1 INTRODUCTION
1.1 Background and scope
Hurricane Katrina struck the southern coast of the United States on August 29, 2005. The surge caused by the storm overwhelmed the flood protection of New Orleans, Louisiana, leading to one of the worst natural disasters in American history. Catastrophic flooding throughout the city led to economic losses of more than 20 billion dollars, and approximately 1100 people lost their lives due to the flooding in Louisiana; 80% occurred in the flooded area (Jonkman et al, 2009). Following the event, the US government committed and provided the people of New Orleans with 100 year level of flood protection. However, an important discussion remains regarding the residual risk associated with the upgraded protection system.
The city’s risk of flooding is expected to increase in the future due to several factors, primarily due to economic and population growth as the city continues to rebuild. While the updated 1/100 per year protection level is defined by national flood insurance policy precedents, considering the risk specific to the city of New Orleans allows acceptable
and optimized protection levels to be discussed. Risk based principals are applied in other engineering fields, such as nuclear and chemical engineering, and the same approaches can be applied in the context of flood protection (Vrijling, 2001). Risk is generally considered to be a function of the consequences and probability of an event. Therefore the analysis of risk allows potential consequences of a flood event to be considered in decision making. Also, effects of potential risk reducing measures can be systematically analyzed, an important aspect of managing and mitigating overall risk. Thus while decision making regarding flood risk management is complex and includes social, political and economic factors, evaluating the risk provides valuable technical information to be used in decision making for flood protection.
The objective of this article is to quantify the risk to life associated with the upgraded protection system of New Orleans. When evaluating risk, various dimensions of risk account for various consequences of an adverse event, such as economic loss, life loss, and environmental loss. Considering the catastrophic loss of life due to flooding as evidenced by Hurricane Katrina and flood events in other parts of the world, the risk to life is an
Risk to life due to flooding in Post-Katrina New Orleans
A. Miller
Delft University of Technology, Faculty of Civil Engineering and Geosciences, Delft, the Netherlands
S.N. Jonkman
Delft University of Technology, Delft, the Netherlands & Royal Haskoning, the Netherlands
M. Van Ledden
Royal Haskoning, the Netherlands
ABSTRACT: After the catastrophic flooding of New Orleans due to hurricane Katrina in the year 2005, the city’s hurricane protection system has been improved to provide protection against storms with at least a 100 year return period. The aim of this article is to investigate the risk to life in the post-Katrina situation for the New Orleans metro bowl. In a risk-based approach the probabilities and consequences of various flood scenarios have been analyzed to estimate the risk to life. A two-dimensional hydrodynamic model has been used to simulate flood characteristics. Results indicate that – depending on the flood scenario – the estimated loss of life in case of flooding ranges from about 100 to nearly 500. The highest life loss value is found for breaching of the river levees. Probability and consequence estimates are combined to evaluate the individual risk and societal risk for New Orleans. When compared to risks of other large scale engineering systems (e.g. other flood prone areas, dams and the nuclear sector) and acceptable risk criteria found in literature, the evaluated risk exceeds acceptable risk levels. Thus despite major improvements to the flood protection system, the flood risk of post-Katrina New Orleans is expected to be significant. Effects of risk reduction strategies on the risk level are investigated to assist in providing bases for decision making. Results indicate the necessity of further discussion regarding the management and reduction of the city’s risk to flooding.
important consideration in determining flood protection levels. The results of this study are compared to acceptable risk to life criterion found in literature and applied in other engineering sectors, providing insight into the level of risk of the city.
This article is outlined as follows: A brief over-view of the applied approach is described in section 2. This section also describes the main assumptions and methods used in the risks analysis.. Section 3 describes the quantification of the risk to life in New Orleans. Section 4 presents an evaluation through the comparison to criterion in literature and other in-dustries. Finally, risk reduction is discussed in sec-tion 5 where measures to mitigate risk are briefly an-alyzed. Concluding remarks are given in section 6.
