 Part 1 - Chapter 3 - 5

The sediment lenses at the accreting margin can amplify the bottom movementcaused by a rupture up to a factor 2 (Synolaksis, 2003). Occasionally also slumpsin this soft layer can be induced by earthquakes, thus amplifying the initial water

displacement.In subduction zones without large amounts of sediment, the plate interface is filledwith soft sediments. When a shallow earthquake occurs at this interface, the slipcan extend to the surface, breaking through a relatively weak plate interface filledwith sediments. the shallow depth and soft sediments on the interface are thenresponsible for efficient tsunami generation and slow rupture propagationrespectively.

Figure 3.5 Accreting and non-accreting margins

3.1.2.3 Discrepancy between M s and M w

Regardless of the mechanism, an important diagnostic feature of tsunamiearthquakes is the difference between the surface wave magnitude M s and themoment magnitude M w.For example the two earthquakes presented in Figure 3.4 are a good illustration ofthis feature: the Hokkaido event has an M s of 6.3 and an M w of 6.3,whereas the

Nicaragua event has an M s of 7.2 and an M w of 7.7. In both cases the M w clearlyshows the tsunami generation potential of both earthquakes, however this is noguarantee their will effectively be a generation of a tsunami, other parameterssuch as fault geometry and depth of the hypocenter also play an important role.


25/220

Part 1 - Chapter 3 - 6 05-01-2004

3.2 TSUNAMI WAVES

3.2.1 Propagation

The earthquake rupture triggers a series of fast-moving, long waves of initial lowamplitude that radiate outward in a manner resembling the waves radiating when a

pebble is dropped in a pond (Figure 3.6). Part of the tsunami travels into the deepocean (distant tsunami) and a part travels to the nearby coast (local and regionaltsunami).Most tsunamis generated by large earthquakes travel as wave trains. These wavetrains contain several long waves with wave lengths that often exceed 200 km in

the deep ocean. Usually one of the waves is more pronounced than the other. Indeep water even the highest wave seldom exceeds 0.5m. Their steepness is sosmall that a ship out at sea does not feel a tsunami pass.

Figure 3.6 Tsunami waves radiating outward like when a pebble is dropped in a pond.

The maximum ocean depth lies between 8 to 10 km and tsunamis in the deepocean have typical wave lengths of hundreds of kilometers. Thus the wave lengthexceeds many times the water depth (L>>d), therefore tsunami waves travel asshallow water waves. The propagation speed of shallow water waves is solely afunction of the water depth:

.c g h= (3.1)

c : wave celerity (m/s)g : gravitational constant (m/s)h : water depth + water elevation (m)


26/220

05-01-2004 Part 1 - Chapter 3 - 7

Due to the propagation speed being solely a function of the water depth, tsunamistravel at speeds of 150-250 m/s ( 550-900 km/h) in the deeper ocean (d>2 km),30-90 m/s ( 100-300 km/h) across the continental shelf (d


27/220

Part 1 - Chapter 3 - 8 05-01-2004

When the tsunami acts as a rapidly rising tide, the resulting incident currentvelocities are relatively low and most initial damage will result from buoyant andhydrostatic forces and the effects of flooding. Once the tip of the wave has

arrived, the velocities will gradually increase and a lot of damage will occur bystrong currents and floating debris. When the tsunami hits the coast as a bore,initial damage will not only occur by buoyant and hydrostatic forces but also byimpact forces of the bore front caused by high turbulence, the high watervelocities and by dragged debris in the bore front. Once the bore front has passedfurther damage can be attributed to strong currents and floating debris.

Figure 3.8 A tsunami arriving at the coast as a non breaking wave. The first picture is the arrival ofthe wave and the last pictures is the retreat of the wave. Notice the absence of the small

truck in front on the last picture. Also notice the darker parts on the seawall, clearlyindicating the maximum water height.

Figure 3.9 A tsunami wave arriving at the coast as a bore


28/220

05-01-2004 Part 1 - Chapter 3 - 9

3.2.2 Terminology

Figure 3.10 Various terms to describe tsunami waves

The terminology used to describe tsunami waves is shown schematically in Figure3.10 . Much of this terminology is similar to the terminology used for wind waves.However, there are some differences. The run-up height is referred to a referencelevel, mostly mean sea level and not to the water level at the actual moment thetsunami occurred. This because the run-up height is mostly measured some daysafter the tsunami occurred. So one should be aware that the actual run-up height ofthe tsunami is the reported run-up height H r corrected with the tidal component.The period of the wave is in principle defined as the time it takes for twosuccessive peaks to pass a fixed point. But due to the complex wave forms it isnot always convenient and two zero successive down crossings can be taken or

even a more artificial definition has to be used. The tidal recording at Port Orfordof the 1994 Kuril Island tsunami (Figure 3.11) clearly shows the probleminvolved in distinguishing the wave period of a tsunami wave in a tidal recording.

Figure 3.11 Tide gauge recording at Port Orford of the 4 October 1994 Kuril Island tsunami

L : wavelength (m)H0 : wave height at deep

water (m)d : water depth (m) : wave elevation (m)h : water elevation (m)H : wave height (m)Hr : wave run-up (m)


29/220


30/220

05-01-2004 Part 1 - Chapter 3 - 11

Nevertheless several aspects of solitary waves and N-waves correspond quitewell; the solitary wave and the N-wave both preserve their wave form, and the

propagation is comparable. But N-waves can become steeper than solitary waves

before they break.This makes that the older work based on solitary waves can still contain a lot ofvaluable information.

