2

Modelling and Simulation of Tsunami Wavesfor Improved Defence Provision in Phuket

สวั�สดี�ครั�บHello

from Phuket, Thailand

Richard N. ZobelDepartment of Computer Engineering

Faculty of EngineeringPrince of Songkla University, Phuket Campus

80 Vichit Songkram Road, KathuPhuket, Thailand 83210

3

Introduction

The author was working in Phuket from three days after the tsunami for the following two and a half months.

He was able to observe the local effects at the beaches around Phuket, the largest island in the Andaman Sea, off the South West coast of Thailand, and took pictures and video.

The situation at each beach is unique and relates to both local factors and the magnitude, phase and direction of the sea wave in relation to the local geography of both the sea bed and the beach area.

Some beaches were badly damaged and others less so, some were badly damaged at the ends of the beaches. Others suffered because the land behind the beach was very flat, and some areas were buffered by the mangrove forest.

4

Introduction

Photographic and video evidence acquired at the time and later.

Used in relation to research for key factors responsible for the

damage, loss of life and injury for this specific tsunami.

Considers possible future tsunami with different characteristics.

Uses simulation to predict the consequences for affected areas.

Possible defences which minimise the effects of future tsunami

are discussed.

Conclusions for those who may need to consider these defences.

Emphasis is on how modelling and simulation can be used to study

the possible effects of future tsunami with respect to affordable

provision of defences at the sites likely to be affected.

5

The Tsunami in Phuket, Thailand

The events in Phuket on Sunday, 26 December:(Thailand is 7 hours ahead of England)

07.59 An earthquake shook Phuket Island. It was also felt in the Capital, Bangkok; in the North West, at Chiang Mai; and in the South East, at Hat Yai.

09.45 The Tsunami arrived on the West side ( Andaman Sea) 09.55 The second wave arrived. Major damage occurred in only ten minutes in some

places10.00 The waves reduced and gradually subsided

Thailand: 5,303 Dead; 3,396 Missing; 8,457 Injured

Map of South East Asia

Earthquake

started here *

Hat Yai

7

The Areas Affected

Phuket is an island off the South West coast of Thailand.It is about 50km North to South, and about 25 km wide.(about the size of the Isle of Wight or Singapore)The main tourist beaches are on the Andaman Sea (West) side

of the Island.

The mainland to the North of Phuket is called Phang NgaPhang Nga is very flat and damage here went as far as 2km

inland.

The mainland to the East of Phuket is called Krabi.This area was less affected as it was partly shielded by Phuket.

8

The Provinces of South Thailand

Andaman Sea

Gulf of Thailand

Malaysia

9

Phuket Island

Patong Beach

Kamala Beach

Surin Beach

Phuket Town

Prince of Songkla University (PSU)

Airport

Nai Yang Beach

10

Tsunami Scenario

11

Tsunami Simulation

The Sea Has Gone Away!

The CD of the Tsunami arriving at Patong Beach shows

some very interesting and illuminating features:

1. The sea has gone away

2. The Tsunami is coming

3. The Tsunami approaches the beach

4. The angry sea

5. Chaotic flows

6. The whirlpool

13

What is Being Done

The Smashed Buildings, Restaurants, Shops and BeachFacilities are now cleared up and the rubbish sorted forre-cycling, removed and stored.

Re-building is underway, and many businesses and facilitiesare already open. However, there are not many tourists yet at Patong Beach,the main tourist beach.

The pictures of the beaches were all taken by the author on

Sunday 23 January, four weeks after the Tsunami

Patong Beach – Undamaged Part of Hotel Resort

15

Patong Beach – Damaged Part of Hotel Resort – Under Re-construction

16

Patong Beach – Clear, Clean and Empty.

17

A Bar at Kamala Beach – Only the tiled floor remains.

18

A Restaurant at Kamala Beach – Already under Re-construction

19

What is Being Done – Re-Building at Kamala

20

Surin Beach suffered only minor damage

21

Surin Beach has Tourists

22

Bamboo Huts at the Restaurant are not Damaged

23

Fishing Boats

24

Damaged Boat – Nai Yang Beach

25

Boat Repair – Nai Yang Beach

26

Bar/Restaurant Repair – Nai Yang Beach

27

Samos, September 2005 - Tsunami here 50 years ago!