1.2 Risk-based approach
A general risk assessment is carried out to achieve the objectives of this research. A risk as-sessment is generally thought to consist of the identi-fication, quantification and evaluation of risk. A general approach to risk management (CUR, 1997) is followed in this thesis. The quantitative risk esti-mate is accomplished through the development of possible flood scenarios, similar to the approach ap-plied in the FLORIS project (Flood Risk In the Netherlands) (Riedstra, 2010). The simulated events give insight into the hydraulic characteristic of flooding and provide information about the potential impacts of flooding. A complete risk estimate would be based on a fully probabilistic analysis, consider-ing all possible load and resistance parameters of the system. However, for the scope of this thesis, a lim-ited number of scenarios are selected based on the way the city can be flooded. Scenarios with the greatest contribution to the overall flood probability are chosen. Failure probability estimates are based on the design criteria. A deterministic damage esti-mate is calculated for each scenario based on the
character- istics
of flood- ing.
Figure 1 Steps of a Risk Assessment (Jonkman, 2007)
The combination of the consequences of the simu-lated events with an estimated event probability of the events determines the risk. The quantified risk estimate is then evaluated by comparing it to limits of ‘tolerable’ or acceptable risk criterion. Proposed limits found in literature and industry are discussed for both flood events and other hazards. Such com-parisons provide insight into the order of magnitude for an acceptable risk and a basis from which risk reduction alternatives can be evaluated. Decisions to reduce the risk are supplemented with information regarding the cost effectiveness of reduction measures. Ultimately, the risk assessment allows de-cisions to be made regarding the acceptability of the risk.
2 SYSTEM DEFINITION
2.1 Threat of flooding to New Orleans
The unique natural environment surrounding New Orleans makes the city highly vulnerable to flooding. Located in the deltaic plain of the Mississippi river as it discharges into the Gulf of Mexico, the city is bordered by Lake Pontchartrain to the north and Lake Borgne to the east. Coastal wetlands surround-ing the city serve as a buffer to the sea. In addition to being largely surrounded by water, it is estimated over half of this metropolitan area is located below sea level, in some areas up to 9.4 feet (3 m) below mean sea level. In this study two types of events that can lead to breaching of the flood defenses are con-sidered, namely storm surge and river flooding
Figure 2 Surroundings of New Orleans (Landsat imagery 2002) Enhanced satellite image of the flood extent due to Katrina (NOAA)
Flooding due to storm surge has impacted New Or-leans repeatedly in the past. In 1965, Hurricane Betsy flooded 164,000 New Orleans homes and took 58 lives in the Orleans parish alone (Boyd 2011). While hurricane Betsy initiated construction of the city’s hurricane protection system, it was unable to withstand the unprecedented surge produced by Hur-ricane Katrina.
Risk management
Qualitative analysis
Quantitative analysis
Risk Evaluation
Risk reduction and control
measures System definition
The second mechanism that can cause flooding to the city is due to a high river event. The Mississippi river delta drains 40% of the United States. Federally mandated flood protection has protected the city since 1927, when massive river flooding inundated much of the alluvial valley including areas of New Orleans. This river protection has been tested most recently in 2011, when record high stages initated the use of emergency levees and bypass structures to divert flow and control water levels in the lower del-ta.
2.2 Study Area and system boundaries
This assessment considers only a portion of the city of New Orleans, that with the highest population and economic value. This area, referred to as the ‘metro bowl’, is an area of roughly 100 km2
(40 square miles) and home to a post Katrina population of roughly 220,000 persons (2011) To the north of the considered area lies Lake Ponchartrain, connect-ed to the sea to the east, and therefore subject to the effects of storm surge. During Katrina, lake storm surge entered the city via drainage canals, causing disastrous flooding. These canals are now closed off where they meet the lake, preventing surge from en-tering the canals and effectively ‘shortening’ the coastline. To the east, lies the IHNC, the Inner Har-bor Navigation Canal which connects to the Gulf
Figure 3 System boundaries for the New Orleans metro bowl
of Mexico. The IHNC is highly vulnerable to the ef-fects of storm surge due to the confluence of two man made waterways at the east end of the canal. The waterways connect to the sea. During Katrina, surge levels in the canal reached 14ft (4.25m). Up-graded protection includes two new barriers that sig-nificantly reduce the opportunity for surge to enter the IHNC. Finally, the Mississippi river flows to the south of the considered area. The water level at this area of the river is affected by both high discharge and storm surge. While there has been no history of flooding to the city due to storm surge at the river
boundary, water levels at the city are becoming in-creasingly influenced by the sea due to continued coastal land loss and ground subsidence and sea level rise.