Some older work can give quantitative insight in the behavior of tsunami wavesand can be used for a preliminary approach to calculate refraction (based on thelinear wave theory), behavior of tsunamis at abrupt depth transitions, wavereflections, resonance phenomena and many more tsunami wave related

phenomena. A well-documented overview of these theories and their validity isgiven by Camfield (1980).

3.2.4 Numerical calculations

Nowadays usually numerical calculations are used to describe tsunami waves. Thestandard equations used to model tsunami waves are the shallow water equations(SW).The SW equations describe the evolution of the surface elevation and of thedepth-averaged water particle velocity of waves with large wave lengthscompared to the water depth. The equations assume that the pressure distributionis hydrostatic everywhere, i.e. there is no variation of depth of any of the flowvariables other than the hydrostatic pressure. The SW equations are valid whenL>>d, , as a practical criterion L>20.d is used.One general form of the SW equations is given below:

The equation of momentum can be written as (Satake, 1995):

( . ) f V V V

V V g C t d

+ = +

(3.3)

The equation of continuity can be written as:

}{( ) ( )d d V t

+ = +

(3.4)

V : depth averaged horizontal velocity vector (m/s)

: water elevation (m)d : water depth (m)C f : nondimensional friction coefficient (-)

g : gravitational acceleration (m/s)


31/220


32/220

05-01-2004 Part 1 - Chapter 3 - 13

than solitary waves before they break (Synolaksis, 2002). So, conveniently the presented breaker criterion (equ.(3.5)) can be regarded as a lower limit for breaking of tsunami waves, as N-waves break later than solitary waves.

3.3 TSUNAMI MODELING

Figure 3.13 Snapshots of a computer model of the 12 July 1993 tsunami in the

East Japan Sea (Byung, 2002). Notice the initial water displacement (00 hours)and the waves radiating outward and hitting the coasts (01 hours).

After which reflection occurs (02 and 03 hours).

Scientists involved in tsunami research have always been very interested inmodels, which can accurately predict the propagation of tsunamis. One of themost important reasons was and still is providing the necessary data for tsunamiwarning systems: an earthquake at A with a certain focal depth and magnitude,will this generate a tsunami, which causes serious inundation of the coastline atB? Another area of great interest is to model the extent of the inundation at aspecific coast.

In this study the techniques used for modeling will be used to derive the designtsunami from deeper water to the two proposed harbor locations.

Most models start from an initial water displacement in deep water and then use a


33/220

Part 1 - Chapter 3 - 14 05-01-2004

2D hydraulic model to derive the waves to the coast. The coast is mostly modeledas a closed boundary, causing reflection at the coast. Sometimes also the coastalzone is modeled in order to model the extent of the inundation.

Severe tsunami events are modeled in order to efficiently calibrate the models andto constantly improve the mathematical techniques used in tsunami engineering.The input for these models is then provided by the derived initial waterdisplacement for the respective tsunami event. Calibration of the model is done byusing the limited available tsunami data; tide gauge recordings, arrival times onshore and reported run-up heights. In the paragraphs below, a more detaileddescription is given on the modeling of a tsunami event.

3.3.1 Tsunami source modeling

The input for a tsunami propagation model can be provided in two ways; bysolving the inverse tsunami problem or by using an earthquake fault model.

Solving the inverse tsunami problem: readings of tide gauges are transformed bymeans of a 2-D hydraulic model to the rupture area and an initial wave profile isderived for this area.When the initial wave profile is known, the 2-D propagation/inundation modelcan be run.This method, solely based on tide gauges readings, gives quite some practical

problems in the implementation and obviously does not give very accurate results.

The most accurate input for a tsunami propagation/inundation model is provided by an earthquake fault model. The tsunamis are treated as gravity waves excited by the displacement of a large volume of water. The displacement of the oceanssurface follows the vertical displacement of the seabed if the length of the ruptureis at least three to four times the water depth (Figure 3.2). Generally the speed, atwhich the displacement of the sea bottom is modeled is taken instantaneous. Theinstantaneous assumption is based on the fact that tsunamis propagate at speeds upto 200 m/s while seismic waves cause rupture to propagate at typical speeds of 2to 3 km/s, so the seafloor motion is much faster than the speed of the tsunamis

(Synolaksis, 2002). There is little question between tsunami specialists, that thetiming of the seafloor deformation by earthquake generation tsunamis can betaken instantaneous (Synolaksis, 2002).

Figure 3.14 Schematic view of the bottom deformation due to the earthquake rupture

and the resulting bottom deformation.


34/220


35/220

Part 1 - Chapter 3 - 16 05-01-2004

3.3.3 Calibration

The results of the model are compared with the scarce available data of thetsunami event: tide gauge registrations and reported run-up heights.Gathering accurate data is always a problem, for recent tsunamis internationalsurvey teams are dispatched a few weeks after the tsunami. One of their mainobjectives is to collect tide gauge recordings and run-up data.The tidal registration is the most accurate data available for calibration; it providesinformation about the wave height, the waveform and period. But most tidegauges are located in or near harbors so the complex layout of the harbor should

be taken into account.Run-up measurements are not so accurate as small variations in topography cancause considerable variations in run-up height (Baptista, 1993) and for a strip ofcoast of a hundred kilometer usually only at ten or fifteen locations run-up heightsare measured. Therefore the reported run-up height should only be regarded as arough indication. This means that comparing the run-up height calculated from thereflected wave height with the reported wave height can only give a check for theorder of the run-up height, as both the calculated and the reported run-up heightare not very accurate.In case of inconsequences between the calculated data and the reported data, themodel has to be adapted. Adaptations can be done at the side of the tectonicsource: the length, width and the height of the bottom deformation can be adaptedkeeping the seismic moment constant (as the seismic moment is a measure for

the energy of an earthquake). Another possible adaptation is refining the of the bathymetry near the coast.

