Reduce Exposure to Reduce Risk

GIT4NDM Reduce Exposure to Reduce Risk ISBN 978-616-90698-2-9 Editors: Hamid Mehmood Ranjith Premalal De Silva Nitin Kumar Tripathi 4th International Conferenece on Geo-information Technology for Natural Disaster Management 7-8 November 2012, Colombo, Sri Lanka

Reviewers Mr. Ali Sheikh, National University of Sciences and Technology, Pakistan Mr. Muhammad Mateen, National University of Sciences and Technology, Pakistan Reduce Exposure to Reduce Risk (Proceedings of Fourth International Conference on GIT4NDM) is edited by Prof. Ranjith Premlala De Silva, Uva Wellassa University, Sri Lanka, Dr. Nitin Kumar Tripathi, Asian Institute of Technology, Thailand, Dr. Hamid Mehmood, National University of Sciences and Technology, Pakistan. It is a collection of papers from GIT4NDM 2012 Publication Year: 2012 Disclaimer: Authors are responsible for the views and contents expressed in their papers. For any general queries, please contact publisher at: Geoinformatics International 165/520 Moo 5, Tiwanon Road, Muang Pathumthani, 12000, Thailand Phone: +66-2-963-9148 E-mail: [email protected] / [email protected] Reduce Exposure to Reduce Risk a

Message from Guest of Honour For Thailand, the year 2011 ended with horrifying consequences of Professor Said Irandoust historical devastations caused by floods which had no parallel in the President, last century. Flooding persisted in some areas until mid-January 2012, Asian Institute of Technology and resulted in a total of 815 deaths and 13.6 million people affected. Thailand Sixty-five of Thailand’s 77 provinces were declared flood disaster zones, and over 20,000 square kilometres (7,700 sq mi) of farmland was damaged. The World Bank has estimated 1,425 billion baht (US$ 45.7 Bn) in economic damages and losses due to flooding, as of 1 De- cember 2011.[2][3] Most of this was to the manufacturing industry, as seven of eleven major industrial estates were inundated by as much 3 meters (10 feet) during the floods. This has caused long term blow to Thai economy and business confidence. Asian Institute of Technology was also severely affected where the water remained at 2-3 m at various locations and damaged the ground- floor completely. This has caused irreversible looses in terms of dam- age to library, laboratories, and useful research data and reports. We had to relocate 250 km away from our campus to continue our aca- demic duties to our students for completion of semester without sig- nificant delay. This was the common nightmare that many of the institutions, organisations and people faced which again I think had no parallel in last one hundred years or more. Fighting natural disaster becomes futile if the magnitude is gigantic. How these loss of lives, economic losses, and disrupting the lives of millions of people can be averted or minimised? Can Remote Sensing, GIS, GPS, Information Technology solve this issue? When the disaster strikes then it is impossible for any technology to protect people and economy 100 percent. The solution these technologies can provide is how to reduce the risk from natural disasters and make us better prepared. The theme of this conference “Reduce Exposure to Reduce Risk” is rightly selected. I am in total agreement that location of eco- nomic infrastructure should be planned to ensure minimum exposure to natural disasters. Geoinformation technology, I am sure can play a very important role in this context. I wish you will focus on this vital angle and propose to decision makers a convincing solution. With my best wishes for a useful outcome of your congregation and interaction. Reduce Exposure to Reduce Risk b

Preface Human society should be endowed with the prospects of applica- Prof. Ranjith Premalal De Silva tion potentials of new technologies so that future generations would Organising Chairman, undoubtedly find a better environment to sustain life. In order to de- GIT4NDM 2012 rive substantial benefits from the inventions and innovations for the Uva Wellassa University Badulla, developing nations, it is of utmost importance to embark on develop- Sri Lanka ment and introduction of new applications to meet the challenging demands of our societies. Hardships faced by the human-kind due to natural disasters are increasing day by day partly as a result of the activities of humans. Flood, landslides, droughts, volcanic erup- tion, tsunamis, wild fires bring life and property damages all over the world. Though the present world is well equipped to successfully face many man-made disasters, it is still a considerable challenge to overcome the consequences of the natural disasters due to their unex- pected nature of occurrence. Long term strategies and huge invest- ments are needed in some occasions to successfully manage these natural disasters minimizing the damages to the man-kind. Within this background, it is a high priority to identify relevant technologies for the successful management of natural disasters. Geo-Information technology can play a vital role in managing natural disasters with its mapping and analytical potentials. It is possible to identify suitable management interventions under present conditions and changing scenarios in the landscape can be modeled to identify viable solutions for the future. Remote sensing provides possibilities to acquire data in inaccessible areas during disaster events. Within this background, the International Conference on Geo-Information Technology for Natural Disaster Management can provide a platform for the researchers from all over the world to present their experienc- es in managing natural disasters using novel technologies involving Geo-Informatics. While wishing all the success for the conference, I am hopeful that this conference will help to raise awareness of Geo- Informatics and allied technologies in managing natural disasters successfully. Reduce Exposure to Reduce Risk c

Preface Climate has changed across the globe and keeps on fluctuating and Dr. Nitin Kumar Tripathi giving shocks in terms of frequent disasters. Regions which had only Organising Secretary 170mm annual rainfall are having the same rainfall in one day. This has GIT4NDM 2012 caused flooding in cities like Jaipur which is surrounded by desserts. Asian Institute of Technology Glacier melts and snow avalanches are common phenomenon. Frequen- Thailand cies of earthquakes are on the rise. Solar activties has enhanced causing frequent solar flares. All these are impacting all of us everyday in some form or other. There is a need to monitor our geosphere, biosphere, tro- posphere and ionosphere regualrly and provide enough information to adapt to changes as quickly as possible and save our people and econo- my to guarantee a sustained life on this planet Earth. This objective can be achieved by effective use of geoinformation tech- nologies for disaster risk reduction. Policies need to be revisted and re- structured. In Asia, there is also alarming growing influx of people to cities from villages. City infrastructures are crumbling under growing pressure of population and life has become chaotic. There is a need to plan and de- velop more infrastructure and enhance facilities to cater to the growing urban population. There is a need to widen roads, create expressways, metro-rails, water management, sewerline management, telephone / mobile network managaement, urban disaster management, education, business, healthcare infrastructure, saftey and security. All the planning using multi-criteria can be done by assimilating the potentials of geoino- formatics into the regular operations of all Government Departments for effective spatial decision support. This time a new parallel meeting / conference is initiated relating to the use of GIT for Infrastructure Management. We are trying to bring the decision makers and the technology providers closer to make them un- derstand the need and availability of solutions. I am sure this will pave the way to achieve our objective to take the technology to the people to help create a sustainable and livable Asia. Reduce Exposure to Reduce Risk d

Content b c Foreword d Message from Guest of Honour f Organizing Chairman Organizing Secretary 1 23 Keynote Sessions 61 87 Technical Sessions 121 137 Flood 155 GIT4IM - I 185 Climate and Disaster, Drought Risk Reduction Drought Risk Reduction Policy 194 Earthquake and Tsunami 215 Landslide 249 GIT4IM - II Coastal Hazards, Forest Fire and Haze 291 Health Hazard IT & NDM Poster Author Index Reduce Exposure to Reduce Risk e

Use of GNSS for geophysical applications: from secular to second The advent of GNSS (Global Navigation Satellite Systems) created an Rui Manuel da Silva Fer- entire revolution on the methodologies for geo-referencing. A multi- nandes tude of new technical and scientific applications are nowadays depen- Head of SEGAL dent of this space-based system requiring accuracies from meters (e.g., Assistant Professor navigation) to millimeters (e.g., deformation monitoring) level. The University of Beira Portugal focus of this talk will be geodetic applications of GPS for monitoring geophysical signals. We start by discussing the accurate estimation and interpretation of secular motions due to tectonic and subsidence/uplift processes. A ma- jor issue is to separate on the estimated time-series the secular motions due to plate tectonics from other signals like atmosphere effects (iono- sphere and troposphere) as well as the effects due to an environment in the vicinity of the receiver (multipath, soil moisture, snow coverage etc.). Various other phenomena, such as loading effects, local crustal movements, postglacial rebounds etc., additionally affect the time-se- ries of a ground monument position and should be taken into account. Additionally, we will also focus on issues related with the application of GNSS to evaluate sudden displacements due to seismic events and their possible applications to Early Warning Systems. Geodetic ob- servations can play a crucial role in TEWS (Tsunami Early Warning Systems). Particularly, if GNSS (Global Navigation Satellite Systems) data are available just after the occurrence of an earthquake, they can be used to model the earthquakes and thus initialize parameters for tsunami modeling. Same examples are provided concerning this issue Reduce Exposure to Reduce Risk f

Making post-disaster damage, loss and needs assessment more efficient: What difference space information products can make? Disaster impacts in the developing countries continue to be upset the Dr. Sanjay K Srivastava pace of their economic and social development. In the recent times, Regional Advisor, Disaster the Damage and loss assessment (DaLA) tools, developed by United Risk Reduction Nations Economic Commission for Latin America and Caribbean (UN UN Economic and Social ECLAC), have been used especially in the context of post disaster im- Commission for Asia and the pacts in the developing countries for resource mobilization in support Pacific of resilient recovery and reconstruction. The DaLA based estimates Bangkok, Thailand have now been used for Post Disaster Needs Assessment (PDNA) to identify short, medium and long term investments in infrastructure, so- cial, productive and cross-cutting sectors. Space information products have helped in making DaLA/PDNA more effective. Harnessing regional cooperation for disaster risk management in Sri Lanka Regional cooperation frameworks enable sharing space information products and enhance the capacities for multi-hazard early warning systems. The United Nations Platform for Space-based Information for Disaster Management and Emergency Response (UN-SPIDER), ES- CAP’s Regional Space Applications Programme for Sustainable De- velopment (RESAP), the International Charter Space and Major Di- saster (the Charter), Sentinel Asia led by the APRSAF (Asia-Pacific Regional Space Agency Forum) and United Nations Institute for Train- ing and Research (UNITAR)’s Programme of Operational Satellite Ap- plications Programme (UNOSAT) have been quite effective in order to provide the access to space information products in the event of major disasters. On the other hand, WMO/ESCAP Panel on Tropical Cyclone, Regional Integrated Multi-hazard Early Warning System (RIMES) and the Humanitarian Early Warning Service (HEWSweb) are in place to enhance the early warning capacities. The paper presents the strategy and way forward for harnessing the existing cooperation frameworks to support disaster risk management activities in Sri Lanka. The paper briefly outlines how the national and regional projects supported under the ESCAP Trust Fund for Tsunami, Disasters and Climate Preparedness helped in enhancing the capacity for disaster risk management in Sri Lanka. Reduce Exposure to Reduce Risk g

Splitting of a Typhoon Passing by Korean Peninsula and an Aftershock of Stormy Wind near The East Sea of Korea Splitting process of a typhoon Songda from September 5~8, 2004 which Dr. Hyo Choi, Professor produced strong wind storm in the coastal and open seas and a severe Dept. of Atmospheric Environ- flood disaster with heavy precipitation was investigated, using a 3D- mental Sciences non-hydrostatical numerical Weather Research & Forecasting WRF-3.3 College of Natural Sciences model. The numerical simulations on wind speeds, relative humidity and Gangneung-Wonju National precipitation amount in three different domains with a horizontal grid of University 27km, 9km and 3km were carried out vertically and horizontally. The Gangneung, Gangwondo typhoon Songda of 950hPa and maximum wind speed of 40m/s for 10 Korea (South) minutes average was originally generated in the western Pacific Ocean and then it had been gradually developed following a typhoon track. h The typhoon track maintained toward northwest in the South China Sea and then it had its turning toward the East Sea (Japan Sea) passing near Korean strait between Busan, Korea and Fukuoka, Japan on September 7, 2004. Before the landfall of the typhoon, the typhoon began to split into two parts of air masses due to the shallow sea depth of the south- ern sea of Korea and surrounding lands - the eastern China, Kyushu Island of Japan and the southern peninsula of Korea. As time went on, the main body of the typhoon passing by the Korea Strait split again into two parts in the vicinity of Jeju Island in the right of Kyusu Island and then further split into four parts as the first one (I) in the Yellow Sea in the west of Korean peninsula, the second one (II) the southern sea of the Korea, the third one (III) in the East Sea of Korea and the fourth one (IV) in the eastern coast sea of Japan near Kyusu Island. Then, after the typhoon passed by the central part of Korean peninsula, the parts I and II were vanished, but part III was reinforced in the East Sea of Korea due to the terrain-induced channel effects by surround- ing lands-the eastern mountainous coast of Korea, the eastern Russia near Vladivostok and the western mountainous coast of Honshu, Japan, showing very stormy wind of 14~22m/s in the East Sea, and Part IV also maintained stormy wind. Simultaneously, geopotential height tendency for 24 hours at 500hPa level was also evaluated for chasing the typhoon track with additional GOES-IR satellite images, radar echo images and NOAA SST images. As negative minimum geopotential tendency area at 500hPa level causes atmospheric layer between the 500hPa level to the ground surface to be shrunken, it directly corresponds to the center of the typhoon. However, maximum positive tendency area causes op- positely the atmospheric layer to be expanded and it followed behind the typhoon center along its track. Thus, at 2100LST September 7, as nega- tive geopotential tendency area covered widely the East Sea of Korea, the shrunken atmospheric layer might cause further the increase of chan- nel flow, resulting in aftershock of stormy winds, continuously. Sea fog and heavy precipitation areas also corresponded to strong surface wind band. “This work was supported by the Korea Science and Engineering Foundation (KOSEF) grant funded by the Korea government (MEST) (No. 2009-0094). Reduce Exposure to Reduce Risk

CITIZEN’S ROLE: REDUCING THE SIZE OF CARBON FOOT PRINT If not you, who? If not now, when? 350 ppm Concentration of Carbon Dioxide is the safe upper limit for CO2 Dr. KVSG Murali in atmosphere. Today we are at 387ppm i.e. 37ppm higher than the safe Krishna or sustainable limit. Climate scientists now warn that if due action is not Professor of Civil taken soon, with catastrophic climate change humans will be on the brink Engineering of extinction. University College of Engineering Don’t be in an illusion that it is not us but our next generation which is in JNTUK, KAKINADA danger. The young are about to see the worst scenarios even before reach- Andhra Pradesh, ing their adulthood. The reason for major floods in Mumbai, Vijayawada, Inida. Kurnool is this climate change which is caused by global warming. Ac- cording to IPCC, all Himalayan rivers (Ganges, Indus, Brahmaputra...) will cause excess flooding as happening now, become seasonal like South In- dian rivers in its next phase and then become dry by 2035. Can you imagine the amount of social instability that we are going to see far before 2035? Do you feel the historic responsibility on today’s youth??? Reduce Exposure to Reduce Risk i