The river protection at the city is therefore designed to protect from both hydraulic mechanisms. The western boundary of the considered area is the parish or county line, and is chosen for simplification.
2.3 Quantitative flood risk analysis
In this section we first describe the approach and main assumptions for assessing the system reliabil-ity, flood effects and consequences. Results are then presented.
2.3.1 Reliability
To determine the reliability of the considered sys-tem, ideally a complete probabilistic analysis is re-quired and would account for the possible strength parameters of the defence and the possible loads which act on the system such as wave and water lev-el. When quantifying the strength variable, the anal-ysis would consider failure of the defence due to var-ious failure mechanisms such as overtopping and piping. Such types of analyses have been made for the New Orleans (IPET, 2008) and Dutch levee sys-tems.
For simplification in this work, failure probabili-ties are based upon the protection design guidelines in combination with expert judgment. The hydraulic design guidelines determine the overtopping rate of the protection due to a 1/100 per year flood event and a design height is iterated to meet specified overtopping criteria. Robust design specifications, e.g. for armouring and seepage protection, have been applied in order to achieve that the probability of failure due to other (geotechnical) failure mecha-nisms is substantially smaller. Thus although multi-ple failure modes can cause a defense to fail, the failure probability estimates are based on overtop-ping.
To estimate the system failure probability, the system is divided into elements of similar strength and load characteristics where the failure of any el-ement leads to flooding of the protected area). The system is assumed to act as a serial system and the overall failure probability assumed to be approxi-mately the sum of the element probabilities (as-sumed independent failures). Correlations or de-pendencies in failure probabilities between different sections, and the exact contribution of various fail-ure mechanisms would be taken into account in a more detailed analysis.
2.3.2 Development of Flood scenarios
Flooding due to failure of hurricane protection and river levees is simulated. Scenarios are chosen to model the various potential load conditions and the behaviour of the various protection elements. Al-so they are chosen to result in a distribution of con-sequences. The 1D-2D SOBEK overland flow model used is developed by Deltares and is based on a pre-viously developed model of the New Orleans metro area (see de Bruijn, 2006). Breaches along the lake, canal and river breaches are modeled. For each simulation, a deterministic breach location, hydraulic load and breach growth rate have been assumed. The hydraulic boundary condition of each scenario models that of an event with a return period at which failure is assumed to occur, e.g. the 1000 year water levels along the river (see results). Breach growth rates are based on Katrina breach evidence. The result of the simulation is progression of flood waters and depth is simulated as the model comput-ed depth and velocity as a function of time.
2.3.3 Consequence Determination
A life loss model relates characteristics of flood-ing to mortality rate, or the number of persons killed by flooding divided by the number of people present during the event (the exposed population). Fatality functions are derived based on the observations of Katrina (Jonkman 2007, Maaskant 2007; Jonkman et al., 2009). For the metro bowl, two zones of flood characteristics have been distinguished correspond-ing to two mortality functions (see fig. 5).
Figure 4 Maximum water depth of two scenarios.
Figure 5 Mortality functions derived from data for New Orleans flooding (Maaskant, 2007).
A spatially varying mortality rate for each scenario is determined. as a result The mortality rate is a charac-teristic of a location, and therefore two further as-sumptions must be made to determine fatalities: the distribution and number of population exposed to the flooding. 2010 US census data provides the dis-tribution of initial population in the form of popula-tion per census tract. The initial populapopula-tion is com-bined with an assumed evacuation rate to determine the exposed population. During hurricane Katrina, the estimated evacuation rate was about 80% (Jonkman et al., 2009; Wolshon, 2006). An evacua-tion rate of 90% is applied as it is considered evac-uation has improved since hurricane Katrina. 2.4 Results
Table 1 presents the results for the various scenarios that have been analyzed. The second column shows the failure probability estimate of the scenario (see above). The calculated mortality rate is associated with the spatial distribution of population to deter-mine an expected fatality value for each scenario. Results range from 0.4% for the breach at the IHNC, to 2.0% for the scenario of a breach due to a high river event. When the mortality is combined with an estimate of the ‘exposed population’, a loss of life
result
Table 1 Results: Loss of life estimates and assumed failure probabilities for the scenarios.