The seismic moment of an earthquake is given by :. . .o M u LW = (Nm)

where is the rigidity around the fault, u is the average slip, L is the fault length and W is the fault width.The seismic moment of the earthquake is a measure for the energy of the earthquake. Thus by keeping theseismic moment constant the released energy is kept constant (Appendix 2).


36/220

05-01-2004 Part 1 - Chapter 3 - 17

3.3.4 Summary and comments

Figure 3.15Schematic overview of the modeling of a tsunami event.

Hydraulic models for modeling the propagation of tsunami waves are developed

by using recent well-documented tsunami events. A schematized overview of themodeling of tsunami events is given in Figure 3.15 .The modeling of such atsunami event starts form the initial bottom deformation as output of aseismological model. The initial water displacement equals the initial bottomdeformation. The initial water displacement is used as input for a hydraulic model

based on the shallow water equations. The calculated data is then compared withavailable data (tide gauge recordings and run-up heights), the comparison is thenused as a feedback for calibration. Adaptations can be done of the bottomdeformations (keeping the seismic moment constant) and in a refining of the

bathymetry.

It is clear that the calculations made with such models are not extremely accurate;the wave height is not represented with an accuracy of centimeters or a decimeter,

but more in the order of a few decimeters to half a meter.

earthquake data

seismologicalsource model

2D-mathematical model

Tsunami data

bathymetry

input comparing

feedback forcallibration

Bottomdeformation

Wave formHeightPeriod

Reflected wave height

1e Input 2e Calculation 3e Feedback


37/220

Part 1 - Chapter 3 - 18 05-01-2004

3.4 TSUNAMI SCALES

At first the most common tsunami scales are presented, together with a shortcritical reflection. Secondly some critical comments from a hydraulic engineering

point of view are formulated: do these tsunami scales carry enough information todescribe tsunami waves and to compare different tsunami events.

3.4.1 Scales

A short review will be given of the most common tsunami scales, based on work

of Satake (2002), Hatori (1995), Bryant (2001) and Abe (1983). These scales areused to quantify tsunami events.

There are two distinct measures for describing tsunamis: intensity and magnitude.The intensity gives an indication of the strength of the tsunami at a given locationand its magnitude is an indication for the total energy of a tsunami .

Values of historical tsunami events can be looked up in publications concerning asingular event or in tsunami catalogues. The most complete and up to datecatalogue at this moment is the Historical Tsunami Database for the Pacific, 47

B.C. to Present, which is the joint project of IUGG/TC and ICG/ITSU . Thisdatabase can be accessed at http://tsun.sscc.ru/htdbpac .

3.4.1.1 Imamura-Ida scale, m

2 ,maxlog r m H = (3.6)

m : Imamura-Idas tsunami magnitude scale (-) H r,max : the maximum observed run-up height (m)

This is the traditional tsunami magnitude scale, designed by Imamura in the midfortys. Because of the obvious strong local influence by using the H r,max , themaximum observed run-up height, it is nowadays regarded as an intensity scale.

With a simple example one can illustrate very clearly the distinction between the two concepts:intensity and magnitude. Imagine two cars crashing against a monolithic wall with the same energy. One car being a Duck andone car being a solid Volvo. The Duck will obviously have more damage than the Volvo. The damagelevel of the car can be regarded as an intensity. The intensity of the crash with the Duck is greater thanthe intensity of the crash with the Volvo, whereas the magnitude (being a measure for the energy) of

the two crashes is the same. IUGG/TC: the tsunami commission of the International Union of Geodesy and Geophysics

ICG/ITSU: International Coordination Group for the Tsunami Warning System in the Pacific


38/220

05-01-2004 Part 1 - Chapter 3 - 19

The maximum run-up height is usually reported along the coast near the tsunamisource.It is especially convenient for older tsunamis, of which no instrumental records

exist, and, of which the evidence can only be found in eye-witness reports, news paper articles or old stories.In order to get an idea of the physical meaning of the Imamura-Ida scale, therelationship between the Imamura-Ida scale ( m) and the maximum run-up ( H r,max )is given in Table 3.1 and. Figure 3.16 .

Tsunamiintensity, i

Mean run-up height(m)

-3.0 0.1-2.0 0.2-1.0 0.350.0 0.71.0 1.42.0 2.82.5 4.03.0 5.63.5 8.04.0 11.34.5 16.0

Imamura-Ida scale, m

0

5

10

15

20

25

- 3 - 2 - 1 0 1 2 3 4

m (-)

m a x

i m u m

r u n - u p

h e

i g h t ( m )

Table 3.1 Imamura-Ida scale, m

Figure 3.16 Imamura-Ida scale, m

3.4.1.2 Tsunami intensity scale, i

Soloviev (1970) was the first to point out the inappropriateness of the termtsunami magnitude as used in the Imamura-Ida scale:The grades of the Imamura-Ida scale must be designated as the intensity of thetsunami and not its magnitude. This is because the latter value must characterizedynamically the processes in the source of the phenomena and the first one mustcharacterize it at some observational point

Soloviev also pointed out that it is more appropriate to use the mean run-up( H r,mean ) at one stroke of coast in stead of the maximum run-up ( H r,max ). Thedifferences between mean and maximum run-up height can mostly be attributed todifferences in bathymetry and topography of the regarded strip of beach.He then defined the tsunami intensity scale i as :

( )2 ,log 2. r meani H = (3.7)

i : tsunami intensity (-) H r,mean : the mean run-up height on the coast nearest to the source (m)


39/220

Part 1 - Chapter 3 - 20 05-01-2004

For a tsunami event the maximum intensity ( i) on the coast nearest to the source isused to quantify the tsunami source. Soloviev did not state how long the regarded

strip of beach should be, so the definition of the mean run-up height is stillsomewhat ambiguous.

