An approach to the model-based spatial assessment of damages caused by urban floods Z. Vojinovic 1 *, J.C.W. Ediriweera 2 and A.K.

Transcription

1 An approach to the model-based spatial assessment of damages caused by urban floods Z. Vojinovic 1 *, J.C.W. Ediriweera 2 and A.K. Fikri 3 1/2/3 Department of Hydroinformatics and Knowledge Management, UNESCO-IHE, Institute for Water Education, Westvest 7, 2611 AX Delft, The Netherlands *Corresponding author, z.vojinovic@unesco-ihe.org ABSTRACT Assessment of floods and flood-related damages is a growing priority due to factors such as increasing urbanisation, unplanned development, changing climate, and increasing operational and maintenance costs. New information and communication technologies offer improved opportunities to address these factors. This paper describes an approach for flood damage assessment as a means of analysing floods and assessing their effects on urban areas. The damage calculation is performed within a framework of the hydroinformatics system where the use of hydrodynamic models, GIS and remotely sensed data is combined within a single platform. For calculation of tangible direct damages the costs are calculated on the basis of model results and cross-referenced against depth damage curves for series of rainfall events. The flood visualisation component of a GIS technology applied here is designed in a way to easily assess potential behaviour of flooding, its rates of rise, evolving flood extents, and associated water depths and velocities. Such framework is useful not only for effective calculation of flood damages but also for evaluation of plans for disaster management activities. KEYWORDS Urban floods; flood damages; modelling; GIS; hydroinformatics. INTRODUCTION Flood management for urban areas is a growing priority due to factors such as the relentless migration to cities, unplanned development, changing climate, and increasing operational and maintenance costs. In particular, the consequences of flash floods in both developed and developing countries have been increasingly devastating over the past decades and the rising trend could reach dramatic proportions with further urbanisation, climate change and rapid population growth. In this respect, new information and communication technologies offer improved opportunities to address these factors. Computer-based flood modelling studies play very important role for identifying flood mitigation measures and for generating flood forecasts as an integral part of urban flood management process. Also, the use of GIS technology has been invaluable for the purposes of flood management. The evaluation of risk due to flooding in urban areas requires a detailed assessment of the potential risks that are possible (see Teng, et. al., 2005, Mark and Parkinson, 2005). In order to evaluate the risk to communities, properties and infrastructure effectively, it is important to estimate the distribution of hazards and the magnitudes of flood-related damages. Such damages are generally divided into tangible and intangible damages. Those that can be estimated and expressed directly in monetary terms are called tangible damages (e.g., damages to properties, infrastructure, etc.); see Penning-Rowsell and Chatterton (1977), Kanchanarat (1989). Tangible damages may be classified further into direct (those that have occurred from a direct Vojinovic et al. 1

2 interaction with the flood) and indirect damages (those that occur because of direct flood impacts). Intangible damages refer to the type of damages that are difficult to identify in monetary terms (e.g., loss of social values, loss of life, anxiety, etc.). Despite the difficulties in quantifying and expressing important variables, some researchers have made their efforts in developing approaches for quantification of intangible damages (see, for example, Lekuthai and Vongvisessomjai, 2001). The methodologies for estimating these categories of flood damages may differ from country to country, but essentially, they all tend to address and quantify various aspects of possible damages; see for example Tang et. al., (1992), Nascimento et. al. (2007). The present paper describes a more generic approach for flood damage analysis and provides some of the main results from an ongoing research. The case study used here is one of the tropical islands located in Caribbean - St Maarten, N.A. In what follows we discuss the importance of urban flood modelling as a means to generate information which is essential for flood damage assessment process. URBAN FLOOD MODELLING Physically based computational modelling has been invaluable for modeling of floods in urban areas. With instantiated models, it is possible to explore the generation of damages and to simulate the consequent effects in response to any control actions. Where flood flows are confined to well-defined conduits, a robust 1D model can usually be instantiated, and used to generate results safe for decision-making. However, the flows generated in urban flood disasters are normally highly complex because the morphology of the urban surface is eminently artificial, with its highly irregular geometry, and is often contrary to natural flow paths. Modelling flows in such complex geometrical situations is difficult. Small geometric discontinuities such as road or pavement curbs can play a significant role in diverting the shallow flows that are generated along roads, through fences and around buildings. Head losses due to flow over or round such structures are difficult to accommodate. Frequently the urban flows are super-critical whereas many of the available modelling products, although they simulate flows that are in reality super-critical, in practice they use modified sub-critical flow algorithms. The use of finite difference methods in conjunction with the reduced momentum equation together with the boundary condition structure inherent to subcritical flow conditions is a standard approach used for numerical simulation of all flow regimes (i.e., subcritical, supercritical and transcritical) in most of the commercial packages. Due to incomplete equations and inadequate boundary conditions used to model supercritical and transcritical flows, such an approach may introduce unrealistic backwater effects, nonamplifying oscillations and other computational instabilities (see, for example, Djordjevic et. al., 2004). There is also the issue of treating the transition from channel flows to over-ground shallow depth flows. This necessitates the coupling of simulations using 1D and 2D modelling systems; see, for example, Hsu et. al., (2000), Djordjevic, et. al., (2005) Chen, et. al., (2005), and Vojinovic, et. al., (2006). In whatever case, Geo-referenced results from either 1D or 1D-2D coupled models can be used to calculate the risk of flooding and to gain insights into the nature of floods and their impacts on communities. In the following section we describe an approach for flood damage calculation which is illustrated on a case study of St Maarten, N.A. For assessment of damages, the rising and lowering of water levels together with the estimation of flow velocities, flow directions, flood durations and inundation extents are important aspects and need accurate representations. Within the framework of the present An approach to the model-based spatial assessment of damages caused by urban floods 2