is calculated, ranging from 100 to 450. Flooding due to the river event is especially devastating, as the dura-tion of the flood wave can be up to several weeks, (compared to a hurricane event the water rises and falls more rapidly), and the metro bowl will continue to fill with water until the breach is closed. This will result in flood depths of over 5m in many parts of the Metro bowl. Given the dependence on assump-tions and uncertainties in the various models, the re-sults give an indication of the order of magnitude of consequences of the various breaches.
Other uncertainties in consequence estimates are as-sociated with the selection and modeling of flood scenarios. The sensitivity of the mortality results is investigated for various selected scenario parame-ters. In general the mortality rate is not very sensitive to breach parameters, as the final flood depths would eventually match the boundary water level in each scenario (de Bruijn, 2006). Hydrograph duration however, was seen to significantly increase mortali-ty, as maximum water depths were increased. This boundary condition could be analyzed further in an effort to provide a more accurate estimate. Overall however, flood extents, characteristics and resulting mortality rates are seen to be comparable to those that occurred as a result of Hurricane Katrina. Thus while significant improvements have been made to the protection system, all scenarios result in cata-strophic flooding of the metro bowl.
3 RISK QUANTIFICATION
Combining the probability of each event and the re-sulting consequences for the event, results in an es-timation of risk to life. The result is quantified as in-dividual risk and societal risk, metrics used to depict and limit the risk.
3.1 Individual Risk
The probability of an individual residing in a given area perishing as a consequence of flooding,
Indi-vidual risk, can be determined by the following for-mula:
IRE(x,y) = Σ Pi, FD|i(x,y)(1-Fe) (1)
Where IR(x,y)= the individual risk at location (x,y) [yr-1]; Pi = the probability of occurrence of flood scenario i [yr-1]; FD|i(x,y) = the mortality at location
(x,y) given flood scenario i [-]; Fe = evacuated frac-tion of the initial populafrac-tion.
The individual risk is a characteristic of a location, and defines the risk level spatially for a permanently present individual. To include evacuation considera-tions, the IR is multiplied by the exposed percentage of the population. Figure 5 shows the results of indi-vidual risk for the metro bowl. The bowl largely has an IR value of greater than 10-5 per year with a max-imum value of 5 x 10-5 per year. As the IR is a func-tion of the depth dependent mortality funcfunc-tion, it largely corresponds with the area topography.
Location Failure Probability Resulting Fatalities Exposed population (with 90% evacuation) Area Flooded (km2) Overall Mortality rate (%) River breach, high discharge 1/1000 450 22120 102 km2 2% River breach, storm surge 1/1000 150 18160 91 km2 0.8%
Lake, West End 1/140 170 16975 95.7 km2 0.9%
Lake, St, John 1/140 167 19980 89 km2 0.8%
IHNC 1/500 55 12662 51 km2 0.4%
Multiple Breach 1/5000 280 22118 102 km2 1.3%
3.2 Societal Risk
Societal Risk is considered to be the probability of exceedance (in a year) of a certain number of fatali-ties due to one event in a given population (Jonkman, 2007). The societal risk is a function of the population directly exposed to the flooding, thus evacuation and shelter are considered. An FN curve, a frequency –number curve, is commonly used to depict societal risk. The curve’s intersection with the y-axis is the cumulative probability of the scenarios in table 1. The studied area’s overall flooding proba-bility: 1.12 × 10-2 yr−1. The intersection with the x-axis is the consequence result of the scenario with the largest consequences, roughly 450 fatalities. The FN curve of this study is plotted against an FN curve based on data from the IPET risk study (IPET, 2009). Loss of life estimates resulting from the IPET study were used for the New Orleans metro bowl for floods with a 100 and 500 year return period and predicted 2011 population levels. As a conservative assumption, the IPET case with no pumping was plotted (as the cases with 50% and 100% pumping capacity would give lower consequence and risk es-timates). Variations in approach exist between the two studies, including the use of post-Katrina popu-lation data in the IPET approach. However, the comparison shows that both FN curves are similar. Also included in the figure is an indicative plot of the Katrina event, demonstrating the risk reduction effects of the upgraded system. It is noted that not all Katrina related fatalities occurred in New Orleans or in the flood zone. If the risks for the current situation are assessed taking into account other areas (parts of New Orleans and Louisiana) the loss of life and risk estimates for the current situation will be higher.