POST COLLISION TECTONICS AND THE PHENOMENON OF LANDSLIDES IN PEN- INSULAR INDIA The recent Geo- information technology based studies show that the Indian Dr. SM. RAMASAMY plate is tectonically active and resultantly the earth surface processes too Vice –Chancellor, are vibrant and controlling the natural disasters. The paper deals with the Gandhigram Rural post collision tectonics and it’s contribution to landslides. Due to the still Institute – Deemed prevalent northerly directed compressive force which has originally drift- University, ed the Indian plate towards northerly and it’s obstruction by the Himalayas Gandhigram from the north, the Indian plate is whirling with East-West trending alter- Tamil Nadu, India nate arches and deeps and fracturing with North-South extension, North- east-Southwest sinistral, Northwest-Southeast dextral and the East-West release fracture swarms. Such post collision tectonics significantly control landslides in many mountains in South India. For example, the Tirumala- Tiruppathi hills exposing the Precambrian Cuddapah quartzites is resting over one of the arches aligned along Mangalore –Chennai. As a result of such arching ,the Tirumala hills is also getting uplifted as revealed by deep fracture valleys ,shearing of the quartzites along the escarpments ,protrud- ing dykes along the obsequent slopes of the granitic basement and active slope processes like debris flow, talus accumulation, colluvial fills , alluvial fan etc., all leading to and also symbolizing landslides. Similarly in the Nilgiris of south India, the landslides dominantly fell along NNW-SSE and ENE-WSW fractures and the landslides of the period from 1900 to 2000 AD seem to gradually shift from NNW-SSE towards NW-SE fractures indicating the anticlockwise rotation of the Nilgiris due to post collision tectonics. The landslides, palaeo scars and the slump scars of Shevroy- Kalrayan hills also seem to fall along ENE-WSW /NE-SW post collision sinistral faults. Thus the paper discusses the same in detail. Reduce Exposure to Reduce Risk j

Technical Session-1 [Hall A]: Flood Rapid Assessment of Flood Affected Population using Geo-Spatial Data Modeling 2 Techniques 8 Raza, Syed Fawad, Muhammad Sheraz Ahsan 13 19 Optimization of Drainage Network to Minimize Urban Floods using Remote Sensing and GIS Techniques 19 Ranjith Premasiri, Nadeeka Paranamange, Dimuthu Niroshani, Srimal Smansiri 20 20 GIS Enabled Integrated Flood Modelling and Warning Management System 21 Oshan Perera, Lesley Karunathilaka , Rizkhan Mohamed , Thasneef Ahamed, 22 Nishani Ranpatabendi , Nimalika Fernando , S. Sivanandaraja An algorithm for Rapid Flood Inundation Mapping from Optical Data using Reflectance Differencing Technique Giriraj Amarnath1, Mohamed Ameer, Pramod Aggarwal and Vladimir Smakhtin Hydrodynamics And Sedimentological Status of the Weligama Bay, Sri Lanka Jinadasa, S.U.P and Wijayadeva, D. A Flood Assessment using with Aerial Photographs WMMNK Bandara Geospatial Database for Vulnerability Assessment: A Case Study from Kosi Floods 2008 Sanchit Suman, Nitin K Tripathi Use of Near Real Time Synthetic Aperture Radar for Flood Inundation Mapping: Case Studies from June 2008 and May 2010 Floods at Western Province Sri Lanka Srimal Priyantha Samansiri Land Use Land Cover Changes in Context to Floods: A Case Study of Delhi Sreeja.S.Nair, Nil.K. Gupta Reduce Exposure to Reduce Risk 1

Rapid Assessment of Flood Affected Population Using Geo-Spatial Data Modeling Techniques Raza, Syed Fawad, Ahsan, M. Sheraz World Food Programme Pakistan, [email protected], Institute of Geographic Information Sciences (IGIS), NUST, Pakistan, [email protected] ABSTRACT: Pakistan is a disaster prone country which encounters various natural or anthropogenic disasters every year. These result in considerable economic, financial and human losses. Floods are the major cause of these losses. The 2010 monsoon flood was the worst in the recorded history of Pakistan. Similar disasters have affected some areas of the country in the two consecutive monsoon seasons since. The capacities of the government, local institutions and communities proved to be inadequate to effectively respond to these unprecedented floods. Timely information on affected populations is central to planning a humanitarian response and optimally allocating available resources. However, this is usually only available following an on-ground assessment and often too late to contribute to the initial decision making that informs a first wave emergency response. The current flood response system is not making the best use of technological leverages available, especially the power of geo-spatial data analysis to address such problems. This study discusses the use of geo-spatial data analysis in disaster response and management. To address the problem, a spatial model was developed for the assessment of flood-affected populations in near real-time scenarios. It gives a daily update of the population living in flooded areas by making use of available spatial datasets, particularly daily satellite images from MODIS Satellite, LandScan Population Distribution Grid and administrative boundaries at the provincial level. MODIS data from Aqua and Terra sensors was interpreted on a daily basis in a specific timeframe to determine the flood extent vector, which was then superimposed on a population distribution grid to estimate the number of people living in the flooded area at any one time. The methodology was both timely and cost efficient for riverine floods, particularly in planning the humanitarian response and allocating resources effectively. The model was tested for its accuracy using an on-ground initial vulnerability assessment and the figures matched 80-90% for the areas affected by riverine floods. The proposed model can be used for any rivirine flood with a confidence level of ±10%. KEY WORDS: Spatial Analysis, MODIS, LandScan, Flood, Rapid Assessment 1. INTRODUCTION Pakistan’s geographic and geo-political disasters that result in huge human, social and location makes it a country, confronted by financial losses. Events during the last seven regular and varying natural or anthropogenic years in particular were notably devastating Reduce Exposure to Reduce Risk 2

and placed the country in an almost based on very little or sometimes no accurate continuous phase of emergency response. The information. The current system of flood earthquake of 2005, monsoon floods in 2007, response in the country is not making the best a population displacement crisis in 2009, use of technological leverages available, monsoon floods in 2010 and again in 2011 especially the power of geo-spatial data and 2012, were amongst the most significant analysis to address such problems. During any crises during this period. Floods are the most emergency, the numbers on affected frequent and major cause of human and population become vital to optimally allocate financial losses. The 2010 monsoon floods available resources, particularly in the case of were the worst in Pakistan’s recorded history. life-saving humanitarian interventions in areas What started as monsoon-related flash such as Food, Shelter, Health, and Water, flooding in the country’s north later developed Sanitation and Hygiene. This paper is trying to into a crisis of national and unprecedented explain the use of geo-spatial data analysis in proportions. As rivers extraordinarily swelled disaster response and management to bridge to more than ten or twenty times their typical these information gaps. To address the size to inundate 160,000 Km2 area (FFC, problem, a spatial model was developed for 2010), which is almost one-fifth of the the assessment of flood-affected populations country’s total landmass. Infrastructure, power in near real-time scenarios: providing a daily and telecommunications systems were update of the population living in flooded severely damaged or destroyed entirely. The areas through the use of available spatial disaster may be attributed to the effects of datasets particularly daily satellite images, climate change and associated increased geo-referenced population distribution climatic/weather variability, resulting in information and administrative boundaries. around 20.25 million (NDMA, 2010) people The methodology proved to be both time- and affected according to government estimates. cost-efficient in the case of riverine floods The capacities of the government, local particularly in planning the humanitarian institutions and communities proved to be response and allocating the resources inadequate to effectively respond to this optimally. unprecedented flood. Timely information on affected populations is essential to plan a 2. DATA AND METHODS humanitarian response and allocate available resources optimally. However, this is usually This study demonstrates the use of geo-spatial only available following an on-ground data analysis in disaster response and assessment and often too late to contribute to management. A geo-spatial model assesses the the initial decision making that informs a first flood-affected population in near real-time wave emergency response. Furthermore, the scenarios and gives almost a daily update of flow of information from the field is not the population living in flooded areas through efficient enough to report frequently during an the use of available spatial datasets evolving disaster such as those caused by particularly the hyper spectral satellite data riverine floods. Timely information can help from MODIS sensor of Aqua and/or Terra to optimally allocate resources in the right Satellites, LandScan Population distribution direction and minimize inefficiencies grid and administrative boundaries at the associated with poor management and provincial level. Weather permitting, MODIS decision making which is conventionally data from Aqua and Terra sensors were interpreted on a daily basis (between August Reduce Exposure to Reduce Risk 3

and October 2011) to obtain the geographic 2.1.2 Population Distribution extent of the flooding. This was then superimposed onto a population distribution The last national population census in grid to determine the number of people living Pakistan was conducted during 1998 and within flooded areas, with the regularity of provides information at the Moza (Revenue updates reflecting the evolution of the state or Village) level. Unfortunately, this situation. statistical data is not geo-referenced. Furthermore, socio-economic, political and 2.1 Data environmental instability experienced by Pakistan in the last seven years has resulted in The model uses mainly three datasets as mass population migration from crisis- inputs, of which two are freely available affected areas to safer locations, precluding (MODIS daily satellite images and GAUL). the application of straight projections for While Landscan population distribution was population estimates in a given geographic also free for the humanitarian community and location. Also note that only two censuses research agencies until 2009, Oak Ridge conducted in Pakistan first in 1981 and the National Laboratory went commercial in 2010 second in 1998 and the growth rate is and this data is now only available at a cost. A calculated based on population growth during brief description of the three datasets and their this period. specifications are discussed below. Using an innovative approach involving 2.1.1 Satellite Images Geographic Information System and Remote Sensing, Oak Ridge National Laboratory's MODIS sensor is installed on Aqua and Terra LandScan™ is the finest resolution (1x1 km) satellites and is providing hyper spectral global population distribution data available images in 36 spectral bands twice a day (10:30 and represents an ambient population (average a.m. for Terra and 1:30p.m. for Aqua) for over 24 hours) for quick estimations in every part of the world at no cost. The emergency situations (NRC, 2002). dimension of one image footprint on ground is 2330km by 10Km. The spectral resolution The LandScan algorithm uses spatial data and (250m for band1-2, 500m for band3-7 and imagery analysis technologies and a multi- 1000m for band8-36) is sufficient to perform variable dasymetric modelling approach to various macro level analyses in relation to disaggregate census counts within an disaster risk management and particularly administrative boundary. Since no single detection of surface water and vegetation. The population distribution model can account for first two bands (i.e. Band1 and Band2) are the differences in spatial data availability, used in this model to extract water from the quality, scale, and accuracy as well as the daily images. These are the only bands out of differences in cultural settlement practices, 36 bands in total in a MODIS image, available LandScan population distribution models are at a spatial resolution of 250m and falls in a tailored to match the data conditions and bandwidth range 620-670nm (Red) for band1 geographical nature of each individual country and 841-876nm (Near Infrared) for band2. and region. Reduce Exposure to Reduce Risk 4

Figure 1: Methodology 2.2. Methodology 2.1.3 Administrative Boundaries The figure 1 explains the process flow and steps involved in the overall model. In this study The Global Administrative Unit Layers (GAUL) are used at admin1 (province) a) MODIS aqua and terra images are level. GAUL is an initiative implemented by downloaded from the MODIS website. FAO within the EC-FAO Food Security The images are made available after the Programme funded by the European capturing time (10:30 a.m. for Terra and Commission (GAUL, 2009). The GAUL aims 1:30 p.m. for Aqua), every day. at compiling and disseminating the most reliable spatial information on administrative b) The two scenes are analysed for the best units for all countries in the world, acquisition particularly in terms of cloud contributing to the standardization of the cover and angle of inclination at the time spatial dataset representing administrative of acquisition. units. The GAUL maintains global layers with a unified coding system at country, first (e.g. c) Spatial algorithms are applied for water regions) and second (e.g. districts) detection particularly Normalized administrative levels. The GAUL is released Difference Vegetation Indices (NDVI) once a year and the target beneficiaries of the using the following standard formula: data include the UN community, universities and other authorized international and national ( )( ) institutions/agencies. ( )( ) d) Applying thresholds on band02 is also used to further refine and validate the results acquired using NDVI. e) Band seven is used to detect clouds and the removal of cloud shadow. f) The flood extent is exported to vector format to further superimpose on Population Distribution data g) The flood vector is used to extract the population distribution grid inundated through performing an overlay analysis. h) The resultant population distribution raster file is further superimposed on GAUL boundary of Pakistan at Admin level 1 (Province) to calculate Zonal Statistics at this level. Reduce Exposure to Reduce Risk 5

i) The sum of all the pixel values falling in the flooded region of a given province will give the estimated population living in a flooded area by province. j) The area inundated can also be calculated using the geometry calculation of flood vector. The process is executed on a daily basis to monitor water recession and suggest which areas are safe for the return of displaced groups living in an informal settlement or an organized camp location. 3. RESULTS In the case of riverine flooding, the Figure 2: LandScan Population Distribution methodology proved to be both time- and cost-efficient, and has greatly facilitated humanitarian response planning and resource allocation by providing an estimate of affected populations. The figure 2 show a population distribution Flood2011 - Affected Population in Error Sindh Province Estimation from Model Surveyed data 5,751,116 5,441,869 5.68% map for the Sindh Province, using LandScan Population distribution data and the figure 3 shows the standing flood waters and the maximum flood extent reached during august 2011 to march 2012 study period. The model was tested against ground assessment data (PIFA, 2011) (RRP, 2011) (MSDNA, 2011) collected during an Initial Vulnerability Assessment and demonstrated 80-90% Figure 3: Flood Extent (August 2011 to March 2012 accuracy. The proposed model can be used for any riverine flood with a confidence of ±10%. The system additionally facilitated planning processes during the recovery and return periods; particularly as water recession Reduce Exposure to Reduce Risk 6