Comparison of Eqs (3.6) and (3.7) suggest that the mean tsunami run-up height isgiven as 1/ 2 times the maximum height (Satake, 2002).

In order to get an idea of the physical meaning of the tsunami intensity scale, therelationship between the tsunami intensity ( i) and the mean run-up ( H r,mean ) isgiven in Table 3.2 and Figure 3.17.

Tsunamiintensity, i

Mean run-up height(m)

-3.0 0.1-2.0 0.2-1.0 0.350.0 0.71.0 1.42.0 2.82.5 4.03.0 5.63.5 8.04.0 11.34.5 16.0

tsunami intensity scale, i

0

5

10

15

20

- 3 - 2 - 1 0 1 2 3 4

i (-)

m e a n r u n - u p

h e

i g h t ( m )

Table 3.2 Tsunami intensity scale, i

Figure 3.17Tsunami intensity , i

3.4.1.3 Tsunami magnitude, M t

Both the Imamura-Ida scale, m, and the tsunami intensity scale, i, are based on therun-up height of the tsunami. Gathering accurate run-up data is not always easy.Most data were not recorded for scientific analysis. Besides, tsunamis propagatingtowards the coast undergo a complicated combination of dynamic processesstrongly related to the local site conditions, which can not be completelyunderstood without intensive study and modeling. As a result a quantitativeanalysis is inevitably accompanied by a large scattering of data.To overcome this problem an other scale is needed.

Abe (1983) defined a scale based on the maximum tsunami wave amplitude, H , asrecorded on tidal gauges. Its original intended use was to estimate the moment

magnitude, M w, of historical large earthquakes from available tsunami data.Therefore Abe used the good correlation between M w and log H .


40/220


41/220

Part 1 - Chapter 3 - 22 05-01-2004

C

Source Region Honolulu Hilo California Japan Aleutian

Peru, Chile +0.2 -0.6 +0.2 0.0 +0.2

Alaska, Aleutian +0.5 0.0 +0.2 +0.3 -

Kamchatka, Kurile, Japan 0.0 -0.4 +0.1 -0.2 -0.2

whole region +0.2 -0.3 +0.2 0.0 0.0

Table 3.3 Values of C for different source regions and registration regions,

as determined by Abe 1983

At present, few values beyond these given in Table 3.3 have been set, so it is not

possible to give regressed values of M t for all tsunami events .

The above formula (3.10) was calibrated with the moment magnitude, M w,(Bryant, 2001) of earthquakes. Therefore, the average tsunami magnitude, M t, fora coastline is in the same order as the M w-value of the source earthquake. Formost earthquake generated tsunamis the correspondence between M t and M w doesnot differ more than 0.2, nevertheless M t and M w can still differ considerably. Themost important disturbance factors for M t not corresponding with M w are:- the local bathymetry: this can seriously affect the recorded wave height H,

although averaging Mt, as calculated from the different tidal recordings, almostcompletely eliminates this effect;

- underwater landslides, triggered by earthquakes can amplify the initial waterdisplacement, thus giving a higher recorded wave height;

- accredited sediment wedges in the generation area can amplify the initial waterdisplacement, thus giving a higher recorded wave height.


42/220

05-01-2004 Part 1 - Chapter 3 - 23

3.4.2 Comments from a hydraulic engineering point of view

A tsunami wave train consists of a small number of waves, usually one of thesewaves is more pronounced. This dominant wave can be used to schematize thetsunami load. For describing the dominant wave essential parameters are the waveheight, the wave length and the form of the wave (not broken, bore or breaking) .

Do the three presented scales carry sufficient information to derive thedominant tsunami wave?

The Imamura-Ida scale, m, and the tsunami intensity scale, i, are both based on therecorded run-up. The run-up of a wave is strongly influenced by the localtopography and the bottom friction (influenced by vegetation and buildings). Incatalogues listing tsunami events, this information is not available, it is thereforeconsidered that m and i do not carry sufficient information to link to the originaltsunami waves.

The tsunami magnitude scale, M t , is based on the recorded wave height on a tidegauge and the distance of the source. However it does not give information aboutthe water depth at the recording side, which is important for the shoaling.Moreover the M t value for one tsunami event is calculated by averaging thecalculated M t values for the different tidal gauges. This averaging filters out the

influence of the local bathymetry and geometry (e.g. amplification by funnelshaped bays) in order to give good correlation with the earthquake momentmagnitude M w. But it also makes it impossible for linking the averaged M t valuewith the wave height at one particular depth.

Also deriving information about the wave length and the wave form (not broken, bore or breaking) is not possible.

Can the three presented scales be used to compare different tsunami events?

The three presented scales were designed to quantify tsunami events, but not all

scales are equally consistent.

The Imamura-Ida scale, m, is based on the maximum run-up. Therefore this scaleis very sensitive for local bathymetry and topography, making it unsuitable tocompare different tsunami events.