28

PLATE TECTONICS

In the situation concerning the recent major event, the Australian-Indian plate is moving towards the Eurasian plate and has been in collision with it for millions of years.

More specifically, the Indian plate has been pressing against the Burma (Myanmar) micro plate.

The pressure has built up over a long period of time, how long is not known (100-300 years).

It was eventually partially released on 26 December 2004 in a large movement, Richter scale 9.3, which saw the Indian Ocean floor move some 15m or more towards Indonesia.

29

The Major Tectonic Plates of the World

30

The 90 East Ridge

There is an ocean ridge, running in a North/South direction, roughly along the 90o East line of longitude, all the way from Australia to India, a distance of some 5000km.

Associated with the zone where the Indian ocean sub-plate meets with the Burma continental micro-plate, there are several deep trenches, notably the Java Trench, depth around 7km, and the Sumatra Trench, depth around 6km.

This is a very ancient structure and is the site where the Indian Ocean floor slides under the Burma Micro Plate.

This type of plate collision is called a Subduction Zone

31

Subduction Zone, Trench and Island Arcs

32

Phuket, Phang Nga and Ranong

Most of Phuket Island is hilly, so perhaps only a few percent of the area of this beautiful island was damaged.

Sadly, Phang Nga, the mainland to the north of Phuket is much flatter, resulting in extensive damage and loss of life.

A marine police launch was found beached, some 2 km inland. Many tourists were living in tents. They had little chance and nowhere safe to go.

In some of the Sea Gypsy villages, the elders had a tribal memory which suggested that if the sea goes out a long way suddenly, it will quickly come back inland as far as it went out, so go up hill fast before it does. Those that did survived, although their villages did not.

In Phuket, Phang Nga and Ranong, many villages lay behind mangrove forest and consequently suffered only minor damage. Others have had their mangrove forest replaced by fish and prawn farms, and were not protected.

It is very clear that the system for whichmodelling and simulation is to be attempted isextremely complex and highly non-linear.

Further, verification and validation is almostimpossible, since any data is likely to besparse and incomplete.

Predicting the occurrence and/or magnitude ofsuch events is also extremely difficult.

MODELLING AND SIMULATION

Such is the advance in technology it is now possibleto measure the precise position of the land and of thedepth of the sea from satellites.

This enables sea depth and boundaries to be established.

So, at least in principle, the problem is computableusing the appropriate three dimensional, time-varying,

non-linear Navier-Stokes partial differential equations to predict the ensuing tsunami sea waves, if the origin orepicentre of the earthquake is known.

Static Features

For the recent tsunami, it subsequently became clear that the totalinverted rupture duration was 200s and the peak slip was about 20m.

The rupture propagated as a series of earthquakes from the original epicentre towards the northwest for about 400km with a speed of 2.0km/s, along the subduction line near to the 90 East Ridge.

This substantially changed the nature of the tsunami, especially in itseffect on Phuket and the adjacent mainland areas of Phang Nga andKrabi.

Following this development, part of the tsunami wave front became a North/South line travelling East towards Sri Lanka and West towards Phuket. This is distinctly different from that emanating from a single point epicentre.

Dynamic features

This subject is complex and concerns the general Navier-Stokes three dimensional (3-D) partial differential equations for viscous, compressible flow in relation to:

Eulerian Conservation of Mass

Eulerian Conservation of Momentum

Eulerian Conservation of Energy

Modelling of Sea Waves

Eulerian Conservation of Mass:

t + U . = - .U

where: U is the particle velocity, and is the water density

= x i + y j + z k; where i, j, k are unit vectorsand U . = Ux i + Uyj + Uz k

Conservation of mass is of fundamental importance to most systems.

In this context the mass balance must include all mass arriving in a

given 3-D volume and all mass leaving that volume during acomputational time period, in addition to the mass already

present in that volume prior to that time period.

Modelling of Sea Waves

Eulerian Conservation of Momentum:

t + U . U = - . + g

= ijP – Sij

where: P is the water pressure, g is the gravitational constant, and Sij are the total viscosity deviators (e.g. Sx = qx – q), where q is the viscosity).

Conservation of momentum is a second fundamental consideration in that the systems sum of the products of individual mass elements and their velocity must remainconstant.

Modelling of Sea Waves

Eulerian Conservation of Energy:

t + U . I = - : U+ 2 T

where: : U =ji Ui/Xj I is the internal energy

is a real viscosity coefficient.