monitoring suggested which areas were safe to also be used for inundated crops using an return for the displaced populations. Figure 4 updated land use data like MERIS MOD44B shows the water recession monitoring curve. 6. ACKNOWLEDGEMENT Thousands Sq Km 15 Technical support from Emergency Preparedness and Response Branch of WFP 10 16-Nov 29-Feb and ITHACA during this study is very much appreciated. Flood extents extracted by 5 UNOSAT for the same area were also used to validate the image interpretations. 0 10-Aug Special thanks to Ms Jedda Constantine and Dr. Amjad S. Almas for their valuable Figure 4: Flood water recession monitoring comments for this publication. 4. CONSTRAINTS AND LIMITATIONS C. 7. REFERENCES FFC. (2010). Annual Flood Report 2010. The major constraint in this methodology is Federal Flood Commission, Ministry of Water that it is using passive sensing to detect and Power, Govt. of Pakistan. standing water and compared with permanent water bodies to detect flood waters. Cloud GAUL. (2009). Global Administrative Unit cover is the key constraint, which may be Layer. Retrieved October 16, 2012, from FAO particularly acute during the monsoon season. Geo-Network: Use of active sensing (radar imaging) can be a http://www.fao.org/geonetwork/srv/en/metada solution to this but is not as cost efficient in ta.show?id=12691 comparison to what is in hand now. MSDNA. (2011). Multi-Sector Damage Needs 5. CONCLUSION Assessment in Sindh and Balochistan Provinces. United Nations Pakistan and The model is useful in situations where water NDMA. remains in place for at least 24 hours so as to NDMA. (2010). Annual Report 2010. be easily detected. Cloud cover is another National Disaster Management Authority, issue during the monsoon season which can Islamabad, Pakistan. impede the analysis. The model is not recommended for micro planning because of NRC. (2002). Down to Earth: Geographic its coarse spatial resolution (1Km for Information For Sustainable Development,. population distribution and 250m for satellite National Research Council 2002,(National data). While proper planning requires social Academies Press). indicators best collected via field surveys, the model does provide a good estimate of PIFA. (2011). Pakistan Initial Flood affected populations which is crucial to initial Assessment. Vulnerability Analysis and planning and resource allocation immediately Mapping Unit of World Food Programme, after the onset of a flood. The same model can OCHA and NDMA. RRP. (2011). Pakistan Floods: Rapid Response Plan. United Nations, Pakistan. Reduce Exposure to Reduce Risk 7

OPTIMIZATION OF DRAINAGE NETWORK TO MINIMIZE URBAN FLOODS USING REMOTE SENSING AND GIS TECHNIQUES Ranjith Premasiri1, Nadeeka Paranamange1, DimuthuNiroshani1 ,Srimal Smansiri2 University of Moratuwa, Sri Lanka ,Disaster Management Centre, Sri Lanka E mail: [email protected] ABSTRACT The frequency of the occurrence of disasters is increasing day by day. Urban flooding has become one of severe problems faced by Sri Lanka seasonally leading to various social and environmental interruptions. With the increasing of the occurrence it has become a necessity to find a proper solution to overcome the problem. This study was mainly focused to optimize the drainage network in Panadura urban council area, Sri Lanka to minimize the urban flood hazard susceptibility.using Remote Sensing (RS) and Geographical Information System (GIS) techniques. Light Detection AndRanging (LiDAR) Digital Elevation Model (DEM) was utilized to delineate requisite drainage and mini water catchments using Environmental Systems Research Institute (ESRI) Arc Hydro Model. Prior to LiDAR DEM processing an accuracy assessment was performed with respect to the ground truth elevation measured by Total Station and GPS surveys. Current available drainage system in the area was assessed in two ways as alignment and capacity for large water volumes in heavy rainfalls with respect to the delineated natural drainage system. Highly flood vulnerable locations in the current drainage system were identified by performing the comparison of natural drainage network and existing drainage paths. Finally adjustments to the current drainage network and new drainage paths were proposed. Keywords: Digital Elevation Model (DEM), GPS (Global Positioning System), Total Station, Arc Hydro Model 1.0 INTRODUCTION November 2010.The causes identified by many authorities for flash flooding are poor drainage Urban flooding has become a severe problem in maintenance (structural damage), lack of drainage Colombo, Kalutara, Gampaha, Kegalle, Matara paths, improper drainage construction along the Galle districts within the country for few years. During last road, water flow blocked by garbage due to illegal two years the problem has gone up to a more dumping, filling of wetlands and unauthorized vulnerable level due to frequent flash floods causing constructions. several social interruptions such as traffic congestions, delays of trains, closure of schools, According to the literature, a number of physical electricity interruptions, disruption of several services studies based on topographic surveying have been and temporary loss of income (DMC Report, 2010 ). conducted to propose a solution for this severe Considering these facts of urban flooding, it is problem. But most of them were failed during the required to propose a long term solution for the implementing phase although they utilized both high problem through detail studies. cost and time. Therefore as a highly applicable Panadura Urban Council in Kalutara district has been approach, this study was carried out to propose an chosen for the case study. This is a highly built up appropriate solution to minimize urban flooding in and industrial area with a population of 33735 Panadura area. The major difference comparing to approximately. Panadura city has been facing the previous studies is the decision making according to problem over five years and sever events were the GIS recorded in May of 2009 and April, May and analysis. As Remote Sensing data, processed Digital belongs to coastal plain where elevation varies from Elevation Model (DEM) created by LiDAR images was used. 0 m to 15.91m. It is within latitude of - and a longitude of - . 2.1 STUDY AREA 2.2 DATA USED The extent of the study area covers about 5.85 km2 of  LiDAR DEM (5m – resolution) as the main Panadura Urban Council in Kalutara district, Sri Remote sensing data (Source - Disaster Lanka. Topographically it is flat low elevated area Management Center, Sri Lanka) Reduce Exposure to Reduce Risk 8

 GIS Layers– Road Network, Drainage Figure 2: Input and output discharges Network digitized from Google Earth of drainages  Field data – Elevation data using Total station survey and Coordinates of flood points, culvent points and water points using GPS survey Figure 1- Study Area of Panadura UC Q2 Q1 Figure 3: Alignment of current drainage network Reduce Exposure to Reduce Risk 9

3.0 METHODOLOGY If Q1>Q2, Drainage point is considered as an incapable one and if Q1≤Q2 it is considered as a Methodology consists of four phases as accuracy capable one. assessment, hydrological processing, data analysing, verification and design. This calculation was performed for both maximum (resulted from 18 years) and average rainfall value in 3.1 Accuracy Assessment of DEM flood seasons. To assess the accuracy, variance of elevation differences was calculated for each total station point 3.4 Verification with respect to the extracted elevation values of Identified incapable drainage points using LiDAR DEM. calculations were compared with ground truth flood locations for further verification of methodology. 3.2 Hydrological Processing Arc GIS can be used to accommodate more reliable 3.5 Design approach of spatial variations in hydrological The design for a new drainage was developed based parameters than using manual procedures. Therefore on the amount of surface runoff it has to cater. LiDAR DEM was processed using ESRI Arc Hydro Surface runoff was calculated using highest rainfall Tool to delineate the stream network (natural flow (450 mm/day) resulted for the area and assuming all path). flow direction, sinks and flow accumulation rain become runoff and no loss of water because of were generated prior to the stream network interception, evaporation, transpiration or loss to delineation assuming that water can flow in from ground water. many cells but out through only one cell. New proposed drainage was divided into segments 3.3 Data Analyzing with respect to the area calculated using maximum flow velocity (0.5 m/s) in flood seasons This was performed by assessing two aspects as Alignment and Capacity of drainage network. 4.0 RESULTS AND DISCUSSION  Alignment assessment 4.1 Accuracy Assessment of DEM Alignment was assessed by overlaying the The resulted accuracy of LiDAR DEM was 0.3377 delineated natural flow path with the current m. Since accuracy is in the centimeter level source available drainage network. DEM was accepted as an accurate one.  Capacity assessment 4.2 Alignment assessment Capacity assessment was performed by Alignment of current drainage network was assessed comparing Q1 and Q2 discharge values by overlaying that with delineated natural flow path. (Maduranga. et al. 2009). The overlaid two layers are represented below (Figure 4). According to the result of the assessment, Where Q1, more deviations were encountered in the top most region of the drainage system than the lower region. Required discharge for natural flow (Q1) 4.3 Capacity assessment Q1=Flow Accumulation * Area * Daily Rainfall / Hours*60*60 The Table 1 represents the results for natural flow accumulation and the discharge which is capable by And Q2 is Capable Discharge of existing drainage current drainage network. paths Based on the delineated natural flow path a new Q2 = Cross Sectional Area*Flow Velocity design was developed. Further, settlements and road network of the area were considered during the design phase. Reduce Exposure to Reduce Risk 10

Table 1 – Calculation of flow rate II. 5.0 CONCLUSION Discharge Water Accumulation ( m3) at Culvert Max. Ave. Average rainfall- ( m3) Rainfall Rainfall flooding C5 0.6567 5.212 0.1144 2.1177 C6 0.7620 7.994 0.1755 3.2484 0.6567 8.524 0.1871 3.4637 C7 1.1675 11.495 0.2523 4.6710 C8 1.5240 12.205 0.2680 4.9608 C9 0.7620 2.294 0.0504 0.9323 C1D 0.4572 1.037 0.0228 0.4212 DE1 0.0000 2.757 0.0605 1.1203 WB1 2.2860 1.549 0.0340 0.6294 WH I. 4.4 DRAINAGE DESIGN The major problem in the drainage network available in Panadura area is the incapability to handle water volumes generated from natural accumulation not only in heavier rainfall but even in average rainfall in flood season. Therefore increasing of drainage volumes is a vital factor. As an alternative for increasing the volume, water blockage in the drainage should also be removed by cleaning the plant cover and wastes. The other identified problem is the improper alignment with the natural flow path in most parts of the network. As the final solution, to overcome all these problems a new drainage which will satisfy all aspects was designed. ACKNOWLEDGEMENTS The authors wish to thanks to University of Moratuwa and Disaster Management Centre, Sri Lanka for providing data and facilities for carrying out this research. Also we appreciate the Chairman of State Engineering Design and Development Corporation, the priest of AbhayaMudalindaramaViharaya and villages of Panadura for valuable support given during the field works. Figure 5: Proposed Drainage Network Reduce Exposure to Reduce Risk 11

REFERENCES Flood Situation Update No.2, (2010 – Sri Lanka, 18 May 2010 at 1830 hrs) Disaster Management Center, GLIDE: FL-2010-000092- LKA Report on Joint Rapid Assessment Disaster Management Centre (DMC) of the Ministry of Disaster Management, World Food Programme,Office for the Coordination of Humanitarian Affairs (OCHA), Date: 12 November 2010 Maduranga H.R and Balasooriya BMRK, 2009, Design Of Surface And Subsurface Drainage System For Johnston Estate Resettlement Site, Nuwaraeliya Using GIS Techniques, National Building Research Organization Reduce Exposure to Reduce Risk 12

GIS ENABLED INTEGRATED FLOOD MODELLING AND WARNING MANAGEMENT SYSTEM

In the present process, the modeling of flood is Data of 15 GN divisions are used for warning management component. not done in a manner which assists in identifying The rest of the paper presents the proposed affected areas at village level. It does not make use methodology, prototype implementation with technologies used and sample results. This is the of any scientific flood modeling methodologies. first version of the proposed approach and possible future directions are also discussed. Further, the warnings send may fail to reach the 2. METHODOLOGY individuals who may be affected by floods as only The proposed framework has four components; the flood modeler (FM), the affected mass communication is used. The communication area Identifier (AAI), the Warning Management Component (WMC) and the Integrator web portal between different parties involved is not efficient (IWP). Following the current process, these components are assigned to be handled by different which leads to delays in decision making. authorities. However a proper and well defined communication mechanism is defined to handle the GIS and communication technologies can be whole process in a collaborative manner. Figure 1 shows the different components of the framework adopted to overcome these weaknesses in the flood with the relevant authorities and the communication flow. mitigation process. Bai et al. (2008) provides DeptI has the responsibility of executing FM details of such framework for Malysia. Jeyaseelan and the AAI components. Row data required for this component are based on hydrology domain. (2003 ) shows evidence of such attempts in India. DeptI has the required knowledge and other resources to execute this In Sri Lanka also there are several research work Figure 1: Proposed Flood Management System done related to flood modeling. There are attempts component once implemented. Therefore DeptI is given the responsibility of these processes. GIS to model flood in Kalu Ganga river (Samarasinghe,2010) and Kelani river (Gunasekara,2008). However none of these Sri lankan studies attempts to integrate the modeling and warning management processes. A flood management framework which deals with (1) flood modeling and (2) flood warning in more effective and efficient manner is proposed in this study to overcome weaknesses in the present process. It uses Geographic Information System (GIS) and Information and Communication Technologies (ICT) together with the conventional knowledge. In the proposed approach, the flood modeling is done based on gauge discharge data using a GIS tool. Based on the modeling, affected GN areas are identified. System offers facilities to register general public and relevant authorities before a disaster situation. Then, once a flood situation is identified, suitable warnings can be issued to them using modern communication methods such as SMS and E-mails. Mobile phone has become a common commodity of the general public and SMS is a feature available in low end phones even. Therefore this mode of communication can expect to reach the target communities effectively than the traditional means. The system will assists in maintaining timely and reliable information flow between Grama Niladhari (GN), District Secretariat( DS), DeptI and DMC also. A prototype of the model is implemented for the lower reach of Kelani river as the sample region. Guage discharge of Hanwella gauge station, which is at the upper boundary of the selected area , from Hanwella to Colombo North is used for the flood modeling. This covers the area of 3242 Km 2. Reduce Exposure to Reduce Risk 14