3 work, several issues that have an influence on accuracy of numerical model results have been evaluated. For 1D modelling, such issues involve the capability of a GIS mapping technique to extrapolate 1D results over the 2D map whereas for 1D-2D modelling the use of different DTM resolution together with different complexity of the terrain features plays a critical role. In this respect, terrain data can be processed to represent either the land surface alone or land surface together with the road network and buildings. For the first case, since the buildings are represented as hollow objects different land use areas are parameterised by means of different hydraulic roughness coefficients, whereas in the second case buildings are represented as solid objects. AN APPROACH TO FLOOD DAMAGE ASSESSMENT Case study area The case study used here involves some of the most affected catchments on St Maarten Island. St Maarten (Dutch: Eilandgebied Sint Maarten) is one of five island areas (Eilandgebieden) of the Netherlands Antilles, encompassing the southern half of the island of Saint Martin/Sint Maarten. With population growth and degree of development activities experienced over the past decades, flooding has become a growing and serious threat to island residents, Figure 2. To combat the floods the island government has undertaken several steps aimed to identify the most effective flood mitigation strategies (see also Vojinovic and van Teeffelen, 2007). Figure 2: Scene from a flash flood occurred in July 2005, St Maarten N.A.. With respect to the modeling work, initially, the entire Dutch side of the island was modeled with the use of one-dimensional hydrodynamic package MIKE 11 (developed by DHI Water & Environment, and later on this model was coupled with the two-dimensional MIKE 21 model and the two models were used together to describe the flood propagation across the floodplains. Tangible damage assessment Vojinovic et al. 3

4 The work concerning tangible damage assessment was carried out in a customised GIS tool where time varying results from unsteady flow numerical models can be geo-referenced to a spatial framework or grid which includes a model of the terrain. Thus for any time step, velocity, depth and thus hazard can be determined for each grid element in the two dimensional spatial framework. Initially, the property (i.e., building) data were imported into the GIS database and the associated polygons were converted into a point feature format of a single point as described in Figure 3. Figure3: Extraction of ground elevation from DTM All buildings within the study area were classified into residential, commercial and industrial and their respective floor levels were incorporated into the GIS database, Table 1. Table 1: Classification of buildings Building Type Classification criteria Residential small Area < 50 m 2 Residential large Area > 50 m 2 Commercial low Area < 100 m 2 Commercial medium 100 m 2 < Area < 1000 m 2 Commercial high Area > 1000 m 2 Industrial low Area < 100 m 2 Industrial medium Area > 100 m2 The damage estimation was based on the general assumption that the monetary damage depends on the type of the building and its size. Based on historical data, flood stage damage curves have been developed for each category of buildings. By considering the types of industry, machinery, turnover and national benefits, the damage values per unit area are assigned with depths, for industrial buildings. In the same manner flood damage values have been assigned for commercial buildings per unit area, based on their turnover and trade. Based on historical data, flood stage damage curves have been developed for each category of buildings. By considering the types of industry, machinery, turnover and national benefits, the damage values per unit area are assigned with depths, for industrial buildings. In the same manner flood damage values have been assigned for commercial buildings per unit area, based on their turnover and trade. An approach to the model-based spatial assessment of damages caused by urban floods 4

5 Seven flood stage damage curves have been established for the study area. All damage curves give damage values per unit area except for residential buildings. For residential buildings, unit values have been assigned for each building, based on residential category as shown in Figure 4. Figure 4: Flood damage curves used in the study. Final direct damage cost was obtained by multiplying the building area and an associated unit cost of flood depth, as calculated by a numerical model (this was done for different rainfall events), Figure 5. Direct Damage Cost ($US) Figure 5: A GIS presentation of direct damage costs. In this work, damage calculation was carried out for five major rain events concerning two scenarios, namely after rehabilitation and before rehabilitation. The associated damage costs are given in Figure 6. Vojinovic et al. 5