Probability and fatality estimates for the scenarios are combined to determine the expected number of fatalities, about 2 fatalities/year.
Figure 7 Results of this study (risk estimate for the Metro bowl) plotted against results of IPET (See IPET, 2009). The event of Hurricane Katrina is plotted.
4 EVALUATION OF FLOOD RISK WITH EXISTING RISK LIMITS
The results are indicative but realistic first estimates of risk level. To evaluate the risk, results are first compared with acceptable risk criterion found in lit-erature. Two sources of criteria include a framework developed by Vrijling (1995), and criteria proposed by recently updated Dam and Levee safety guide-lines of the US Army corps of Engineers (USACE, 2010).
4.1 Individual Risk Criterion
For individual risk criterion, both investigated crite-ria are based on the analysis of fatality statistics. First the approach of Vrijling proposes that individ-ual risk should add less than 1% to the existing probability of death (10-4 per year). The following equation is proposed to describe an appropriate indi-vidual risk level:
IR<β*10-4
(eq 2) In this expression, the policy factor β represents the level of voluntariness of the activity, with a higher level of voluntariness corresponding to a higher β value. In investigation of the fatality rate of various activities, voluntary activities correlate to a higher risk level than that of involuntary. The context of flood protection in urban areas is considered a rela-tively involuntary situation (although this is a topic for discussion). A β value of 0.1 represents this level of voluntariness, resulting in an IR limit of 10-5 per year. This framework is applied in other fields, for example in industrial safety, where a risk limit of 10 -6
per year is determined to be appropriate (Vrijling, 2011).
Figure 8 Proposed IR as proposed by updated USACE guidelines for Dam and Levee safety.
USACE guidelines define tolerable risk as the risk that society is willing to live with so as to secure cer-tain benefits. Considering existing limits currently used in industry (dam design, chemical contamina-tion and land use planning), a ‘tolerable risk’ limit of 10-4 per year was found to have achieved consensus among government and private sectors institutions engaged in safety management. While this criterion is more risk seeking than that proposed by Vrijling, an intermediate ‘ALARP’ (As Low As Reasonably Practical) zone is included in the guidelines. This zone characterizes a ‘tolerable’ and ‘acceptable’ risk level between 10-6 per year and 10-4 per year and provides for the many complex factors involved in determining acceptable risk such as cost effective-ness and societal concern. When assessing the indi-vidual risk for New Orleans, the risk is relatively low in comparison with the probability of an auto accident for example (10-4 per year), and falling within the ALARP zone of the USACE (between tolerable and acceptable risk). However, the majority of the metro bowl results in IR values higher than 10-5 per year, exceeding the acceptable level of indi-vidual risk proposed by Vrijling.
4.2 Societal risk Criterion
Criterion for societal risk is generally more con-servative than for individual risk. Society as a whole is more averse to large, multi- fatality event and such events provoke a greater social political response (Baecher, 2009). To determine if societal risk is ac-ceptable, the FN curve representing the societal risk is plotted against an acceptable ‘limit line. Again based on accident statistics, the threshold for a na-tional, socially acceptable level is proposed by Vrijling:
1 − F(n) < CN/nα (eq 3)
where F(n) = cumulative distribution function of the number of fatalities, CN = a constant that determines the vertical position of the limit line [yr− 1fat− α]. And α
= risk aversion coefficient that determines the slope of the limit line.
CN is determined by ( β *100/k)2 , again including the policy factor β and a coefficient k generally set to 3 (Vrijling, 1997). The resulting CN represents the relative size of the impacted area/population at a na-tional scale for the Netherlands. To apply the limit to a specific system, the limit scale must be adapted for the relative magnitude of the installation assessed. This concept however is not examined well in this paper, and is not elaborated on further.