based flood modeling will be used by the FM The FM and AAC components are component. Flood modeling is a complex process implemented using GIS technologies. The FM as flood levels can be vary based on many factors. component is implemented using hydrology Rate of Flow (Discharge), Rainfall, Evaporation, technologies namely HEC-RAS 4.1.0, HEC- Wind Velocity, Temperature, Solar Radiation and GeoRAS 4.3 and GIS tool ArcGIS 10.0. ArcGIS Sedimentation are hydrological measurements offers a strong support regarding GIS hydrology which can be used in this regard (Dias, 2006). modeling. However it is a proprietary software tool However other factors such as soil type of the area, and ownership cost is considerably high. Free and sand excavation along the river bank and the open source GIS tools are not yet powerful enough constructions blocking the water flows etc affect in this regard when compared with ArcGIS. the actual flood levels. In the proposed flood model Therefore, even with the cost barrier, this only the gauge discharge data are used following proprietary tool is used in the study. the current process used. AAI component is developed using open The output of the FM component is the flood source tool MapWinGIS 4.7 with support of its inundation data of the selected area. This API and geo-processing libraries. AAI is information will be passed to the AAI to identify implemented on the top of .NET Framework 3.5 DS divisions and GN divisions which will be and with the help of Microsoft Visual Studio IDE affected by the flood. AAI uses GIS based over with C# language. The IWP and the WMC is layering methods in this regard. developed using PHP language. The warning dispatch is implemented using SMSGlobal SMS The extracted information will be passed to gateway, which offers a 30 free SMSs for 1 mail the WMC to issue relevant warnings to individuals account. and organizations involved. It will reside at the DMC. As DMC is the authority which works 3.2. Data used for test implementation closely with the administrators and community itself in a flood situation, WMC has to be handled 3.2.1. Topographic and GIS Data by them. In addition to the warning handling , this  1:10000 contour maps component offers support for pre-registration of  1:10000 land-use maps in the selected individuals and organizations to the system via a web interface. The GN of the area can register as a catchment user and then he/she can register individuals in that  1:10000 administrative boundary maps area. The organizations such as DMC or Public  Spot heights digitized in 1:10000 scale places such as schools, hospitals etc in the area also can be registered. Based on the affected areas Figure 2 : Geographical area considered for the identified, the target affecting people and study organizations can be filtered. Simple Message Services (SMS) and Emails are used to issue 3.2.2. Hydrological Data warnings directly to them within few minutes.  The model uses the Hanwella station , the IWP component, which also resides at DMC upper stream gauge station‟s discharge data contains functionalities to handle the  It uses cross section data of 5 main river gauge communication flow between different components. It pulls the information from the stations starting from Hanwella gauge station DeptI end and assists in distribution of required towards down side information to the general public as well. 3. TECHNOLOGY AND SAMPLE DATA SELECTION The prototype of the framework is implemented for lower Kelani river area. The area considered and the locations of the gauge stations are shown in the Figure 2. 3.1. Technologies Used Reduce Exposure to Reduce Risk 15

 Manning values of Kelani river lower reach. corresponding data will be passed to IWP to use for the management activities collaborating with 3.2.3. Individuals/ organizations data DMC. The process within the AAI is illustrated in  Individual‟s telephone numbers and personal Figure 4. information for testing SMS ( 20 sample) Study 1:10000  Data on DS, GN and DMC area TIN Topographic 4. IMPLEMENTATION Pre- Implementation details of main Processing components are presented in the following section. Cross Gauge 4.1. Flood modeling Component (FM) section Data Discharge Figure 3 shows the work flow of the FM Data component. This involves three main sub Figure 3 : ImplementaPtriocneossfiFlood Modeling components to handle pre-processing, processing and post-processing. Compongent  Pre-processing Post- Accuracy of the model and the quality of the management decisions depends on the quality of Inundation the data used for the model. Therefore pre- processing of data is an important step in the Figure 3 : Implementation of Flood Modeling modeling process. The contour line and elevation Component points data are used for generating a TIN model. In addition to that feature classes such as river 4.2.3. Warning Management Component center line, flow paths, bank lines, XS cut lines are (WMC) also used at this stage. HEC-GeoRAS extension for Hec-RAS tool was used for the task. Figure 5 shows the work flow of the WMC component. As the end result, a message will be  Processing generated by respected administrative authorities FM component also covers further processing of and will be delivered to the general public in flood river bathymetry and surface roughness. The affecting area. processing is mainly carried out based on the cross section data. Finally gauge level discharges are 4.2.4. Integrator Web Portal (IWP) entered to the implemented model and initial execution is performed. In the processing stage, IWP is implemented as a web based HEC-RAS tool is used. component run  Post-processing 4.2.5. Integrator Web Portal (IWP) In this stage, model requires previously created TIN as a main input parameter. Finally the IWP is implemented as a web based analysis is performed and flood inundation is component run generated as result of the analysis. on top of an Apache server. It provides support 4.2. Affected Area Identifier (AAI) to interaction between the people who live in the flood prone area and the officials who responsible The flood inundation data itself is of little use for taking necessary flood mitigation and recovery for flood management activities. These data needs actions. to go through GIS operations to identify flood affecting DS divisions and GN divisions. Therefore 5. SAMPLE RESULTS a standalone customized GIS interface which can be used to view the flood inundation data and then process them with the other required GIS layers is developed. It offers geo processing and spatial analysis functionalities such as clipping, buffering etc. Once the flood prone area is identified, Reduce Exposure to Reduce Risk 16

The model is tested with sample gauge discharge AAI Grama values available with DeptI for 2005 and 2008 Input Nladari floods occurred in the selected area. Integrator Web Portal (IWP) FM Authority Level: Grama Niladare modeler General Public Registration Affected Area Identifier (AAI) Authority Level: Disaster Management Center Authority Level: Irrigation Department Reporting Basic GIS Component Operations Clipp Bufferin Notificat Warn ion E-mails ing Identified Inundation Area Disaster General (Grama Niladari Division Management Public Wise) Center Pass bulk data to User Figure 4 : Work flow within Affected Area Figure 5 : Implementation of Warning Identifier Component Management Component The testing process is done in three stages; The model is tested with known gauge discharge 6. DISCUSSION values. Then the resulting affected area map is verified with a field visit. The filed visit data is The proposed approach assists in again used to calibrate the model. overcoming the present flood mitigation process in three ways. Mainly it improves the communication The figure 6 and figure 7 show sample flood flow between different parties involved in the flood affected area resulted from the system for the flow management process. Secondly as the flood parameter of 1265 m3/sec ( 2008 flood data) and modeling is done with GIS enabled tools, it is 122m3/sec in same station ( 2005 flood data) for possible to identify exact GN divisions which Hanwella gauge station respectively. would be affected by the flood. Thirdly, it assists in identifying flood prone people and public places up The calibration of the model is done by to GN level and issue warnings directly targeting at performing a comprehensive field survey along the them via a SMS system. All these would help to Kelani River in the selected study area from coordinate post-disaster management process more Kaduwela to Kelaniya. It is used to obtain effectively. independent reference data to test the accuracy of resulting GIS model. The WMC is tested with two The main limitation of the system is the telecommunication service providers, namely simple nature of the flood modeling process Dialog and Mobile. adopted. System uses a modeling technique based on a single gauge level discharge. Flood levels can vary based on many more parameters. When developing a proper flood model, it is required to build ground profile cross sections across channel flow with 100m interval which are spanning over the entire flood extent. In the selected river catchment, only 84 cross sections are Reduce Exposure to Reduce Risk 17

created with equal interval between cross sections.  Adoption of multiple input parameters to The interval size is greater than 100m in this case. flood model such as rainfall data, temperature data, wind velocity, solar In a more accurate flood model, bridges, radiation and sedimentation. culverts, ineffective flow areas and obstructions are also required to be incorporated as and when they ACKNOWLEDGEMENTS are present in the selected catchment. Due to the Authors wish to offer their sincere insufficient data limitations only the river flow paths and tributaries are considered in this study. gratitude to Dr. Hemalie Nandalal, Senior lecturer While the area affected will be calculated by the at Department of Civil Engineering, University of system, the timing is based on the staff of the Peradeniya, Sri Lanka and Mr. S.M.J.S. experience of the DeptI. Samarasinghe, Superintendent of Survey, GIS Branch, Dept. of Survey, Sri Lanka for the Figure 6 : Flood affected area identification ( inevitable guidance and technical support offered sample data : 1265 m3/sec gauge discharge) in flood modeling stage. The assistance provided by the staff of Department of Irrigation and the Figure 7: Flood affected area identification ( Disaster Management Center are also sample data : 122 m3/sec gauge discharge) acknowledged. This study is done an undergraduate 7. CONCLUSION final year project of Sri Lanka Institute of Information Technology and the support provided The purpose of this study is to explore the by the institute is also acknowledged. applicability of GIS technologies and modern communication methodologies in the process of REFERENCES flood modeling and warning management. Kalani river basin is used as the sample area. The Bai,R, Akbari, A, Datu and Azizan.,2008, prototype system developed as the proof of concept “GIS-Based Flood Modelling”, 2008, shows acceptable results. The proposed approach University of Malaya, Kuala Lumpur can be further improved considering following http://www.apru.org/awi/workshops/climatec features. hange2008/Ppt/AWI03rdCMASUM- Azizan.pdf  Integrate automatic sensing methods to Dias, P. P. G., 2006, “Hydrometric Network read gauge discharge values. & Flood Mitigation”, Hydrology Division, Irrigation Department, Sri Lanka.  Apply remote sensing technology and http://www.itu.int/ITU- satellite images in calibrating the flood D/asp/Events/EmergencyTelecomWorkshops/ model more accurately. SriLanka_Workshop/Presentation/7%29%20I rrigation%20Department.pdf Gunasekara, I.P.A., 2008, “Flood Hazard Mapping in Lower Reach of Kelani River”, In Proceedings of ACCIMT Conference, 27th August 2008. Jeyaseelan, A. T., 2003, “Droughts & Floods Assessment and Monitoring using Remote Sensing and GIS”, Department of Space, Govt. of India, Hyderabad. http://www.wamis.org/agm/pubs/agm8/Paper- 14.pdf Samarasinghe, S. M. J. S, Nandalal H. K, Weliwitiya, D. P, Fowzed , J. S. M , Hazarikad , M. K. and Samarakoon,L., 2010, “Application of Remote Sensing and GIS for Flood Risk Analysis: A Case Study at Kalu- Ganga River, Sri Lanka”, in International Archives of the Photogrammetric, Remote Sensing and Spatial Informatio. Reduce Exposure to Reduce Risk 18

AN ALGORITHM FOR RAPID FLOOD INUNDATION MAPPING FROM OPTICAL DATA USING REFLECTANCE DIFFERENCING TECHNIQUE Giriraj Amarnath1*, Mohamed Ameer1, Pramod Aggarwal2 and Vladimir Smakhtin1 1International Water Management Institute (IWMI) HQ, Colombo, Sri Lanka 2International Water Management Institute (IWMI), New Delhi, India Email: [email protected]; Phone: +94-11-2880000; Fax: +94-11-2786854 ABSTRACT: There is a need for developing innovative processing chains for generating remote sensing-based products supporting near real-time flood management. This paper presents an algorithm for flood inundation mapping in the context of emergencyresponse. The analysis of spatial extent and temporal pattern of flood-inundated areas is of primeimportance for mitigation of floods. With the development of remote sensing techniques, floodmapping for large areas can be done easily. The algorithm involves the use of shortwave infrared (SWIR), near-infrared (NIR) and green spectral bands to develop a suitable band rationingtechnique for detecting surface water changes. This technique is referred to as NormalizedDifference Surface Water Index (NDSWI). The proposed approach is applied to the July 2010 flood in Pakistan. The NDSWI-based approach produces the best results for mapping of flood inundated areas when verified with original satellite data. Analysis of results reveals that NDSWI has the potential to detect floodwater turbidity, which was verified using principal component analysis.The data used by the proposed algorithm is available for a large portion of the world, which makes the approach suitable for automated near-real time flood mapping. KEYWORDS: Flood, Rapid mapping, Remote sensing, Band rationing HYDRODYNAMICS AND SEDIMENTOLOGICAL STATUS OF THE WELIGAMA BAY, SRI LANKA Jinadasa, S.U.P and Wijayadeva, D. A Oceanography Division, National Aquatic Resources Research and Development Agency (NARA) Crow Island, Colombo 15, Sri Lanka, Email: [email protected], Fax: +94-11-252192 ABSTRACT Weligama bay is a semi enclosed bay with the dimensions of ~2.9 and ~2.5 km’s length and width respectively, highly abundant with coral reefs and rock outcrops. Polwathumodera river is the major discharge carrying heavy sediment load during the southwest monsoon. Severe erosion on west bank of the bay and narrower the access channel by accumulation of sediments is the current burning issues for safe navigation. Sedimentological status of Weligama bay was studied by collecting sediments on vegetation, high water and low water levels. Comparison of sediment analysis data indicated that they are formed in diverse sedimentological environments. In association of the sediments from high and low water lines suggests that the sediments are derived in outsource environment and deposited on the navigation channel. Further the, aggregation of coarse grained carbonated materials on Kapparatota Point suggests that are originated from offshore sources. The results of hydrodynamic studies of the bay show westward strong current patterns with the speed of ~0.12 m/s. Also, the results suggest that the intensity of the currents increase due to impact of coral reefs and small islands in inner bay. However, mechanism of heavy sediment formation and accretion is not explained yet scientifically. Though this study was mainly focussed on hydrodynamics and sedimentological status to explain high sediment concentration on navigation channel, further studies on bottom current behaviour are required to conclude sedimentological status of bay. KEYWORDS: sediment, current, bay, hydrodynamics Reduce Exposure to Reduce Risk 19