The tsunami intensity scale, i, is based on the mean run-up on the nearest coast.Hereby preventing a strong influence of very local bathymetry and topography

The three parameters wave length, wave height and form of the wave (not broken, bore or breaking), arenot sufficient for an exact representation of the tsunami wave. But as the exact form of a tsunami wave is

hard to derive exactly and because each tsunami event is different, it is believed that these parameters aresufficient for representing a design load (Kamel 1970). Because of the inaccuracy involved a largesensitivity analysis should be done.


43/220

Part 1 - Chapter 3 - 24 05-01-2004

(e.g. the influence of a funnel shaped bay). Obviously the bathymetry andtopography of the regarded coast does influence the mean run-up, making itdifficult to compare different tsunami events.

However tsunami events occurring in a region with coasts with comparable bathymetry and topography, can be compared with each other, making it possibleto estimate a reoccurrence period of the tsunami events in that region.

The tsunami magnitude scale, M t , is based on the recorded wave heights on tidegauges and the distance to the source. The M t scale has a strong correlation withthe moment magnitude, Mw, of the generating earthquake.These factors make the M t scale the best base for a comparison of differenttsunami events. However an M t value is not available for all tsunami events.

Are there alternatives for deriving tsunami waves?

Since the early nineties a deepwater tsunami registration program has been set up.These recordings register the entire tsunami profile. However the registration

program only covers a small part of the Pacific. Consequently these detailedmeasurements are not available for all tsunami events.

At this moment the only possible way to get accurate information about thetsunami wave profile is by modeling the tsunami event right from the initial

bottom deformation, caused by the earthquake.


44/220

05-01-2004 Part 1 - Chapter 4 - 1

4 DETERMINING THE CONCEPT OF THE DESIGNTSUNAMI

In civil engineering the common method of defining a design load is by astatistical approach. For example defining a design water level for a dike. Thedesign water level should be higher than the already observed highest water level.When an extensive dataset of observed water levels is available, a distributionfunction is fit to this dataset. And a design water level, for a certain return period,can then easily be determined. However in reviewed tsunami literature no

examples are found of a statistically defined design tsunami. Mostly for designinga structure a large historical event is taken and the wave height and the durationare slightly increased (e.g. Kamel, 1970 - Hitachi, 1994)

At first it will be investigated if a design tsunami can be determined statistically,with the available tsunami data for Central America from Chapter 2. If this provesnot to be possible, the largest historical event will be taken and slightly amplified.

4.1 ANALYZING THE TSUNAMI DATA FOR THE PACIFIC SIDE

OF CENTRAL AMERICA

In this paragraph it will be examined if the dataset (Table 2.1 and Figure 2.1)contains enough information for a statistical derivation of a design tsunami.

4.1.1 Homogenizing the dataset

First a homogeneous dataset has to be obtained so that the different tsunami

events can be compared.Two criteria can be distinguished for the homogenization of this tsunami dataset:the generation mechanism and the location of the generation.


45/220

Part 1 - Chapter 4 - 2 05-01-2004


Year Mo Da Hr Mn Sec Lat Long Dep Ms Mw Mt Int Hr,max D F C TR Source Region

1 1906 1 31 15 36 1 -81.5 25 8.6 8.7 8.7 3 5 M 1000 T SAM Columbia-Ecu ador:TUMACO

2 1907 4 15 6 8 16.7 -99.2 25 8.2 7.8 1 2 S T CAM Mexico

3 1928 3 22 4 17 9.6 16.84 -96.02 15 7.5 7.4 T CAM MEXICO


5 1930 8 31 0 40 36.6 33.9 -118.6 0 5.2 -0.5 6.1 M L CAM S.CALIFORNIA

6 1932 6 3 10 36 54.2 19.46 -104.17 25 8.1 7.6 8.2 -0.5 0.7 N T CAM Central Mexico,Jalisco







13 1950 10 5 16 9 10.4 -85.2 60 7.7 7.7 7.8 -2 0 .1 N T CAM Nicaragua,Costa Rica-Panama

14 1950 10 23 16 13 14.3 -91.8 30 7.1 7.5 7.5 -1 0.2 N T CAM Guatemala-Nicaragua




18 1958 1 19 14 7 25 0.99 -79.48 20 7.3 7.7 1.5 S 20 T SAM Colombia-Ecu ador

19 1962 3 12 11 40 15.3 8.09 -82.68 19 6.8 -3 0.1 N T CAM Costa Rica-Panama

20 1962 5 11 14 11 56.8 17.14 -99.7 46 7.2 7 0 0.8 N 4 T CAM S.Mexico

21 1962 5 19 14 58 13.1 16.99 -99.69 16 7.1 S 30 T CAM S.Mexico

22 1965 8 23 19 46 1.5 16.17 -95.85 10 7.8 7.5 -1 0.4 N T CAM MEXICO


24 1979 3 14 11 7 15 17.7 7 -101.23 24 7.6 7.6 -1.1 0.4 N 5 T CAM S.Mexico:GUERRERO

25 1979 12 12 7 59 4.6 1.6 -79.36 24 7.7 8.1 2.5 5 M 600 T SAM Colombia-Ecu ador

26 1981 10 25 3 22 15.8 18.12 -102 21 7.3 7.2 -3 0 .1 N T CAM MEXICO



29 1988 4 30 15 42 32.6 -117.3 5 N 0 M CAM San Diego, S. California

30 1992 9 2 0 16 2.7 11.72 -87.39 45 7.2 7.7 7.9 2.8 10.7 M 170 T CAM Nicaragua

31 1994 1 17 12 30 55.5 34.16 -118.57 14 6.8 6.7 -3 0.1 N 0 T CAM Southern California




35 1999 9 12 18 17.62 -101.6 2.5 8 S 0 U CAM Guerero, Mexico

Table 4.1 Databank for the tsunamis at the Pacific side of Central America from 1900 to 2002.

the crossed out events are removed to homogenize the dataset.