6 before rehabilitation Direct Damages after rehabilitation Cost ( $ miil) yr 50yr 20yr 10yr 5yr Rain event Figure 6: Direct damage assessment for different scenarios. Indirect flood damage assessment In this work, the value of indirect damage was calculated as a fixed percentage of the direct damage. This assumption was believed to be acceptable for practical reasons as the time required for a detailed analysis of indirect damage is far too great to be justified in an individual flood study (see also James and Lee, 1971). To complete the work of this study and to calculate indirect damages, the values proposed by Kates (1965) were considered as reasonable enough and as such they were applied in the present work. These values are: 15% for residential land, 35% for commercial, and 45% for industrial. An overview on total tangible damage is given in Tables 3 and 4. Table 2: Estimation of indirect damages before rehabilitation Rainfall Direct damage Indirect damage Tangible Event Res Com Ind Res(15%) Com(35%) Ind(45%) damage 1:5 year :10 year :20 year :50 year :100 year Table 1: Estimation of indirect damages after rehabilitation Rainfall Direct damage Indirect damage Tangible Event Res Com Ind Res(15%) Com(35%) Ind(45%) damage 1:5 year :10 year :20 year :50 year :100 year Intangible damage assessment An intangible damage assessment has always been regarded as a very complicated and difficult task due to the range of issues as well as subjectivity. In our work we have adopted the methodology introduced by Lekuthai and Vongvisessomjai (2001). In order to find the relationship between anxiety, productivity and income, directly affected following predominant factors were identified. An approach to the model-based spatial assessment of damages caused by urban floods 6

7 Anxiety Productivity = f (Flood depth, Land use) =f (Anxiety, Income) From the sociological survey carried out on St Maarten Island, it was found that, over 80% of people feel that they will not be affected by a flood depth of less than 0.1 m. At the same time, over 80% of people feel that they will be greatly affected if the floodwaters are higher than 0.9m. This can be explained by the tragic experience of flash floods occurred in the past. Other details complied from this study are given in Figure 7. This figure gives the value of intangible damages calculated for each scenario and converted into the monetary value. In the present work, the so-called API methodology (Lekuthai and Vongvisessomjai, 2001) was modified according to the field data. Following this, the new relationships were established to estimate intangible damages for St Maarten Island: Anxiety 1. Relationship of flood depth and anxiety was derived as FloodDepth = 0.05e 2. Relationship between the flood depth and loss of income was derived as 6th order polynomial function. Figure 7: Estimation of intangible damages. Annual benefit calculation The total damages caused by different flood events (i.e., different flood return periods) are used here in order to determine the probability-damage relationship, as presented in Figure 8, Graph-C. The expected annual flood damage can be determined from the probability-damage Vojinovic et al. 7

8 curve. The benefit calculation determines the value of the expected annual benefit from a rehabilitation measure (in this case, construction of a single detention basin was used as an example of a rehabilitation measure), constituting a factor by which all damage data are proportionately reduced. This is achieved by calculating the contribution that each successive flood event makes to the annual average flood damage. Thus, the mean damage expected from two flood events of similar magnitudes is multiplied by the probable interval between the two occurrences (see also Parker et al., 1987). In other words, the expected annual flood damage is the damage divided by its return period, or the damage multiplied by its exceedance probability. Furthermore, in practice, the area under the curve which describes the probability-damage relationship is equivalent to the expected annual flood damage. Table 4 summarizes the estimated total damages for a 1:100yr ARI rainfall event. Table 4: Total flood damage for a 1:100 yr ARI. Average Annual Scenario Flood damage(mill$) Total Recurrence Exceedence Direct Indirect Tangible Intangible (mill$) Interval (ARI) Probability (AEP) 1:100 year 1% Existing Rehab The expected annual benefit can be calculated by calculating the difference in the expected annual damage estimates provided by two flood mitigation scenarios, Figure 8, Graph-C. Figure 8: Results from annual benefit calculation Figure 8, Graph-B, illustrates that the peak discharge and flood depth would reduce form 68 cms and 7.2 m depth to 32 cms and 6.4m respectively. This also implies that the benefit of a detention basis for a 1: 100 year flood event (1% AEP) can be reduced to 1: 15 year (6.5%AEP) flood event as shown in Figure 8, Graph-A. Further to that, it can be concluded An approach to the model-based spatial assessment of damages caused by urban floods 8