FLOOD ASSESSMENT USING WITH AERIAL PHOTOGRAPHS Willora Mudiyanselage Mevan Nishantha Kumar Bandara Survey Department of Sri Lanka, Photogrammetric Unit, [email protected] ABSTRACT Flood attacks in Sri Lanka are mainly due to excessive rainfall received during monsoons and receive as a result of development of low-pressure in the Bay of Bengal. Floods are directly related to rainfall and therefore a proper understanding about the distribution of rainfall becomes important. In this case, Gampaha district was considered and collected the aerial photographs with some Ground Control Points to create the Aerial Triangulation. After that, the same Data can be used for the Digital Terrain Model creation. The scale of the Aerial Photography is 1:8000. Therefore, that data is more sensitive to high accurate DTM generation and the flood assessment of above mentioned area. This project has few steps to be covered. Block Preparation, Bundle Block Adjustment, Digital Terrain Model Creation, Digitize Water Features and create buffer zones, Assess the flood with the rain fall data, Testing with rain fall Data, As a Photogrammetric Technician, the concept of flood modeling is easier to distinguish the various kinds of obstacles in terrain such as forest covers, mountainous areas, urban areas etc. In many times, we experienced and did algorithms for generating DTM for given coordinates and water level displaying on that given surface. KEY WORDS: Aerial Photographs, Ground Control Points, Aerial Triangulation, Digital Terrain Model (DTM), Block Preparation, Bundle Block Adjustment, Digitize, Water Features, Buffer zones, Photogrammetric Technician. GEOSPATIAL DATABASE FOR VULNERABILITY ASSESSMENT: A CASE STUDY FROM KOSI FLOODS 2008 Sanchit Suman 1and Nitin K Tripathy School of Engineering and Technology, Asian Institute of Technology, Pathumthani, Thailand ABSTRACT On 18 August 2008 an embankment of river Kosi in South Nepal breached, leading to change of the river course completely and finally resulted in one of the worst flood disasters in Bihar, India. The flood inundated large areas of Nepal and the state of Bihar in India, affected nearly 4 million people (Nepal and India put together) and caused immeasurable sufferings to poor people in one of the most backward areas of the region. Thanks to the efforts made by the Government of Bihar, National Remote Sensing Agency, Government of India and International Charter for Space and Major Disasters, satellite data capturing the diverse and dynamic flood waves were available on their respective websites. However, what was not available was the profiles of vulnerability and risk of the people who were to be targeted for humanitarian assistance. Poverty and vulnerability to the natural disasters have strong spatial correlations in the Kosi basin. The present study deals with vulnerability mapping taking into account the experiences of 2008 Kosi floods. The spatial and non-spatial data used used in the study were collected from online resources available in the public domain. The vulnerability atlas of India, developed by the Building Materials & Technology Promotion Council, Ministry of Urban Development and Poverty Alleviation, Government of India, was used in conjunction of near real time satellite data capturing the different flood waves, to assess the vulnerability profiles of the affected population in the riverbasin. Though at coarser scale, vulnerability atlas served as a baseline spatial data for assessing flood 1 Sanchit Suman, Student of Department of Computer Engineering, 2006-2010 academic session at SV National Institute of Technology, Surat, Gujarat, India, contributed this paper under the mentorship of the second author. Reduce Exposure to Reduce Risk 20

impacts. The paper intends to present an approach how online satellite images of flood affected areas can be combined with socio-economic data to assess and target the people at risk. Mindful of the fact that Kosi floods hit the poor of the poorest region, assessing the vulnerability and risk was of the significance for targeted post- flood relief, recovery and reconstruction efforts. Quite often, valuable spatial data which are generated by multiple agencies using great deal of technical expertise with huge investments are not put to use when demands for them were the most on the ground. The barrier continues to exist when it comes to understand, share and use the data. Creating a quick geo-spatial database in the Kosi river basin presents a simple methodological approach to convey the powerful message of tracking vulnerability and risk in the event of floods. USE OF NEAR REAL TIME SYNTHETIC APERTURE RADAR FOR FLOOD INUNDATION MAPPING: CASE STUDIES FROM JUNE 2008 AND MAY 2010 FLOODS AT WESTERN PROVINCE SRI LANKA Srimal Priyantha Samansiri Disaster Management Centre, Sri Lanka, E-mail: [email protected]; [email protected] ABSTRACT Lessons learned in the last several years have given clear indications that the prediction and efficient monitoring of disasters is one of the critical factors in decision-making process. In this respect space-based technologies have the great potential of supplying information in near real time. Earth observation satellites have already demonstrated their flexibility in providing data to a wide range of applications: weather forecasting, person and vehicle tracking, alerting to disaster, forest fire and flood monitoring, oil spills, spread of desertification, monitoring of crop and forestry damages. Ministry of Disaster Management is member of the Joint Project Team (JPT) of Sentinel Asia (SA) since 2007 and later DMC, actively involve with the SA operation since February 2009. It is very important to capture satellite imageries in emergency situation as Sri Lanka is prone to frequent floods and landslides. The flood event which occurred in June, 2008 displaced more than 300,000 people and the death toll reported was 23. More than 5 districts were affected, and most of the supply routes became impassable. It would have been more appropriate if disaster management authorities get timely and accurate information of the ground situation for effective emergency response. Satellite imageries can capture such occasion to map affected areas which can provide overview of event for possible emergency planning. Disaster Management Centre requested satellite observation through SA on 17th May 2010 at flood inundation of Colombo, Gampaha and Kalutara districts. JAXA has provided ALOS Palsar radar imageries in near real time basis and was able to detect and analyze flood inundation in the above districts successfully. Radar analysis with ground verification proves the near real time image acquisition and inundation mapping can be utilized for emergency response, future flood mitigation and development planning activities. As an overall outcome of this project, following factors can be highlighted. Firstly, a protocol has been developed to acquire satellite imageries in emergency situation within 48 hours period. Secondly, Data Analysis Node (DAN) has been initiated to analyze imageries and provide valuable maps as an output. Finally, dissemination mechanism has been setup to ensure efficient usage of such information. Reduce Exposure to Reduce Risk 21

LAND USE LAND COVER CHANGES IN CONTEXT TO FLOODS: A CASE STUDY OF DELHI Vidya Satija, Sreeja.S.Nair, Anil.K.Gupta RMS India Pvt. Ltd., Noida; National Institute of Disaster Management, IIPA Estate, New Delhi, India E-mail: [email protected], Mob no. – 9891580811 ABSTRACT Due to the increased population and the scarcity of lands, people are increasingly choosing the unsafe areas for settlements and authorities are choosing the areas for unsafe development of infrastructure and public utilities. Whenever the physical land use planning have underestimated the disaster risk in a hazard prone area, the overall risk is been increased. Delhi has experienced six major floods during the year 1924, 1947,1976,1978,1988 and 1995, 2008 and recently in September 2010. Being the national capital of India, Delhi witnessed significant changes in land use pattern in last 2 decades. The present study was carried out to understand the land use land cover changes and its implications on floods. Eight villages severely affected during the 2010 flood from the north and north-east districts of Delhi were selected for the study. Detailed analysis of all the eight villages was performed to understand the changes land use land cover. Satellite imageries for the year 1992, 1998,2001,2005,2008 and 2010 were used for the study. Land use classification followed by deriving of spatial statics to understand the changes for 5 land Use classes viz, Settlement, water body, open area, agriculture and natural was performed for all the villages. It was observed that all the selected villages have experienced substantial changes in the Land Use and Land Cover pattern. Increasing trend was observed in the built-up area in all the selected villages. On contrary a decrease in open area and water body was also observed till 2008. Post 2010 flood there was an increase in the water body and natural vegetation observed in many villages. Comparative analysis was also made between the Land Use Land Cover patterns derived from classified image of October 2010 of the study area with the proposed land use map of National Capital Region Master Plan 2021. In the NCR Master Plan 2021 is taken consideration of the prevailing flood risk and the proposed Land use for the flood plain was predominantly open areas, play grounds, natural vegetation etc. In the flood prone areas/river beds/banks, no construction or habitation activities should be permitted. On the contrary the comparison has shown significant deviation of the present LULC with the proposed LUSE map of NCR 2021. KEY WORDS: Land use- land cover, Floods, Image Classification, Settlement, Spatial Statistics Reduce Exposure to Reduce Risk 22

Technical Session-2 [Hall B]: GIT4 Infrastructure Management Analysis of Spatio-Temporal Land Use/Cover Change of Devils Lake Watershed using 24 NDVI and NDWI Data 30 B. D. Madurapperuma, P. G. Oduor, K. A. J. M. Kuruppuarachchi, J. U. Munasinghe, 35 L. A. Kotchman 40 46 Measuring the Housing Loss and Analyzing the Flooding Impacts in Ho Chi Minh City, 52 Vietnam 58 Nguyen Thi Phuong Chau, Le Thanh Hoa, Michael Schmidt, Harry Storch, 59 Nigel K. Downes 59 Designing Model of Database and Web GIS Application for Road Performance 60 Management Insannul Kamil, Buang Alias, Abdul Hakim Mohammed, Aditia Andrika Climate Change Affected To Land Suitability: Case Study “Eucalyptus Suitability in Nakhonratchasima Province, Thailand” Rujee Rodcha, Chudech Losiri, Asamaporn Sitthi GIS-Based Site Location Procedure For Solid Waste Management In Sri Lanka: A Case Study of Moratuwa Urban Council Suresh Pakshaweera, NandanaPathirage Feasibility of using Energy Cost Based Geoinformatics Models for Decision Supporting In Infrastructure Development and Management Projects in Elephant Ranging Areas M.S.L.R.P. Marasinghe, Ranjith Premalal De Silva, N.D.K. Dayawansa A Plot Based Land Conversion Mapping on Human Behaviour Process in the Context of Household Stability on Low-Lying Areas GPTS Hemakumara, Ruslan Rainis Spatial Analysis and Geographic Information to Reduce Risk in Community; Nong Pling Model Ladachart Taepongsora1, Choosak Nithikathkul, Pipat Reungsaeng, Bangon Changsap, Supaporn Wannapinyosheep, Pissamai Homchumpa, Anothai Trivanich Chalobol Wongsawad The Role of Geo-Spatial Tools In Rice Supply Chain Management During Disasters Suby Anthony Geoinformation Technology for Efficient Infrastructure Management - Assessment of Geotechnical Property Variations In Five Soils From Sri Lanka P.L. Dharamapriya, H.A.H. Jayasena Reduce Exposure to Reduce Risk 23

ANALYSIS OF SPATIO-TEMPORAL LAND USE/COVER CHANGE OF DEVILS LAKE WATERSHED USING NDVI AND NDWI DATA B. D. Madurapperuma1*, P. G. Oduor2, K. A. J. M. Kuruppuarachchi3, J. U. Munasinghe4 and L. A. Kotchman5 1, 2 Department of Geosciences North Dakota State University, P.O. Box 6050, Fargo, ND 58108, USA. Email: [email protected], * corresponding author, [email protected]. 3Department of Botany, The Open University of Sri Lanka, Nawala, Nugegoda, Sri Lanka. Email:[email protected]. 4Department of Zoology, The Open University of Sri Lanka, Nawala, Nugegoda, Sri Lanka. Email: [email protected] Dakota Forest Service, 307 – 1st Street East, Bottineau, ND 58310, USA. Email: [email protected]. ABSTRACT Devils Lake is the largest natural lake(9,800 km²) in North Dakota, United States. It is a closed basin lake, which is characterized by saline, hyper-eutrophic, and historic flooding. In this study, we attempt to determine mappable water-related in dice svis-à-vis to land-cover data within the Devils Lake sub-watershed region using Landsat 5 Thematic Mapper (TM) images from 1991 to 2005.Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI)vector datasets were derived from Landsat 5 TM satellite imagery using ENVI EX®to evaluate the multi-year changes of vegetation and water resources. In addition, land-use and land-cover change were assessed using National Land Cover Database (NLCD). ArcGIS Explorer online platform was used to map Federal Emergency Management Agency (FEMA) 100 year flood zones and associated land-cover change. Results show that the NDVI is negatively correlated with NDWI. For example, in the periods of 1991-1994 and 2000-2005 NDVI values greatly increased in this region, while NDWI values considerably decreased in those time steps. The dramatic decline of greenness was observed in the 1994-2000 period. Devils Lake experienced a historically unprecedented rise, especially from 2000 onwards, causing the overflow of the Sheyenne River and overtopping downstream Stump Lake leading to prospects of catastrophic flooding in the area. This study also provides a summary of how land-use and land-cover have changed due to human activities and the current flood-prone condition of the Devils Lake. KEY WORDS: Devils Lake, Flooding, NDVI, NDWI overflown to the Stump Lake and further downto the Sheyenne River, which causes downstream flooding. 1. INTRODUCTION Sincethe Sheyenne River is a tributary of the Red River, floodingposes potential environmental risks Devils Lake is a closed basin lake in North Dakota, downstream. North Dakota State Water Commission characterized by large fluctuations in water levels in (2011) proposed three strategies to mitigate the response to climatic variability (Anar, 2011). It is the catastrophic flooding in Devils Lake, consisting of: largest natural water body in North Dakota, and is (1) upper basin water management to reduce the situated in the Devils Lake Basin and adjacent to Red amount of water inflow into the Devils Lake (2) River Basin at the north-eastern region of North protection of properties such as roads levees, and Dakota (Fig. 1).Historic flooding revealed that the relocation (3) make emergency outlets to carry runoff from agricultural lands caused a nutrient spike excess water from the Devils Lake to reduce the due to circulation of saline water from the inundated flooding impacts at the downstream area.For the third areas. The dissolved solids concentration in Devils strategy,land managers are interested in finding Lake is considerably high and fluctuates with water which lands may be available for emergency outlets level (Sando and Wiche, 1990). Considering the or for temporary water storage. spatial characteristic of land-use at watershed scales, the development of an integrated approach that can This study takes into purview historic land-use and simulate land-use changes and their effects on water land-cover change, climatic variability, and water resources at the watershed is critical (USGS, 2011). balance in the Devils Lake watershed; critical information which can be used as baseline data for It is also important to understand how land-use and management of the Devils Lake watershed. Even land-cover changes affect water balance and river though, the surrounding Devils Lake hills are hydrology. For example, if there is significant land- delineated as one of the top five high-priority areas of cover along the stream channel, the immediate North Dakota upland forests, land-use and land-cover implication would be an increase in water retention change has severely altered the hydrology of this capacity in the particular geographic area locale. closed basin. In this study, we address impact of Devils Lake has continued to rise and consequently Reduce Exposure to Reduce Risk 24