The USACE guidelines provide criteria for societal risk again based upon existing acceptable risk limits applied in other sectors including the US Bureau of Reclamation in their development of their dam safe-ty program (USBRC, 2003). The USACE applies a base point of the annual probability of 0.001 per year corresponding to the loss of one life, in combination with a limit line slope of 1 for acceptable risk asso-ciated with flood protection. Considerations for an intermediate realm beween tolerable and acceptable
Figure 9 New Orleans Societal risk compared to limits proposed in literature
risk, the ALARP zone, are included. The varying risk aversion coefficient α (slope of the limit line of 1 or 2) applied in the two approaches reflects a dif-ference in reflects risk aversion toward large multi-ple fatality accidents. Discussion regarding this coef-ficient is ongoing, as an appropriate level of risk aversion varies amongst approaches.
When results of this study are plotted against pro-posed limit lines, it is seen that the societal risk for New Orleans exceeds criterion limits, most notably so when compared against criterion found in the up-dated USACE tolerable risk guidelines. Comparison can also be made with societal risk of other sectors, such as is seen figure 9. Here, failure probability and fatality data of several industries is plotted (Lambert and Associates, 1982). By doing so, insight into the relative magnitude of the risks in different domains and magnitude of risk to life of the city of New Or-leans can be gathered.
Figure 10 Results of this study plotted in orange.
5 EFFECTIVENESS OF MEASURES TO REDUCE RISK
As the risk of flooding exceeds discussed acceptable risk criterion, the effects of various risk reduction strategies can be determined. The quantitative risk estimate allows the reductive effects to be analyzed in a systematic and consistent way. Reduction measures can be categorized into measures that re-duce event probability and measures that rere-duce event consequences. Measures to reduce the proba-bility of While the risk reducing effect of load reduc-ing measures, such as barrier islands, wetland resto-ration and river diversions, is difficult to quantify and outside this research scope, it is expected that such measures play an important role in the future of
flood protection for New Orleans. These measures could reduce the (return periods) hydraulic loads on the system and thus improve the level of protection flooding include increased protection as well measures that reduce the load on the system. . The effect of increased protection however can be deter-mined through the use of the quantified risk metrics, FN curve and risk maps, and protection levels that meet the various risk criterion are in table 2.
Criterion Resulting Pf (yr-1)
Individual risk <10-5 yr-1 1/4000 Individual risk <10-6 yr-1 1/10,000 Vrijling (National) 1/25,000 FN criterion, USACE <1/50,000 FN 1982 Plot 1/10,000 Economic optimization* 1/1000-1/5000
Table 2 Protection levels that would meet proposed risk criterion (* economic optimization based on Jonkman et al., 2008)
The effects of measures to reduce the consequences are also determined. The strategies for which risk re-duction effects have been discussed include: elevat-ed homes, improvelevat-ed evacuation and relocation of population to higher areas. When the reductive effect of each measure is combined with a cost estimate for each proposed measures, the cost effectiveness of reduction measures can be assessed. The CSX value, or the cost of saving an extra statistical life, is ex-pressed to describe this cost effectiveness:
CSX = I/Δ E(N) (eq 4) Where: CSX = the cost of saving an extra statisti-cal life per year ($/Fat/Yr), I = investment ($), and E(N)= expected number of fatalities (fat/yr).
The lower the ratio between the investment and the saved lives, the more cost effective the measure is. Results of the analysis show that increasing evac-uation is the most cost effective as it limits the ex-posed population at a relatively low cost. It is noted that this evaluation focuses on reduction of risk to life. Improving evacuation would hardly reduce the economic damage that would result from widespread flooding. Comparing remaining alternatives, the 1/500 year protection level is most cost effective. The elevation of homes is second most cost effective as it reduces the depth of flooding having a signifi-cant reduction on the mortality rate. The reductive effect of the relocation of population within the Met-ro bowl is minimal due to the large flood extents of the scenarios modeled.
Table 3 Cost and resulting CSX values for risk reduction measures.