9 that there is a 6.5% probability of getting a flood with peak flow of 32 cms and depth of 6.4m in each year. Flood damages have been estimated for five flood events and evaluated for two scenarios: before rehabilitation (i.e., existing case) and after rehabilitation (i.e., after construction of a detention basin), as shown in Figure 87, Graph-C. Therefore, the benefits of constructing detention basin can be estimated as follows: Flood damage for existing scenario = 5.2 mill$ Flood damage after rehab = 2.9 mill $ Rehab benefit for 6.5% AEP = 2.3 mill$ Annual expected benefit = 2.3 * 6.5% = 0.15 mill $/ Annum This can be also explained as, if this project is implemented, the society will gain benefits (or be protected from the losses) in the value of 0.15mill $ for each year. CONCLUSION This paper presents an approach that can be used for estimation of flood damages. The study presented here is an attempt to incorporate the GIS technology with computer-based flood modelling results for flood damage assessment and disaster planning. St.Maarten was chosen as a case study for this work. The study suggests several findings. First, there are many issues with respect to the use of a particular modeling technique (i.e., 1D, 2D, etc.). In this respect, terrain features and DTM resolution are very critical aspects that need careful consideration. In case that the 1D model is used, special care must be given to the GIS mapping technique for representation of model results over a 2D map. This is particularly important for irregular and steep terrains. In terms of the damage calculation, the use of a hydroinformatics system which incorporates numerical model results and GIS within the same framework is essential. The flood damage assessment for both tangible and intangible damages was carried out and the results obtained suggest that the use of methodology applied here can be effective to quantify the benefits (or damages) of different rehabilitation alternatives. With the recent provision of some actual flood damage data it is our intention to undertake comparison against this data and to further comment in another publication on the validity of this approach. ACKNOWLEDGEMENT The authors would like to express their gratitude to the Executive Council of the Island Territory of St. Maarten for their cooperation and providing valuable info to prepare this paper. Also, we would like to thank to the Department of Public Works in Philipsburg St. Maarten Neth. Antilles. REFERENCES Chen, A.S., Hsu, M.H., Teng, W.H., Huang, C.J., Yeh, S.H., and Lien, W.Y., 2005, Establishing the Database of Inundation Potential in Taiwan, Natural Hazards, Springer, Djordjevic, S., Prodanovic, G., A., Walters, 2004, Simulation of Transcritical Flow in Pipe/Channel Networks, Journal of Hydraulic Engineering, Vol. 130, No. 12, December 2004, pp Djordjevic, S., Prodanovic, D., Maksimovic, C., Ivetic, M. and Savic, D., 2005, SIPSON Simulation of Interaction between Pipe flow and Surface flow in Networks, Water Science and Technology, IWA Publishing, Vojinovic et al. 9

10 Hsu, M.H., Chen, S.H., and Chang, T.J., 2000, Inundation simulation for urban drainage basin with storm sewer system, Journal of Hydrology, 234, 21-37, Elsevier. James, L. D. and Lee, R. R.: 1971, Economics of Water Resources Planning, McGraw-Hill, New York. Kates, R. W.: 1965, Industrial Flood Losses: Damage Estimation in the Lehigh Valley, University of Chicago, Department of Geography Research Paper No. 98, The University of Chicago Press, Chicago. Kanchanarat, S (1989) Estimation of Flood Damage Function for Bangkok and Vicinity, MSc Thesis, Asian Institute of Technology, Bangkok, Thailand. Lekuthai, A. and Vongvisessomjai, S., 2001, Intangible Flood Damage Quantification, Water Resources Management, Vol 15, October 2001, (20), Springer. Mark O. and Parkinson J., 2005, Urban Stormwater Management in Developing Countries, IWA Publishing, London. Nascimento, N., Machado, M. L., Baptista, M., de Silva, A., 2007, The Assessment of Damages Caused by Floods In The Brazilian Context, Urban Water Journal, Vol 4, Issue 3, September 2007, Penning-Rowsell, E. C. and Chatterton, J. B.( 1977), The Benefits of Flood Alleviation: A Manual ofassessment Techniques, Gower Publishing Company Limited, Aldershot, England. Tang, J., C., S., Vongvisessomjai, S., and Sahasakmontri, K., 1992, Estimation of Flood Damage Cost for Bangkok, Water Resources Management, 6, 47-56, Kluwer Academic Publishers, The Netherlands. Teng, W.H., Hsu, M.H., Wu, C.H., and Chen, A.S., 2005, Impact of Flood DIsasters on Taiwan in the last Quarter Century, Natural Hazards, Springer Vojinovic, Z, and van Teeffelen, J., 2007, An Integrated Stormwater Management Approach for Small Islands in Tropical Climates, Urban Water Journal, Vol 4, Issue 3, September 2007, Vojinovic, Z., Bonillo, B., Chitranjan, K. and Price, R., 2006, Modelling flow transitions at street junctions with 1D and 2D models, 7th International Conference on Hydroinformatics, Acropolis - Nice, France, September, An approach to the model-based spatial assessment of damages caused by urban floods 10