land-use on downstream water quality by elucidating exported to ENVI® EX environment to detect the the primary loss of riparian arboreal ecosystems that would otherwise mitigate nutrients. changes between the 1991 to 2005 time steps based on Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI) using an image difference tool.Each NDVI and NDWI data was imported to ArcGIS 9.3® to create change detection maps of the study area. 2.1 Data processing Figure 1: Location map of the Devils Lake basin in 1992, 2001, and 2006 National Land Cover Database North Dakota. FEMA 100 year floodplain is used to (NLCD) datasets originally derived from classified derive the extent of Devils Lake. LandSatTM satellite imagery were imported into ArcMap-ArcInfo® 9.3. A set coordinate system and 2. MATERIALS AND METHODS projection, namely, Universal Transverse Mercator (UTM) North American Datum (NAD) 1983, was Landsat images of the study area were acquired from chosen since the generic linear measurements are in the Global Land Cover Facility. These images were meters and works well with NLCD datasets. NLCD acquired over the span of four different yearsviz., data are raster datasets with 30 m spatial resolution. 1991, 1994, 2000 and 2005. Table 1 shows the details Each imported NLCD dataset was spatially clipped to of the acquired images including the Worldwide Devils Lake watershed boundary spatial extents using Reference System (WRS), path and row information. a basic Minimum Bounding Rectangle (MBR). In the The images are on a scale of 30 m spatial resolution. GIS environment, the resulting image was exported using the clipped extents and the pixel size set to 0.00027777 decimal degrees (30 m) for the associated latitude. The total area of the clipped raster was determined by multiplying the number of pixels by each set pixel area. Raster reclassification was performed where minor NLCD classes were collapsed into defined major categories, for example, class 41 (deciduous forest), class 42 (evergreen forest), and class 43 (mixed forest) were reclassified as forest. The total number of classes was 7, namely, (i) Forest (ii) Urban Developed (iii) Grassland (iv) Pasture/Hay (v) Cultivated Crops (vi) Wetland (vii) Water. 2.2 Accuracy Assessment Table 1. Landsat time series scenes used in the study. Classification accuracy refers to the extent of correspondence between the remote sensed data and Satellite Sensor Date Resoln WRS Path Row reference information (Congalton, 1991). In this Landsat 5 TM 31/08/1991 30 m 2 31 27 study, accuracy assessment for the NLCD land-cover classifications was performed by generating error Landsat 5 TM 22/07/1994 30 m 2 31 27 matrices. The total number of cell counts in each Landsat 7 ETM+ 30/07/2000 30 m 2 31 27 land-cover type is set out in a square array of rows Landsat 5 TM 18/06/2005 30 m 2 31 27 and columns. The columns in the matrix represent the reference data (actual land-cover) and the rows This data was freely available and was obtained in represent assigned (mapped) land-cover types the form of individual bands ranging from 1 to 7. (Congalton, 2008). Error matrices have been used These individual bands were then stacked to to correlate between two data sets with a high represent satellite imageryusing ENVI® 4.5. Layer coefficient of agreement, Khat, Kˆ ,(Kappa) value indicating how two datasets are similar or dissimilar stacking creates a new multiband file from the input bands which are resampled and re-projected. Upon (Oduor et al., 2012). The set years for the land- layer stacking of the Landsat scenes; they were also cover data were aggregated using Combine () georeferenced to UTM Zone 14 North from WGS-84 function in Raster Calculator of Spatial Analyst extension of ArcMap-ArcInfo®9.3, for example, Datum, then converted to calibrated radiance using ENVI®4.5 software. Thereafter each data set was Combine ([„landcover_92], [„landcover_01]). The Reduce Exposure to Reduce Risk 25

attribute table of the resulting calculation was periods (Fig. 4), which span 15 years, are compatible for detecting land-cover change in the Devils Lake exported as a .dbf file (for example Oduor et al., basin. Figure 4 clearly shows that the extent of the Devils Lake increased especially in 2001 and 2006. 2012). A crosstab query table for the .dbf file was However, in 1992 the Devils Lake shrunk mainly in done using Microsoft® Access. The crosstab result those areas that occupied wetlands and the forests in was exported to Excel® and Kappa derived from (see the western and central parts respectively. The grassland extent in 1992 is prominent. It is interesting also, Jensen, 2005): to note that the pastureland, which is anthropogenic land-use, is mainly confined to the lower basin of the kk Devils Lake (Fig. 4).    Kˆ Figure 5 shows the land-use and land-cover dynamics  N  xii   xi  xi for the years 1992 to 2006. In 1992 the land-cover i1 i1 acreages of grassland, wetland, and forest were comparatively higher than in 2001 and 2006. By N 2  k xi  xi contrast, urban, water, and crops in 1992 extents were comparatively lower in 2001 and 2006. The  pasture land extent for the entire period is more or less similar. The land-use and the land-cover extent i 1 for 2001 and 2006 are comparable. Of all the classes, crops exceed 30% of coverage for the entire period. Khat is coefficient of agreement, N is the total number of sites in the matrix, k is the number of rows Figure 2: Land-cover change in the Devils Lake basin in the matrix, xii is the number in row i and column i, for 1991-1994, 1994-2000, and 2000-2005. Image x+i is the total for row i and xi+ is the total for column difference was performed using NDVI and NDWI i. using ENVI EX® software. 3. RESULTS 3.1 NDVI and NDWI relationship Figure 2 shows the land-cover change based on the NDVI and the NDWI indices for the periods of 1991- 1994, 1994-2000, and 2000-2005. The positive and the negative changes in land-cover correspond to the big increase and big decrease of the particular area respectively. Results show great increase of NDVI between 1991- 1994 with a great declining of NDWI. In contrast, in the 1994-2000 periods there was no change for NDVI, but NDWI increased considerably. Between 2000-2005 NDVI showed both an increase and a decrease; on the other hand NDWI showed only a decrease. 3.2 Climatic variability Figure 3 shows the historic rainfall and temperature variation in the Devils Lake for the years 1991, 1994, 2000, and 2005. Rainfall increased from April to June with a major peak in June. The rainfall exceeded 100 mm threshold for 1994, 2000, and 2005. A minor peak of rainfall was observed for September to November months. The temperature was below zero from November to March and then it increased to a maximum of 21°C for the months of March to August. The temperature then declined to zero from August to November. 3.3 Land-use and land-cover change Figure 4 depicts how land-use and land-cover changed from 1992 to 2006. Although for the NDVI/NDWI analysis in Figure 2 we used Landsat TM images for years 1991, 1994, 2000, and 2005, we had to utilize classified land-cover images of 1992, 2001, and 2006 in Figure 4 since those were the only available images that covered the time line. We assumed that 1991-2005 (Fig. 2) and 1992-2006 Reduce Exposure to Reduce Risk 26

Figure 4: Land-use and land-cover change detection using National Land Cover Database (NLCD) for the years 1992, 2001, and 2006. Figure 3: Climate diagrams for the Devils Lake basin 3.4 Cellular automata (CA) model for the years 1991, 1994, 2000, and 2005. The area above the dotted line represents rainfall values The constrained CA-based simulation models were greater than 100 mm. The annual rainfall for 1991, developed using numbered land-use and land-cover 1994, 2000, and 2005 was 576 mm, 627 mm, 601 types and 3D models were generated from mm, and 499 mm respectively. The climatic data was commission error (Fig. 6a, b and c). In the 3D graphs acquired from http://www.prism.oregonstate.edu/. diagonals showed a significant correlation as indicated by relatively high values. For example, in 1992-2001 forest and cultivated crops showed 0.77 (error = 0.23) and 0.75 values respectively. In 2001- 2006 all diagonals showed a high correlation, where water category showed a value of 0.96. Similarly in 1992-2006, 1.0 (no error) value was observed for water, and a 0.85 value for the developed category. Figures 6d, e, and f show omission errors and conditional Kappa results. In 2001-2006 land-use and land-cover categories showed high Khat values with low omission errors. We calculated the overall coefficient of agreement (Khat) and the results showed that it was 0.39 in 1992-2001, 0.84 in 2001- 2006, and 0.38 in 1992-2006. 4. DISCUSSION This study provides how land-use and land-cover changes influence the water budget, when an area is exposed to historic flooding scenario. We used remote sensing techniques to address impact of land use on downstream water quality addressing the primary loss of riparian arboreal ecosystems that would otherwise mitigate nutrients. Reduce Exposure to Reduce Risk Figure 5: Land-use and land-cover change for years 1992, 2001, and 2006 in the Devils Lake basin. NDVI and NDWI are useful parameters in predicting temporal vegetation changes. In the 1991 to 1994 period NDVI showed a big increment due to high occupancy of forest and cropland. According to land- use and land-cover data from 1992 (Figs. 4, 5) 43% of land was occupied by vegetation. The NDVI 27

increment for the 2000-2005 periods is mainly due to North Dakota State Water Commission may increased extent of cultivated crops (Fig. 5). The compensate land owners for temporary water storage forest decline from 9% to 1% for the period of 1992 during a flooding stage. This may reduce the impact to 2006 may considerably influence the water budget of catastrophic flooding at the downstream areas such in this region. The Devils Lake water level has risen as Stump Lake and Sheyenne River. Prior to rapidly since 1993 in response to above-normal implementing a flood mitigation project, it is precipitation and in 1993; 50,000 acres of land worthwhile to do a feasibility study using cost and around the lake have been flooded (Wiche, 1998). In benefit analysis. the 1994-2000 period the Devils Lake water level changed from 433 m to 441 m, which overflowed to Devils Lake hills are delineated as one of the five the Stump Lake at about 433 m priority areas of North Dakota upland forests threshold.(http://nd.water.usgs.gov/devilslake/images (Kotchman, 2010). The adjacent riparian forestmay /DLPOR_Mar2010.pdf). This could be the primary control flooding minimally through uptake and reason for a significant change of NDWI between majorly by forming a natural flood-wall preventing 1994 and 2000. The other reason would be high bank erosion or collapse to a certain extent. Forest annual rainfall, which also contributed to a marked Stewardship Program between North Dakota Forest increment of the NDWI value (Fig. 3). Service and private forest landowners is beneficial in mitigating effects of flooding in North Dakota. From Fig. 4 the prominent land-use type at the lower Cultivation of high water up-taking crops will be an Devils Lake basin is pastureland, which can be advantage of maintaining water balance in this utilized to extend the catchment of the watershed. region. ( (( ( (( Figure 5: Error analysis of NLCD data for the Devils Lake basin Reduce Exposure to Reduce Risk 28

5. CONCLUSIONS Sando, S.K., and Wiche, G.J. 1990, Variability of Dissolved Solids in Devils Lake, North Our study shows that NDVI negatively correlated Dakota.Proceedings of the North Dakota Water with NDWI, which implies that densevegetation Quality Symposium, Fargo, North Dakota. contributed to water budget balance in the Devils Lake catchment. Increasing conversion of primary USGS, 2011, Devils Lake Basin in North Dakota. land cover to agricultural crops results in declining http://nd.water.usgs.gov/devilslake. forest cover.The significant loss of riparian forest is lucid in relation to nutrient uptake and may pose Wiche, G. J., 1998, Lake Levels, Streamflow, and severe ramifications for ambient water quality. Our Surface-Water Quality in the Devils Lake Area, results may be useful forland managers to grasp North Dakota, Through 1997.USGS Fact Sheet impact of land-use practices and land-cover change FS-033-98. over a15 year period. Furthermore, State Water nd.water.usgs.gov/pubs/fs/fs03398/pdf/fs03398 Commission may use this information when they .pdf implement the Devils Lake emergency outlet project. ACKNOWLEDGEMENTS This projected was funded by U.S. Department of Agriculture Forest Service award #10-DG-11010000- 011 and CFDA Cooperative Forestry Assistance # 10.664. Flow through funding was handled by North Dakota Forest Service. REFERENCES Anar, M.J., Madurapperuma, B.D. and Oduor, P.G., 2011,Surface hydrological modelling for identifying infiltration zones in Devils Lake watershed, North Dakota.Proceedings of the North Dakota GIS Users Conference 2011, October 11-13, Grand Forks, ND. Congalton, R., 2008, Thematic and positional accuracy assessment of digital remotely sensed data.Proceedings of the 7th Annual Forest Inventory and Analysis Symposium, 2005, edited by G. Reams, P. Van Deusen, and P.McWilliams, General Technical Report WO- 77, USDA Forest Service. Congalton, R. G., 1991,A review of assessing the accuracy of classifications of remotely sensed data. Remote Sensing of Environment, 37, 35- 46. Jensen, J.R., 2005, Introductory Digital Image Processing: a Remote Sensing Perspective, 3rd edition, Prentice Hall, Upper Saddle River, NJ. Kotchman, L. A., 2010,North Dakota Statewide assessment of forest resources and forest resource strategy. Resource Bulletin, North Dakota Forest Service. North Dakota State Water Commission, 2011.Mitigation Plan. http://www.swc.nd.gov/4dlink9/4dcgi/GetConte ntPDF/PB-1956/Mitigation%20Plan.pdf. Oduor P.G., Kotchman, L., Nakamura, A., Jenkins, S., and Ale, G., 2012, Spatially-Constrained Forest Cover Dynamics using Markovian Random Processes.Forest Policy and Economics,20, 36-48. Reduce Exposure to Reduce Risk 29