Conservation of energy is a little more tricky. However, it must be remembered that energy is

only transferred from one form to another and is consequently not lost.

Modelling of Sea Waves

• This set of equations may be simplified by assuming the flow is incompressible ( = 0 )

• The equations may be further simplified by excluding viscosity.

• Viscosity terms may be included if the fluid is incompressible, with or without shearing forces.

• Further details and a full discussion of these equations may found in Numerical Modelling of Water Waves, 2nd Edition, 2004, by Mader [6].

• This is the definitive work

Modelling of Sea Waves

• In computational terms, each of these equation sets must be converted into finite difference equations before solutions may be computed digitally.

• In order to do this, assumptions must be made in relation to the most appropriate form for the representation of differential terms to achieve an acceptable accuracy of result.

• To represent such a large volume of water, it is necessary to carefully consider the spatial resolution, which needs to be adequate to represent the spatial frequencies, or wavelengths of the expected sea waves, but also needs to minimise the computational cost. This is always a difficult compromise.

• The temporal aspects must also be considered in terms of minimum acceptable sample rate in this respect.

Modelling of Sea Waves

• It is important to appreciate that for these systems, such as water wave systems, the consequences of having relatively low damping needs to be considered in relation to the convergence of the solutions, and of their divergence with time from an initial state.

• In relation to the shallow water simulations, these aspects also need to be carefully looked at, since in shallow water the wavelength becomes considerably shorter and the wave height considerably larger. In addition the volume under consideration becomes much smaller.

Modelling of Sea Waves

It is clear that a “simple” solution of the Navier- Stokes equation withan input of a point source earthquake and tsunami will not suffice.

Further than this, the wave comes and goes with declining amplitude and is modified by obstructions, plus reflection and refraction from islands and the shoreline.

If this is not enough, each surge results in progressive pile up and/or removal of tens of metres of silts and sediments where tsunami

move in shallow waters.

This also causes concern for large ships bringing relief supplies in respect of harbour access and movement through narrow busy waterways such as the Malacca Strait. Much surveying will be necessary to enable vessels to proceed safely.

Tsunami Wave Dynamic Features

But there is more, it turns out that the sequence of earthquakes, mainly to the North of the original earthquake involved a shift of the ocean floor by perhaps 15m in a matter of minutes. The Indonesian Island of Sumatra and the Indian Nicobar and Andaman Islands may have moved

as much as 20m.

The scale is staggering; the total force involved was estimated at around 600 Megatons of TNT.

Dynamic features

Tsunami arise from: Earthquakes (Indian Ocean, Pacific Ocean, etc), Landslides (Lytuya Bay), Cavity collapses, Asteroids (Cretaceous-Tertiary Chicxulub) and other related sudden major movements of land above or below sea or water surface level.

Few can be foreseen

Prediction is difficult

Need to employ effective warning systems

Need to employ affordable defences to minimise death, injury and destruction

Possible Defences against Tsunami

In Krabi, villages which lay behind mangrove forests were far less affected than those where the forest had been removed to make way for fish farms or other commercial activities

Minimisation of the scale of mangrove forest removal could minimise the damage caused by future tsunami.

Alternatively, man-made structures might be created which in some way mimic nature in this respect.

Defence Possibilities

Defence Possibilities

Mangrove forest Phang Nga Bay

Other possibilities:

• Refuges to be constructed to provide escape to an upper story when stranded in a flat area

• Changes to the shallow seas near the beaches by dredging, or of dumping sand or rocks.

• Modification of the built environment at the planning stages.

• Changes to vehicle parking policy and provision.

Defence Possibilities

• Source of future tsunami which would affect western coasts of South Thailand, is most likely to be the subduction zone earthquakes associated with the fault line which includes the Sunda Trench, to the west coast of Sumatra, to the Andaman and Nicobar islands in the north. Other possibilities include cataclysmic explosion of volcanoes such as Krakatoa (between Sumatra and Java) and Tambora on the island of Sumbaya.

• Both single and multiple earthquakes (as for the recent events) need to be considered.

• Simulation of the effects of such earthquakes on possible tsunami generation is needed, but not easy to model.

Modelling and Simulation ofPossible Tsunami in the Indian Ocean

• Hydrological data is required to model the seafloor boundaries.