46/220

05-01-2004 Part 1 - Chapter 4 - 3

Figure 4.1 The epicentral locations of the tsunamigenic earthquakes.

The crossed out events are removed to homogenize the dataset.Also the tectonic setting of Central America is given.(MAT: Middle American Trench, PFZ: Panama Fracture Zone)

4.1.1.1 The generation mechanism

In general earthquakes are distinguished as the predominant cause of tsunamigeneration. The dataset for Central America confirms this idea; 32 on a total of 35tsunamis are generated by earthquakes. Therefore the dataset will be limited toearthquake generated tsunamis.

The following tsunamis events are removed (Table 4.1):- the 1930 tsunami in Southern California (5), caused by a landslide- the 1988 tsunami in Southern California (29), caused by meteorological

circumstances- the 1999 tsunami in Mexico (35), with an unknown cause (this means the cause

was certainly not an earthquake, because even the smallest earthquake isalways detected by sensitive seismographs)


47/220

Part 1 - Chapter 4 - 4 05-01-2004

4.1.1.2 The location of the generation

The predominant cause of tsunami generation are earthquakes. Earthquakes aregenerated at plate boundaries. The type of the plate boundary affects the

bathymetry and the occurring earthquakes (Appendix 2). Therefore the generationregion will be limited with regard to plate tectonics. A similar approach is used byHatori (1995).

Tectonic setting The tectonic setting of the Pacific side of Central America (Figure 4.1) is given bythe interaction of the Cocos, Caribbean and Nazca plates (fig.1). The Cocos Platesubducts under the Caribbean Plate along the Middle American trench; fromMexico to Central Costa Rica. The limit between the Cocos and Nazca plates isthe Panama Fracture Zone (PFZ). The PFZ is composed of north-south trendingfaults located in front of the Pacific Coast of Costa Rica and Panama. The

boundary between the Caribbean Plate and the Nazca Plate is still ambiguous,some authors consider it as a subduction zone and others as a strike-slip fault.

Most earthquakes are related to the Cocos-Caribbean subduction zone, defined bythe Middle American trench. The Middle-American Trench is from 4 to 5 kmdeep (Shah et. al., 1975) and extends, approximately 130 km off the CentralAmerican Pacific coast (Montero, 1990), from Mexico to Costa Rica.Research of Fernandez (2000) reveals that 93% of the total released moment, wasreleased along the Middle American Trench. The earthquakes can be separated intwo populations, shallow depth (0-30km) and intermediate depth (40-200km)earthquakes. There are no earthquakes deeper than 200km.

Because shallow and intermediate earthquakes along subduction zones, have the biggest possibility of causing vertical displacement of the ocean bottom, theseearthquakes are mostly held responsible for tsunami generation. Therefore thedataset will be limited to earthquakes generated along the Middle AmericanTrench

The Panama fracture zone is an ocean ridge. Ocean ridges primarily generate

earthquakes without predominant vertical displacement, which is necessary fortsunami generation.Therefore it is somewhat ambiguous if the earthquakes of 1934 (9) and 1962 (19)in Costa Rica-Panama are generated on the middle American trench or on thePanama fracture zone, these tsunami events are kept in dataset.

The following tsunami generating earthquakes are removed (Table 4.1):- the 1906 earthquake in Colombia-Ecuador (1)- the 1930 earthquake in Southern California (5)- the 1958 earthquake in Colombia-Ecuador (18)- the 1979 earthquake in Colombia-Ecuador (25)

- the 1988 earthquake in San Diego (29)- the 1994 earthquake in Southern California (31)


48/220

05-01-2004 Part 1 - Chapter 4 - 5

Comments

It is clear that tsunamis generated in Southern California (5, 29 and 31) do notaffect the coast bordered by the Middle American Trench. But the tsunamisgenerated at the Columbian subduction zone can obviously affect this coast. Thiswere the 1906 Columbia-Ecuador tsunami (1) with intensity 3, the 1958Columbia-Ecuador tsunami (18) with intensity 1.5 and the 1979 Columbia-Ecuador tsunami with intensity 2.5. These tsunamis are so called regionaltsunamis as they are generated at a certain distance of the Pacific coast of CentralAmerica. This means that the large intensities due to the nature of the intensityscale ( defined for the mean run-up on a nearest coast , see 3.4) have nosignificance for the Central American coast. They 1906 tsunami was observedalong the entire coast of Central America, but there is no specific reference madeof large damage along this coast due to this specific tsunami (Fernadez, 2000).Therefore for statistical calculations these three tsunamis are still kept out of thedataset.


49/220

Part 1 - Chapter 4 - 6 05-01-2004

4.1.1.3 Resulting dataset

By restricting the generation cause to earthquakes and the generation area to theMiddle American Trench, a final dataset is obtained with 28 tsunami events forthe period 1900-2002.



2 1907 4 15 6 8 16.7 -99.2 25 8.2 7.8 1 2 S T CAM Mexico

3 1928 3 22 4 17 9.6 16.84 -96.02 15 7.5 7.4 T CAM MEXICO















20 1962 5 11 14 11 56.8 17.14 -99.7 46 7.2 7 0 0.8 N 4 T CAM S.Mexico

21 1962 5 19 14 58 13.1 16.99 -99.69 16 7.1 S 30 T CAM S.Mexico

22 1965 8 23 19 46 1.5 16.17 -95.85 10 7.8 7.5 -1 0.4 N T CAM MEXICO



26 1981 10 25 3 22 15.8 18.12 -102 21 7.3 7.2 -3 0 .1 N T CAM MEXICO



30 1992 9 2 0 16 2.7 11.72 -87.39 45 7.2 7.7 7.9 2.8 10.7 M 170 T CAM Nicaragua




Table 4.2 Homogenized dataset


50/220

05-01-2004 Part 1 - Chapter 4 - 7

Figure 4.2Homogenized dataset.