6 CONCLUDING REMARKS
An evaluation of the risk to life for the city of New Orleans has been carried out. Evaluation of re-sults show that the risk, while reduced from the pre-Katrina status, is still significant. The quantified in-dividual risk to life exceeds some proposed risk to life criterion and falls within the tolerable and ac-ceptable criteria of the USACE. The resulting socie-tal risk also exceeds criteria proposed in literature and is high compared to societal risk found in other sectors. Thus while there is not a consensus on an acceptable risk to life criteria, the risk to life for New Orleans, both individual and societal, is outside some of the risk limits proposed in literature. Inves-tigation into the quantified effects of risk reduction strategies show that evacuation is the most cost ef-fective and should thus continue to be encouraged. An increase in protection level is the next most cost effective, promoting discussion for further increase in protection levels.
The results presented have been based on only a limited number of flood scenarios. For a more com-plete evaluation, it is recommended to include a more complete set of scenarios to better represent possible load situations, breach combinations and flood conditions. As the approach has considered solely a portion of the city, application to the re-maining metropolitan areas of New Orleans provides a better overall picture of the risk. A complete relia-bility analysis should be carried out to verify event probabilities estimated in this research. Reduction measures should be analyzed in greater detail and include measures to reduce the load, such as wetland restoration and barrier islands, and a multiple lines of defense approach, combining coastal habitat res-toration and engineered flood protection (Lopez, 2007). Economic risk considerations should be in-cluded to provide a more complete picture of the risk of the city. Based on the results of this study and an-ticipating increased flood risk in the future, further discussion is recommended regarding the acceptable level of flood risk for the city is necessary.
7 REFERENCES
Aalberts, M.L., 2008. The cost effectiveness of
compartmentation of the Orleans Metro Bowl. Msc thesis, Delft University of Technology.
Asselman, N.E.M., Heynert, K., 2003, Consequences of floods: 2D hydraulic simulations for the case study area Central Hol-land, Delft Cluster report DC1-233-5
Baecher , G. , Zielinski , P. Tolerable Risk for Coastal Protec-tion, draft, The Water Collaborative, University of Mary-land.
Baecher, G., 2009, Quantifying flood Risk. ASFPM Founda-tion Symposium,
http://www.asfpmfoundation.org/forum/Quantifying_Flood_ Risk_FINAL.pdf
Boyd, Loss Of Life due to hurricane Katrina, Phd thesis. 2011 Bos, Anja,2008, Error! Reference source not found., Error!
Reference source not found., traineeship report, Royal
Haskoning.
Bruin, K.M. de (2006) Improvement of casualty functions based on data of the flooding of New Orleans in 2005 WL Delft Hydraulics Report Q3668.00, 2006.
CUR (Civieltechnisch Centrum Uitvoering research en Regelgeving) (1997) Kansen in de Civiele Techniek, Deel 1: Probabilistisch ontwerpen in theorie. CUR rapport 190
Hoss, 2010, A comprehensive assessment of Multilayered Safety (Meerlaagsveiligheid) in flood risk management Msc Thesis, Delft University of Technology.
IPET, 2009. Performance Evaluation of the New Orleans and Southeast Louisiana Hurricane Protection System, L.E. Link (editor). Volume VII, – Engineering and Operational Risk
and Reliability Analysis, see https://ipet.wes.army.mil
IPET, 2009, A General Description of Vulnerability to Flood-ing and Risk for New Orleans and Vicinity: Past, Present, and Future Supplemental Report of the Interagency Perfor-mance Evaluation Task Force
Jonkman S.N., van Gelder P.H.A.J.M., Vrijling J.K. (2003) An overview of quantitative risk measures for loss of life and economic damage, Journal of Hazardous Materials. A99, pp.1-30
Jonkman SN, Vrijling JK, Kok M. Flood risk assessment in the Netherlands: A case study for dike ring South Holland. Risk Analysis, 2008; 28(5):1357–1373.