MEASURING THE HOUSING LOSS AND ANALYZING THE FLOODING IMPACTS IN HO CHI MINH CITY, VIETNAM Nguyen Thi Phuong Chau1, Le Thanh Hoa2, Michael Schmidt3, Harry Storch4, Nigel K. Downes5 (1) and (2) Geography Department of the University of Social Sciences and Humanities of Ho Chi Minh City, Vietnam [email protected] and [email protected]; (3), (4) and (5) Environmental Planning Department, Cottbus Brandenburg Technology University, Germany. ABSTRACT: Ho Chi Minh City, a coastal city, has pressures of climate change with sea-level rise, strong-storms effect. The heavy rains usually visit the City together with high tide and cause mass flood. The rapid urbanization has caused the infrastructure problems, especially drainage system. This has pushed the flooding more seriously in this City and affected heavily on the housing area. This research aims to identify the losses on housing typologies and the socioeconomic impacts by flood based on GIS and remote sensing techniques with GeoEye image 2010, cadastral data, population census data, and the field-surveys in urban-flooded area. The 1/2000-scale-cadastral map combined with GeoEye image are updated and classified to identify housing typologies in flooded area by remote sensing technique. The flooding statistic data is analyzed on land- use map and the survey-data are used to identify the socioeconomic impacts in urban flooded-area. With above results, the GIS technique measures the urban flood impact on housing categories to calculate the total loss on urban housing area. The research results can support for re-development of the City, especially for infrastructure new-planning. KEY WORDS: climate change and rapid urbanization, flooding impacts, housing categories, flooded-housing area, housing loss by flood. 1. INTRODUCTION al., 2012). Flooding has destroyed properties, impacted 1.1. Problems of Flooding in Ho Chi Minh City on people socioeconomic aspects, and loss of their lives Ho Chi Minh City is the fifth of the world’s twenty of the cities (James et al., 2001; Tucci, 2001; Martin et coastal cities with high risk of flooding by climate al., 2007). In which, poor people is the highest change (Nicholl et al, 2007). It had reported to receive vulnerable group in the floodplain areas (Braga, 1996; more storms and rainfall intensity, and the more heavy Doracie et al, 2000; Parkinson, 2003; Sulong et al, rains have visited in the summer monsoon (Martin et al., 2012). 2007; ADB, 2010; The World Bank, 2010). Long-term urbanization in Ho Chi Minh City has been mentioned in 2. METHODOLOGY many reports and it has caused the expansion of built-up Base on the GIS analysis (GIS - geography information area into the agriculture and low land area (Le, 2007; Ho, system), the flooded housing in the floodplain area was 2008; Luong, 2008). The rapid urbanization in this City identified on the City’s flood map of year 2010. has pushed the increase of housing area, the Floodplain area was statistically collected during the year encroachment of roads and other landuse in the low land for the largest flood area in 2010. And on this housing areas. The master plan cannot follow up the urbanization area, with cadastral and population census data, the GIS process. It leads the City’s infrastructure with many helps to find out the socio-economic aspects of the problems, especially drainage system (Ho, 2008). These vulnerable population in flooded area, such as flooded result in the rising up of flash flood in Ho Chi Minh City population, population density and housing density in in the monsoon season. flooded area, poor population and poor housing in flooded area. 1.2. Reviews of Flooding Problems in Developing Floodplains in Ho Chi Minh City expand by the Countires geographical and urbanized factors. Therefore, the The similar problems and reasons of flooding have research uses the field-surveys to understand the types of occurred in many coastal and humid tropic cities in flood in flooded area and in different times of seasons. developing countries (Tucci, 1996; Pangilan and Lau, And through the surveys, the socioeconomic and housing 2011; Ramachandraiah, 2011; Sulong et al, 2012). losses are more understanding. Where there are problems with high population and With the 1/2000-scale-cadastral map integrated with urbanization and inadequate landuse, monsoon flood and GeoEye image 2010, GIS and remote sensing techniques flash flood risen up especially in rainy season and even in help to classify the housing typologies and calculate the dry season. And there is an extremely filling of rivers’ total area of each type of housing in floodplain area. tributaries, canals, and low land areas during Housing types are categorized by the structure and the urbanization (Tucci, 1996; Doracie et al, 2000; Haque et Reduce Exposure to Reduce Risk 30

quality. The different types of houses have different rain season, the tidal flood and flash flood can both housing losses. occupy the same areas and cause the new flood type – the The housing loss was identified and calculated by: mixed flood of both tidal and flash floods. Where: 3.2. Flooding occupation L = Total cost of loss With the above description of flooding, this research Si = Total area of housing types in flooded area analyzed the flooding area of three flood types: the tidal Ci = Cost per square meter for flat rising up of flood, the flash flood, and the mix of tidal and flash each type of housing, and floods. The flood level was calculated by the highest time i = Value of each housing type in the rain season of the area. The map in Figure 1 showed flash flood occupied about 56.4 percent of total With the average cost per square meter of housing flat flood area. The level of flood ranked from 0.25 to 0.4 rising-up of each housing type collected from the field- meter. And the highest area was concentrated in the old- survey, GIS combines them to the total flooded housing urban areas and the lower was in the new-urban areas. area to calculate the total loss of housing in floodplain The area of tidal flood was about 30.8 percent of total, area of the City. The cost value was in Vietnam Dong. It distributed near the water catchment of Saigon River, its was exchanged for US dollar (USD) value at the same tributaries and canals. The flood level was from 0.2 to time of the surveys in Ho Chi Minh City. 0.6 meter. And the mixed flood of tidal and flash floods was mainly distributed in two big areas along the canals 3. FLOOD EXPOSURE IN HO CHI MINH CITY of Nhieu Loc – Thi Nghe and Kenh Doi – Kenh Te - Tau 3.1. Types of flooding Hu – Ben Nghe. These canals are affected strongly by Flooding in Ho Chi Minh City comes by two main tide sediment. The highest level was up to 0.5 meter. It reasons: by tide and by rain. Another reason is by the could be seen in the map, the flash flood areas spread far water release of the hydroelectric systems from the and wide in the urban area, while the tidal flood areas upstream areas. However this reason does not much trended to concentrate in big areas. The high-flood-level impact on the City because there is the upstream-flood- area of flash flood was in the old Districts, where the water-control-system on this area (To, 2008). Flooding long-time urbanization has led big problems on old and by tide, as tidal flood, comes in every month even in the overload drainage system. dry season, normally in the low land of downstream Flooding in Ho Chi Minh City is mostly affected by areas. But in the rain season it becomes stronger and geographical and typographical conditions. However, exposes in larger area. Flooding by rain, can be called many flash flood areas with high flood level of 0.4 to 0.5 flash flood, usually is in the rainy season. It mostly meter were in the high land areas. One example was in happens in the urbanized area. Urbanization of this City Ward 12 of District 10 with three meter of elevation but has pushed the flash flood increase because of the 0.4 meter of flood. Another area was in Ward 3 of Go following reasons: Vap District with over eight meter of elevation and 0.25 meter of flash flood. It means that the drainage system is 1. There is a transformation of built-up area onto the inadequate to the urban development in these areas. agricultural and green areas, Figure 1: Flooding Exposure in Ho Chi Minh City 2. The development planning has ignored the protection and conservation of rivers and canals, Tidal flooding visits the areas longer than flash flood, the low land areas, and the natural water two times a day and about three to four hours a time in reservoirs, 3. There are the encroachment of rivers and canals and low land areas for housing; the siltation and sediment of riverbeds, reservoirs and irrigation canals during urbanization, 4. There is no drainage system planning integrated in the master plan during the development planning. The current drainage system is overload in the inert-center and incomplete in the new urbanization areas, 5. The stormwater was not separated from the waste water in the drainage system, and 6. Many packages of the current environmental sanitation projects, water environment and improvement projects have blocked some important drainage flows and pushed up the water-flows on the ground surface. However the Saigon River goes through this City with 80 kilometers long of main stream and has many tributaries widely on the City’s area. When tidal flow comes in the Reduce Exposure to Reduce Risk 31

the areas near the canals; or nearly whole day in the areas However, the vulnerable population in area of tidal flood surrounded the main river. Flash flood usually visits the and in the mixed area of flash and tidal floods, described City’s areas just fifteen to thirty minutes, however, it has in Figure 3, was little higher than in the flash flood area. many impacts on the areas and has brought about This could be explained clearer by the field-surveys. tangible and intangible losses to people. Most of poor and low-income people live near and along 4. FLOODING HAZARD ON THE SOCIO- two canals where the flash flood and tidal flood both ECONOMIC ASPECTS occupied (see in Figure 1 and in section 3.2). This is the Flooding in Ho Chi Minh City did not zone in area but encroachment result of the poor along the canals and low wildly spread on the City’s ground surface. It not only land areas during the urbanization. The incomplete impacts on infrastructures and urban services, but it also drainage and sanitation systems have caused more floods affects on housing and living environment of people. and impacted more people of the areas. Flooding impacts on socio and economic aspects can explain to the losses of their livelihoods, daily activities, 4.2. Housing Damage and Economic Loss as well as their response capability to future flooding Flood water when visits the house, it brings sediment and adaptation. bad environment into the house. Depends on the flood level and the time of visit, housing may have stronger Figure 2: Flooded area and Flooded Housing Area in Ho damages. Through the surveys, households lost their Chi Minh City monthly incomes when flood visited on their business 4.1. People in Flooding Areas areas, commercial shops, restaurants, food shops, rent From the flooded area and the housing layers from the houses, and small selling areas, etc. map, the total area of housing impacted by flood Responding to the flood visits and their economic losses, calculated by GIS was about 1,739.8 hectares. The people have trend to rise their houses up and upgrade the Figure 2 explained the housing density in the flash flood housing drainage and sanitation. Flats may be risen up area was the highest and in the tidal flood area was several times because flood still coming up. There was lowest. Housing area in the flash flood area was about 64 no support or fund from City and local governments for percent; in tidal flood area was 22 percent; and in the last this activity. The high-income people may afford to rise flood type was 14 percent. up their flats even several times. However, it is a big problem to the poor and low-income people, even to their Figure 3: The Vulnerable Population in Floodplain Area losses on furniture and other household’s utilities. The of Ho Chi Minh City very poor households are forced to live with flood and stand with their livelihood loss and daily difficult conditions. 4.3. Working and Living Conditions in Flood Hazard Area The daily activities would be disordered when flood comes into the house. People have no place for living and working. The furniture and house-wares must be moved up until flood gone; in contrast, they have been damaged by flood. The earning mean such as private transportation would be ruined by the traffic congestion in street flooding. Their assets have been lost and pushed them to the poverty risk in the future, even to the high- and upper-low-income people may become poorer when flood would exist in many years in the City. Other intangible losses such as stresses and health risks would be increased during flood time. Most of poor households in the surveys worried when heavy rains come. They had to move up their house-wares before flood coming and cleaned their houses after it gone. Flood has left the pollute environment and threaten the people’s health. 5. HOUSING LOSS IN FLOODPLAIN AREAS 5.1 Housing Categories in Floodplain Area On 22nd March 2011, the Vietnam Ministry of Construction (MoC) has promulgated the Decision 295/QD-BXD on the collection of construction investment cost for the year 2010 (MoC, 2011). In which housing typologies were described and attached with the construction cost per square meter for each type. From the GeoEye image 2010 and the 1/2000-scale-cadastral map, housing typologies were interpreted in flooded area. Reduce Exposure to Reduce Risk 32

And in combination to housing description of the above damages and looses in the whole flooded area lead the Decision and to the housing condition from field- risk of economic loss of the whole City. surveys, this research divided housing types in flooded area of Ho Chi Minh City into the following four CONCLUSION categories: Flood hazard in Ho Chi Minh City is not only the problem by climate change, but it also by the long - Type A: includes villas, single solid houses with periods of urbanization and the loose control of urban at least three floors, and high-rise buildings with master plan on infrastructure, especially drainage system. more than six floors; The encroachment of river catchment and low land areas leads the disappearing of natural water reservoirs, - Type B: includes single solid houses up to two narrowing of water flows on river and canals, etc. People floors, and apartment-buildings from three to have to face to the socioeconomic difficulties and lose five floors; their capability for the future survival. In planning process, drainage system has not integrated into the - Type C: includes single houses with one floor, master plan. The City and local governments do nothing and apartment-buildings with less than three to support people, they should solve along with people to floors; and improve the City’s drainage system, instead. Future master plan should have landuse control, drainage system - Type D: includes low quality or temporary planning integrated with separate stormwater drainage houses with no floor. system, and environmental planning and solid waste management. It should not also forget to build the flood- Villas are normally located in low land area or near the control data system and integrate into the planning rivers, the former suburban area of the City. They are system. now in the tidal flood area or in the area of both flash and tidal floods. The temporary or low quality houses are REFERENCES located near and along the canals, and in the new ADB - Asian Development Bank, 2010, Ho Chi Minh urbanized area. City Adaptation to Climate Change, The Summary 5.2. Housing Loss Analysis Report, Philippines. The high-income households in flooded area mostly live http://www2.adb.org/documents/reports/hcmc- in the high-flat houses. Many of them had flat rising or climate-change/hcmc-climate-change-summary.pdf house re-building before the visiting of the flood. In Braga B., 1996, Proceedings of a Symposium Hold at contrast, the poor and low-income people had Kingston, Jamaica in November 1996 – Hydrology in experiences with flood in several years before doing The Humid Tropic Environment “Urban Water something. Households who could not afford to rise up Resources Management in Tropical Climates” their housing flats totally; they rose up partly or built the https://itia.ntua.gr/hsj/redbooks/253/hysj_253_275.pd high doorsteps or pavements to prevent the incoming f flood. Many other poor households had to lend money Doracie B. Z. N., 2000. Flood Hazard in Metro Manila: for flat rising or even house rebuilding, especially in the deep-flooded area. Recognizing Commonalities, Differences, and The costs for housing upgrading were collected from the Courses of Action, The Online Journal of Social field-surveys. With the different housing qualities of Science Diliman (January – June 2000), 1(1), 60- housing types, the cost for flat rising is different. To rise 105. up one square meter of housing flat of type A, B, C or D, Haque A. N., Grafakos S., and Huijsman M., 2012, the average costs in turn were 119.7 USD, 71.8 USD, Participatory Integrated Assessment of Flood 52.7 USD, and 43.1 USD. The total housing loss in the Protection Measures for Climate Adaptation in flooded area of the City was described in Table 1 below. Dahka. International Journal of Environment and Urbanization, Vol. 24 (1), 197-213. Housing types Housing area Cost for total flat Ho L. P., 2008, Proceedings: The Sixth Scientific Conference, HCMC University of Sciences (square meter) rising up (USD) “Inundation and Stormwater Drainage in Ho Chi Minh City” Type A 304,000 36,388,800 http://www.nsl.hcmus.edu.vn/greenstone/collect/hnkh bk/archives/HASH018a.dir/doc.pdf Type B 10,976,000 788,076,800 Ho Chi Minh City Website: Introduction of Ho Chi Minh City: The Natural Conditions - Water Resources and Type C 4,451,000 234,567,700 Hydrography http://www.hochiminhcity.gov.vn/thongtinthanhpho/ Type D 1,667,000 71,847,700 gioithieu/Lists/Posts/Post.aspx?CategoryId=17&Item ID=5499&PublishedDate=2011-11-07T16:00:00Z Total 17,398,000 1,130,881,000 James J. McCarthy, Osvaldo Canziani, Neil A. Leary, David J. Dokken, Kasey S. White, 2001, Climate Table 1: Housing Loss by Flat Rising in Flood Area in Change 2001: Impacts, Adaptation and Vulnerability, Chapter 11: Asia, Contribution of Workgroup II to Ho Chi Minh City The Third Assessment Report to The Upgrading flats or pavements and doorsteps of houses in 33 flooded area were just the temporary responds, not a sustainable flooding adaptation. When flood would come again in the higher levels, flat rising would be risen higher. Then the total loss of housing would be much more. People would lose more their capability to deal with flood hazard in the future, not only the poor but also the high- and upper-low-income people. The total income loss of people in flooded area as well as the other Reduce Exposure to Reduce Risk