• However, the recent tsunami has caused changes to the sea floor

• The possible effects of reflections from islands and undersea rock formations also need to be considered.

• Such studies will provide samples of input data for

shallow water studies in the coastal areas.

• Full Navier-Stokes compressible water wave simulations needed for validation of simpler models used to reduce simulation costs for both deep and shallow water studies.

Modelling and Simulation ofPossible Tsunami in the Indian Ocean

• Tsunami need to be simulated for the Indian Ocean and the Andaman Sea

• Required as input to the shallow water simulation studies for the general and specific bays and beach areas of Phuket, Phang Nga and Krabi.

• Problems of change of scale and consequent spatial resolution, resolved perhaps by interpolation.

• However, there are also a number of potential problems in achieving a reasonably accurate representation of the resulting sea wave motion.

Tsunami Modelling Problems

• Need to obtain recent or current data of the sea floor hydrological data, in relation to the type of deposits and exposed rock formations as seen in video footage of the sea receded to the horizon at Patong beach.

• Need to obtain estimates of the sand and other materials picked up by the first wave. This affects the local water density, and also the way in which such mixtures of solid and liquid matter behave.

• Debris from damage caused by the first wave will be deposited, changing the sea bed data, which requires changes to the model data for the second sea wave.

• Small islands, sand banks and bars have disappeared; new islands have appeared, and beaches modified.

Tsunami Modelling Problems

Small island between Phuket and Krabi after the tsunami

Tsunami Modelling Problems

• Simulation of possible tsunami scenarios at specific beaches requires interfaces with other simulations:

• Apply to a variety of building structures with a view to determining their susceptibility to such sea waves.

• Two storey restaurant as example.

• Propose new or modified building designs.

• Avoid basements in areas where flooding can occur.

• Reconsider the built environment.

Application of Simulation Resultsto Affordable Defences

• Consider construction of multi-storey car parks to minimise the number of parked vehicles on the beach road.

• Use the same building as an escape for people to a higher level above the tsunami wave.

• Employ two or more storey buildings with ground floor supports designed to resist tsunami sea waves.

• Buildings with supporting columns are common in Thailand

• They provide storage and vehicle parking underneath and reduce the risk of entry of snakes and other undesirable visitors to domestic apartments as may be seen in the next slide.

Application of Simulation Resultsto Affordable Defences

Accommodation building, PSU, Phuket Campus, Kathu district

Application of Simulation Results to Affordable Defences

Modelling and simulation might be used for other aspects:

• Design of the cross section of the pillar supports needs to be optimised to provide minimum resistance to sea waves, in order to maximise the possibility that upper storeys survive a tsunami attack.

• After the tsunami waves had initially receded, the sea took on a violent chaotic appearance and that this was replaced by a large whirlpool. Perhaps these damaging activities could have been ameliorated by modifying the sea floor close to the beach by dredging or by dumping sand or rock to damp out this motion.

• The possibilities of creating moles out to sea to reduce access of large sea waves to the beach areas, as can be seen at Schevningen, in the Netherlands and at other places, should be considered.

Application of Simulation Resultsto Affordable Defences

• A tsunami warning system to enable coastal areas of the Indian Ocean likely to be affected is under construction.

• The flat, low lying areas of Khao Lak in Phang Nga, Thailand and Aceh, in Sumatra, Indonesia provide little in the way of possible escape to a higher level above the invading sea.

• This paper considers the possibilities of the use of computer modelling and simulation of sea waves to determine the feasibility of making economic defence provision for affected areas in Thailand, which may also be relevant in other countries.

Conclusions

• The paper outlines the stages of the work required, and highlights some of the likely major technical difficulties in carrying out such a novel study.

• Perhaps the most important of these difficulties lies in the problems of interfacing the deep sea model to the shallow sea model and in turn to the models required to determine the effects of the tsunami on building structures and the built environment.

• The problems of determining the details and modelling of the effects of sand and debris following the arrival of the first tsunami wave are seen to be formidable.

• Finally, possible economic solutions are proposed in relation to more robust and safer building practices for the affected areas and to the contribution of purpose-built or dual purpose refuge buildings for escape from such damaging sea waves.

Conclusions

EPILOGCome to Phuket, Come to Beautiful Thailand, We are Open for Business!

62