Jonkman, S.N. (2007) Loss of life estimation in flood risk as-sessment, TU Delft PhD Thesis
Jonkman, S.N., Kok, M., Van Ledden, M., Vrijling, J.K.,( 2008). Risk based design of flood defence systems – a pre-liminary analysis for the New Orleans Metropolitan Area. 4th International Symposium on Flood Defense: Managing Flood
Measure Description Cost, $106 Source Resulting E(n) % E(n) Reduced CSX Value Elevate Homes 5ft (1.52m) water depth reduction 4000 Estimate .52 73% 2.7x109
Increased Protection 1/500 year level of protection ( Hurricane protection only )
2000 Bos (2007) 1.03 48% 1.5x109
Evacuation 95% of population evacuated 10 Estimate .01 98% 2x107
Relocated population 100% of persons in high risk zone relocated (roughly 50,000 persons)
Risk, Reliability and Vulnerability, Toronto, Ontario, Cana-da.
Jonkman S.N., Jongejan R.B., Maaskant B. (2010) The Use of Individual and Societal Risk Criteria within the Dutch Flood Safety Policy—Nationwide Estimates of Societal Risk and Policy Applications, Risk Analysis, Vol. 31 No. 2, pp. 282-300
Jonkman, S.N., Maaskant, B., Boyd, E., Levitan., (2009) Loss of Life Caused by the Flooding of New Orleans After Hurri-cane Katrina: Analysis of the Relationship Between Flood Characteristics and Mortality. Risk Analysis, Vol. 29, No. 5, 2009
Lopez, J. A. (2009) The Multiple Lines of Defense Strategy to Sustain Coastal Louisiana. Journal of Coastal Research: Spe-cial Issue 54 - Geologic and Environmental Dynamics of the Pontchartrain Basin [FitzGerald & Reed]: pp. 186 – 197. Maaskant, B.(2007), Research on the relationships between
flood characteristics and fatalities. Based on the flooding in New Orleans caused by hurricane Katrina. Msc thesis, Delft University of Technology.
USBRC (2003), United States Department of the Interior, Bu-reau of Reclamation, "Guidelines For Achieving Public Pro-tection In Dam Safety Decision-making”, 15 June 2003. Reidstra, D, Failure modes and risk Presentation presented at
the "International Flood Risk Management Approaches: From Theory to Practice", 2010, Washington, D.C, National Flood risk management program.
T.W. Lambe and Associates. (1982). “Acceptable riskat Kawa-saki site 400” Report prepared for Towa Nenryo Kogyo co. Ltd. (now, TONEN Corporation), Kawasaki, Japan , by G.B. Beachear and W.A.Marr, Cambridge.
USACE 09 Feb 2010 USACE Levee safety program and toler-able risk guidelines, Draft report:
http://www.nfrmp.us/TRG2010/docs/9_Feb_2010_Levee_Sa fety_ TRG_ Workshop_ Discussion_Paper.pdf
USACE, (2007). HSSDRS Guidelines, Elevations for design of hurricane protection levees and structures.
U.S. Census Bureau. (2011). 2010 Census of Population and Housing. Washington D.C.: US Census Bureau.
Van Dantzig, D. (1956). Economic decision problems for flood prevention. Econometrica, 24, 276–287.
Vrijling, J. K., Hengel, W. van, & Houben, R. J. (1995). A framework for risk evaluation. Journal of Hazardous Materials, 43(3), 245–261.
Vrijling, J. K.,& Van Gelder, P.H.A. J.M. (1997). Societal risk and the concept of risk aversion. In C. Guedes Soares (Ed.), Advances in safety and reliability. proceedings of ESREL 1997 Lissabon, 1 (pp. 45–52). Oxford, UK: Pergamon. Vrijling, J.K. et al. (1998) Acceptable risk as a basis for design. Vrijling, J.K., 2001. Probabilistic design of water defence
sys-tems in the Netherlands. Reliability Engineering and System Safety, 74, 337-344.
Vrijling, J.K., Schweckendiek, T., Kanning, W. (2009) Safety Standards of Flood Defenses. Keynote paper at ISGSR 2011 (International Symposium on Geotechnical Safety and Risk), Munich, June 2011. (in press)
Vrijling J.K. , van Gelder, P.H.A.J.M. An analysis of the valu-ation of a human life, Delft University of Technology, fac-ulty of Civil Engineering, P.O. Box 5048, 2600 GA Delft, The Netherlands
Wolshon B. Evacuation planning and engineering for Hurri-cane Katrina. Bridge, 2006; 36(1):27–34.