4.1.2 Do the available data for these events contain enoughinformation for deriving a wave or a wave train?

A homogenous dataset was obtained, which enables us to make statisticalcalculations. It is now possible to take a parameter, fit a distribution to it andextrapolate it for a certain return period. But the question is whether theseavailable parameters can be coupled to wave characteristics? This is examined

below. First it is described how the wave can be schematized. Afterwards it is

described what can be concluded out of the available tsunami data and seismic parameters.

4.1.2.1 Tsunami wave vs. tsunami wave train

A tsunami is a wave train consisting of a small number of waves, usually less the10, before reflection occurs. Usually one of the waves is a lot more pronouncedthan the others (Murty, 1977).In literature usually one wave is used to characterize the tsunami load, as well intheoretical reflections as in experiments (e.g. Kamel, 1970 Ramsden, 1996

Silva, 2000 Tanimoto, 1981).Therefore for design purpose it is believed to be sufficient to schematize the


51/220

Part 1 - Chapter 4 - 8 05-01-2004

tsunami to this one dominant wave, characterized by its wave height, length andthe form of the wave (not broken, bore or breaking) (3.4.2). So it is sufficient to

derive this one dominant wave.

4.1.2.2 Tsunami data

The relevant tsunami data available in the dataset are the maximum run-up( H r,max ), the tsunami intensity ( i) and for some tsunamis the tsunami magnitude( M t ). in paragraph 3.4 an extensive overview of the scales is given. It is concludedthat nor the maximum run-up ( H r,max ), nor the tsunami intensity ( i), nor thetsunami magnitude ( M t ) can be used to derive a tsunami wave

.The tsunami intensity ( i) can be used to compare different tsunami events as thecoast near to the Middle American Trench is assumed to have a similar

bathymetry and topography along the entire coast. The best base for a comparisonis the tsunami magnitude scale ( M t ), but the M t values are only available for 8tsunami events.

4.1.2.3 Earthquake scales

The available earthquake data are the depth of the hypocenter ( Dep ), the surfacewave magnitude ( M s) and the moment magnitude ( M w). This data cant be coupleddirectly to wave parameters and they contain insufficient information for derivingan ocean bottom deformation.

4.1.2.4 Conclusion

The effect of a tsunami wave train can be represented by one dominant wave,characterized by its wave height, wave length and the form of the wave (not

broken, bore or breaking).The data available of the tsunami events contains insufficient information to linkto wave characteristics. Therefore a statistical approach for deriving a design

tsunami out of the available data is not possible.As a statistical approach proves not to be possible, the largest tsunami event in theCentral American region will be identified. The largest event, will then be takenas a basis for a design tsunami; the wave height and the period will be increasedwith 20%.

The conclusions concerning the possibility of linking the tsunami data and the seismic data to tsunamiwave parameters, were the result of an extensive investigation. It was concluded that linking the limited

available data to wave parameters is not possible. However due to the importance of this conclusion forthis study, the expert opinion was asked of some tsunami researchers. They were contacted by mail andthere replies confirm my conclusions. These mails are shown in Appendix 4.


52/220


53/220

Part 1 - Chapter 4 - 10 05-01-2004



2 1907 4 15 6 8 16.7 -99.2 25 8.2 7.8 1 2 S T CAM Mexico















20 1962 5 11 14 11 56.8 17.14 -99.7 46 7.2 7 0 0.8 N 4 T CAM S.Mexico

22 1965 8 23 19 46 1.5 16.17 -95.85 10 7.8 7.5 -1 0.4 N T CAM MEXICO



26 1981 10 25 3 22 15.8 18.12 -102 21 7.3 7.2 -3 0 .1 N T CAM MEXICO



30 1992 9 2 0 16 2.7 11.72 -87.39 45 7.2 7.7 7.9 2.8 10.7 M 170 T CAM Nicaragua




Table 4.3 Final dataset


54/220

05-01-2004 Part 1 - Chapter 4 - 11

Figure 4.3Final dataset

4.2.1 Occurrence

The occurrence of tsunamis with a certain intensity is shown in Table 4.4. Alsothe mean run-up heights (H r,mean ) corresponding with the tsunami intensity (i)(equ 3.7) are shown.


55/220

Part 1 - Chapter 4 - 12 05-01-2004

i H r,mean (m) Number of

tsunamis4 11,3 0

3,5 8,0 03 5,7 1

2,5 4,0 02 2,8 0

1,5 2,0 41 1,4 3

0,5 1,0 10 0,7 2

-0,5 0,5 1

-1 0,4 5-1,5 0,3 0-2 0,2 3

-2,5 0,1 0-3 0,1 6

Table 4.4 Tsunami intensity versus occurrence.