Intergovernmental Panel on Climate Change (IPCC), Maka Catchment in The Town of Tanah Merah, (New York: Cambridge University Press). Kelantan”. Le V. T., 2007, Economic Development, Urbanization Parkinson J., 2003, Drainage and Stormwater and Environmental Changes in Ho Chi Minh City, Vietnam: Relations and Policies. Management Strategies for Low-income http://www.cicred.org/Eng/Seminars/Details/Seminar UrbanCommunities, International Journal of s/PDE2007/Papers/LE-VAN- Environment and Urbanization: Drainage and THANH_paperNairobi2007-2.pdf Stormwater Management, Vol. 15 No. 2, 115 – 126. Luong V. V., 2008, Urban Development and the Trend of Ramachandraiah C., 2011, Coping with Urban Flooding: A Study of The 2009 Kurnool Flood, India, Climate Change in Ho Chi Minh City, The Selected International Journal of Environment and Papers of the Tenth Sciences Workshop, Vietnam Urbanization, Vol. 23(2), 431 - 446. Institute of Meteorology Hydrology and Sulong M., Noorazuan M. H., Kadaruddin A., and Environment, 10, 369-375. Toriman M. E., 2012. Flash Flood and Martin Parry, Osvaldo Canziani, Jean Palutikof, Paul van Community’s Response at Sg. Lembing, Pahang. der Linden, and Clair Hanson, 2007, Climate Change The Journal of Advances in Natural and Applied 2007: Impacts, Adaptation and Vulnerability, Sciences, Vol. 6 (1), 19-25. Chapter 10: Asia, Contribution of Workgroup II to http://redac.eng.usm.my/html/publish/2011_34.pdf The Fourth Assessment Report to The The World Bank, 2007, Climate Risks and Adaptation in Intergovernmental Panel on Climate Change (IPCC), Asian Coastal Megacities, A Synthesis Report, (New York: Cambridge University Press). (Washington, DC) MoC – Vietnam Ministry of Construction, 2011, To V. T, 2008, Water Drainage in Ho Chi Minh City. Decision 295/QD-BXD - The Promulgation of http://www.vncold.vn/Web/Content.aspx?distid=136 Collection of The Construction Investment Cost for Tucci C. E. M., 1996, Proceedings of a Symposium Hold The Year 2010, Hanoi. at Kingston, Jamaica in November 1996 – Hydrology Nicholls R.J., Hanson S., Herweijer C., Patmore N., in The Humid Tropic Environment “Urban Drainage Hallegatte S., Corfee-Morlot J., Château J., Muir- Planning in Brazil” Wood R., 2007, OECD Environment Working Paper https://itia.ntua.gr/hsj/redbooks/253/hysj_253_275.pd No.1. “Ranking Port Cities with High Exposure and f Vulnerability to Climate Extremes” Tucci C. E.M., 2001, Urban Drainage in Specific http://www.oecd.org/environment/environmentworki Climates, Volume I: Urban Drainage in Humid ngpapers.htm Tropics, International Hydrological Programme, Pangilan R. S and Lau T. L., 2011, The Third UNESCO, 2001, No. 40 (1), (Paris). International Conference on Managing Rivers in the Twenty-first Century: Sustainable Solutions for Global Crisis of Flooding, Pollution and Water Scarcity, Penang, Malaysia. “Flooding in Sungai Reduce Exposure to Reduce Risk 34

DESIGNING MODEL OF DATABASE AND WEB GIS APPLICATION FOR ROAD PERFORMANCE MANAGEMENT Insannul Kamil1, Buang Alias2, Abdul Hakim Mohammed3, Aditia Andrika4 1,2,3) Faculty of Geoinformation and Real Estate, Universiti Teknologi Malaysia, Johor Bahru, Malaysia 4) Department of Industrial Engineering, Faculty of Engineering, Andalas University, Padang Indonesia Email: [email protected] ABSTRACT: Nowadays, the use of web based geographic information system has been increasingly used for the purpose of mapping an area. Web based geographic information system can be an integrated system with various data owned and will facilitate making decision for planning, maintaining and technology of information system presenting maximum information on evaluating road performances as public infrastructure facilities. Engineering model of database and web GIS application for road infrastructure management in the research done uses two approaches, namely, designing model of system and geographic information system using unified modeling language. GIS application is made based on Web by using php and MySQL database and to present the map, alovmap v 0.96 is used. The made database and web GIS application are able to present information on road condition in real time, where it will facilitate making decision for road performance improvement. The application resulted can check road conditions, road distribution pattern, contribution of an area toward road performance improvement in management system for road infrastructure facilities, so that public infrastructure management can be done effectively, efficiently, and transparently. KEY WORDS: Database, Web GIS, UML, Road Performance Management 1. INTRODUCTION (database) that will facilitate to make good decisions to 1.1 Background plan and maintain. Integrated technology of information September 30, 2009 earthquake in West Sumatera system can contribute to maximum inventory, plan, and Province had destroyed about 85% infrastructures maintain road infrastructure that is called GIS existing in West Sumatera. Data from Satkorlak (Geographical Information System), due tu the system (Coordinator Executive Unit for disaster in West can gather, store, combine, and manage, transform , Sumatera) said that in amount of 178 roads were in manipulate, analyze data which is closely related to highly damage, 63 moderate damage, and 51 light spatial planes of the geo-information, such as, areal damage. planning, urban development, inventory, transportation and other business and economic matters. The other Considering the impacts of the disaster on road damages strengths of the computerization are the improvement of that is highly damage and road function as a motor for time efficiency. Because, the integrated data does not social economic growth, so process of planning and take longer time to make a policy on overcoming the management of roads must be done computerizedly, so existing problems. that the process of post-disaster road planning can run effectively and efficiently. 1.2 The Objectives of the Research The objective of the research is to develop a model of Road planning must be set carefully and efficiently so as database and design an application of geographic to give convenience to the society. In addition, road information system of road maintenance as public assets. maintenance facilities are an obligation which is routin in a period of time to make sure that the condition is 2. DESIGNING INFORMATION SYSTEM maintained. In fact, so far Service of Road Infrastructure Designing information system made is Web Geographic of West Sumatera Province has been still using picture Information System of Road Maintenance. The stages are data in form of manual maps (blueprint, papers, and so designing model of system, GIS, database, and web GIS. on) that can cause trouble in the light of storing and more fatally, they can make lateness and less accurate in 2.1 Designing a Model of System planning, maintaining or analyzing of the development of Designing a model of system is done by using business new roads, since they are done separately and partially by process, data flow diagram, context, system requirement combining separate road map and data. specification, use case documentation, use case diagram, class diagram, activity diagram and user interface, as The use of computer, therefore, is essential, because by well as deployment diagram using integrated information system technology, so mapping an area can be a wholly unity with various data 35 Reduce Exposure to Reduce Risk

2.1.1 Data Flow Diagram 3. Facilitating administrator in managing road due to the The objective of the making of DFD is to demonstrate map directly pointing to the geographical location of the function of the designed system. the road. 4. The ability to recapitulate related data on roads 5. Availability of searching facilities to facilitate administrator to get information from designed GIS. 6. Administrator can print out information on road. 2.1.4 Use Case Diagram To see the relationship between use case with other usecase or with the system, so use case diagram is made (Figure 3). Figure 1: Data Flow Diagram of Road GIS 2.1.2 Context Diagram Context Diagram is a high level flow diagram that describes a whole network of input-output. The aimed system is to describe information flow in a designed information system to identify the start and finish of the input-output into the system. Figure 3: Usecase diagram of road maintenance information system 2.1.5 Sequence Diagram Sequence Diagram draws dynamic collaboration between one object and the other object. Activity Diagram shows an activity flow in sequence based on made sequence diagram. These diagrams are shown in Figure 4. Figure 2: Context Diagram of Road GIS Figure 4: Sequence diagram of road input master 2.1.3 System Requirement Spesification 2.1.6 Class Diagram System requirement spesification is done to identify the Class diagram demonstrates the relationship between needs for the system based on business process, namely, objects that have similar attribute and operation, not actors and functional spesification. about what will happen if they relate each other. A. Actors Actors in the system are: 1. Administrator (Service of Road Infrastructure) 2. NASSRA 3. Officers in the field (Rehabilitation and Maintenance) 4. Body of Regional Development Planning (BAPPEDA) 5. The Government of West Sumatera Province 6. Users B. Functional spesification (System requirements) Functional spesification in the system are: 1. The ability of the system to manipulate data on road. 2. Informing in real time Reduce Exposure to Reduce Risk 36

Table 1: Class Diagram of Designing Road Maintenance 2.2.2 Joining Database Table Information System Having transformed and normalized, then relation of each entity will be obtained. No. Class 1 Road 2 Bridge 3 Administrator 4 Check Sheet for road maintenance 5 Check Sheet bridge maintenance 6 User Figure 7: Relational Data Basis 2.3 Designing User Interface Designing user interface based on the needs for designed system. User Interface is a medium of communication between user and the system, so that designing must facilitate user in using the system in order to achieve efficiency. Figure 5: Class Diagram of Designing Road Maintenance Information System 2.2 Designing Database System Designing database system is done by transforming class diagram in to related relation, normalization of table of data, joining of table of data, implementation of database with MySQL. 2.2.1Transforming Class Diagram into Related Relation In the stage, we will determine the relation happening inter-class. Figure 8: Web Road SIG page Home Figure 6: Entity Relationship Diagram of Road SIG 37 Reduce Exposure to Reduce Risk

3. ANALYSIS 3.1 Actual System and Designed System Comparison between actual system and designed system can be seen in Table 2. Table 2: Comparison of strengths between actual system and proposed system User Comparer Actual Proposed   Information   Flow   Structure of   database  storing  Facilitation of  administration process Figure 9: Map page Facilitation of Figure 10: Road GIS Simulation in WEBGIS monitoring Figure 11: Screenshot Menu of Road in Page asset Administrator Facilitation of 2.4 Designing Web Geographic Information System (GIS) for Road Management data searching Designing Web (SIG) using Software dreamweaver 8 Service of Road Data security with programming language PHP, and database MySQL using software xampp Versi 1.6.7. Infrastructure Back up data Map used in web SIG is the map of West Sumatera Province. All information on the map is made by using Facilitation to software Arcview 3.3 and Map Info Profesional 8.5 . Information to be input in the map is data on road get information maintenance. Reduce Exposure to Reduce Risk Accessible anywhere and anytime (internet) Data processing is done computerizedly Faster in making decision Remark: (better) 3.2 Data and Information Flow Conventional system used by Service of Road Infrastructure makes data and information flow unsmooth. Information is only got from reports that contain all information on roads. To get expected information, we have to work harder. Like, finding a document on number of roads existing in the West Sumatera Province, bad condition roads, number of bridges in the process of repairmen, knowing location of certain roads, and so on. The officers or person in charger in this matter must be careful and concise to check the report. Data searching is done among hundreds of road documents which are inefficient actions. 3.3 Database System System of management of database uses concept of RDBMS. Therefore, in the future, it can be developed without destructuring all subsystems of information. 3.4 Values of Web SIG The main objective of making web is to accelerate making decisions in maintaining road. The abilities had by GIS web are: 38

a. Easiness to accessible data [6]. Riyanto. (2009). Pengembangan Aplikasi Sistem b. Integrated data with technique RDBMS that has Informasi Geografis. Yogyakarta: Gava Media. data unity, so that [7]. Hariyanto, Bambang. (2004). Rekayasa Sistem c. Time to proceed data faster Berorientasi Objek. Bandung: Informatika d. Better resulted data quality Bandung. e. Faster data searching. [8]. Munawar. (2005). Pemodelan Visual dengan UML. 3.5 Asset Management Yogyakarta: Graha Ilmu. Roads are public assets that must be kept, and indirectly are one of factors of motor of social economy of West [9]. Nugroho, Adi. (2005). Rational Rose untuk Sumatera Province. Good conditions of road facilitate Pemodelan Berorientasikan Objek. Bandung: transportation or economic movement from one place to Informatika. another place. Problems related to asset are unstructured and not standardized asset inventory, periodic reports on [10]. Roff, Jason.T. (2003). UML A Beginner’s Guide. asset conditions, analysis of asset data for the sake of California: Corel VENTURA. policy making by government, and control of asset distribution in some areas. [11]. Faried Irmansyah. Pengantar database. http://www.IlmuKomputer.com/Artikel/ Pengantar Designed Web GIS can manage asset or road points database. April 2008. existing in West Sumatera. Through good management, Regional Government can inspects road conditions, [12]. Sistem Informasi Manajemen Aset.2010. pattern of road distribution, contribution of an area to http://www.cits-indonesia.co.id/ road maintenance, and transparancy. Thus, such index.php?mod=&idMenuAtas=67&idMenuTab=1 information facilitate Regional Government to develop a 01. 23 Agustus 2010. strategy or policy on road management as well as their impacts on West Sumatera. [13]. Abowo. Aset Manajemen. http://www.abowo.com/2010/aset-manajemen-3/. 4. CONCLUSION 23 Agustus 2010. Of the research, model of database and designed application of Web GIS can be generated where they are [14]. Kendall, Kendall. (2003). Analisis dan able to inform road performances, so that system of road Perancangan Sistem. Jakarta : PT. Indeks management can be done well by basing performance and condition in real time, so that planning, operating, and maintaining road can be done effectively, efficiently and transparently. 5. REFERENCES [1]. Kamil, Insannul, et al. (2012). Analysis of Factors Influencing Awareness of Road Maintenance as Public Asset Using Analytical Hierarchy Process (Case Study: Padang City Indonesia). International Conference on Asset and Facility Management. Padang Indonesia. [2]. Kamil, Insannul, Alias, Buang and Hakim, Mohammed. (2012). Developing Model of Infrastructure Asset Maintenance in Developing Countries Using Fuzzy Screening System-AHP Approach (Case Study: Main Roads in Padang City, Indonesia). International Real Estate Conference. Kuala Lumpur. Malaysia. [3]. Pemerintah. (2006). Peraturan Pemerintah Nomor 6 Tentang Pengelolaan Barang Milik Negara/Daerah: Pemerintaha Republik Indonesia. (Government Decree. (2006) Government Decree No. 6 on the management of Regional/National Assets: Government of Republic of Indonesia) [4]. Jogiyanto. (2005). Sistem Teknologi Informasi. Yogyakarta: Andi Yogyakarta. (Jogiyanto. (2005). System of Information Technology. Yogyakarta: Andi Yogyakarta.) [5]. Mcleod, Raymond. (2008). Management Information Systems. Jakarta: Salemba Empat. Reduce Exposure to Reduce Risk 39