(the intensities are rounded of to a half)

Tsunam i Data

0

2

4

6

8

43210-1-2-3

i

n u m b e r

Figure 4.4

Tsunami intensity versus occurrence

Probably the observation of the small intensity tsunamis (i


56/220

05-01-2004 Part 1 - Chapter 4 - 13

Figure 4.5

Tsunami exceedance probability versus intensity for tsunamis with i 0 for theCentral American region from 1900 to 2002

In Figure 4.5 the tsunamis with an intensity i 0 are plotted against theirexceedance probability. This gives for the 1992 Nicaraguan tsunami (i = 3) areturn period between 85 and 850 years. The best fit distribution function gives an

return period of 120 years for the coast bordering the Middle American Trench.As only local tsunamis occurred in this region, only a small part of the coast isaffected. In the case of the Nicaraguan tsunami only 250 km of coast wasseriously affected (Bryant, 2001) of the total of 4,000 km along the MiddleAmerican Trench. This is a normal value for tsunamis of such a large intensity.

Figure 4.6 A tsunamigenic earthquake, similar to the 1992 tsunami event, can only affect a

certain location at the coast as the epicenter lies in the zone in front of the location.

As for the 1992 Nicaraguan earthquake the coastal region affected was 250km, thelength of the regarded strip in which the epicenter has to lie is 2x125 km.


57/220

Part 1 - Chapter 4 - 14 05-01-2004

To get an impression of the occurrence at one location along the coast, the Middle

American Trench can be divided into strips of 250 km (Figure 4.6). The chancefor a tsunamigenic earthquake, as in Nicaragua1992, can be taken uniformlydistributed over the 16 cells of the Middle American Trench. Thus the chance thatthe once in a 120 year tsunami affects one exact location, i.e. occurs in one of the16 cells, is once in 1920 years. Taking into account the spreading, then the return

period for one location lies between 1360 and 13600 yearsClearly this is only a rough estimation of the return period. The tsumanigenicearthquakes probably are not uniformly distributed along the Middle AmericanTrench (Figure 4.3), most tsunamigenic earthquakes seem to occur in Mexico. Butin order to give account for such heterogeneities a dataset for a longer period andmore detailed insight in geological processes is needed.

Nevertheless this simple calculation clearly illustrates the infrequency of largetsunami events at the Pacific coast of Central America.

4.2.2 Compared to the entire Pacific Region

Compared to the entire Pacific region, one could wonder if this is what should beexpected for Central America .Looking at the entire Pacific region ( Table 4.5 ) forthe period of 1900 to 2002 only 23 tsunamigenic earthquakes with an intensity i 2.8 occurred. So Table 4.5 clearly shows that large tsunami events are infrequent

phenomena. Most of these large tsunamis only affected a few hundred kilometersof the nearest coast. Figure 4.7 shows that in the North Western part of the Pacific(Japan, Kamchatka and Alaska) more than half of the large tsunami events weregenerated. Therefore this is the most important generation area for large tsunamis.This has two reasons: first of all this region consists of very active seismicsubduction zones and secondly very local amplification effects can occur. Thesevery local amplification effects are attributed to local bathymetry such as bays orstraits or large Archipelos (Ida, 1981). Because these two conditions are fulfilledfor the North Western Pacific, most of the large tsunamis occur in this region.Figure 4.7 is consistent with Figure 2.1, where the percentages of tsunamigenicearthquakes for all the regions in the Pacific are displayed. Also Figure 2.1 clearly

shows that Central America is a rather modest region for tsunami generation(6.5%).

The maximum run-up heights of tsunami events comparable to the Nicaraguantsunami (i ~ 3), lie mostly between 10 and 15m. Sometimes very local bathymetryeffects can seriously increase the maximum occurred run-up, e.g. the 1964tsunami in the gulf of Alaska (H rmax =67m). The enormous maximum run-upheight of the 1958 tsunami in South East Alaska is caused by a massive rockslidetriggered by an earthquake in a confined bay (Bryant, 2001).Figure 4.7 and Figure 2.1 clearly show that Central America is a rather modestregion for tsunami generation. However large tsunamis, like the 1992 Nicaraguan

tsunami, do occur and they have to be taken account for.


58/220

05-01-2004 Part 1 - Chapter 4 - 15

Figure 4.7

For each region in the Pacific the number of generated large tsunamis (i 2.8) areplotted. (based on Table 4.5)

Year i H max Source Region1900 3 Bismark Sea,New Guinea1906 3 5 Tumaco, Colombia-Ecuador1918 3 12 E.Urup,S.Kuril Islands

1923 3 8 SE off Kamchatka1923 3 14 E.Kamchatka peninsula.USSR1923 3 13 Sagami Bay, Tokaido, Japan1927 3 15 Celebes Sea.Indonesia:SULAWES1933 3.5 29.3 Sanriku, Japan1939 3 10.5 Solomon Islands Papua New Guinea

1946 4 35 Unimak Is.,E.Aleutian Islands1946 3 30 East of Vancouver Is., Canada1952 4 18.6 SE of Kamchatka peninsula

1957 3.5 16.2 Central Aleutian Islands1958 3.5 525 Lituya Bay, SE Alaska1960 3 0.25 Peru: Arequipa,Chuquibamba,ca1960 5 25 Corral, Southern Chile

1963 3 15 Urup,Kuril Islands.USSR1964 4.5 67.1 Gulf of Alaska, Alaska pen.1968 3 10 N.Celebes, Banda Sea.Indonesia

1969 3 15 Kamchatka,Bering Strait1975 3 14.3 Kalapana, Hawaii Is.USA1977 3.5 15 Sunda islands.Indonesia1992 2.8 10.7 Nicaragua

Table 4.5Earthquake generated tsunami events with an intensity i 2.8, for the entire

Pacific region, for the period 1900-2002 (data compiled out of the Histori

breakwater stability under tsunami attack-for a site in nicaragua.pdf

Documents