Risk Analysis and Uncertainty in Flood Damage Reduction Studies (2000)

The National Flood Insurance Act of 1968 sought to reduce suffering and economic damage from floods. This congressional act created the National Flood Insurance Program (NFIP). Revisions to this act have included the Flood Disaster Protection Act of 1973 and the Flood Insurance Reform Act of 1994. The National Flood Insurance Program was originally placed under the authority of the secretary of Housing and Urban Development (program authority today lies with the administrator of the Federal Insurance Administration). In 1979 the Federal Insurance Administration and its programs, including the National Flood Insurance Program, were transferred to the newly created Federal Emergency Management Agency (FEMA).

Congress sought to accomplish two main goals in the National Flood Insurance Act of 1968 and the Flood Disaster Protection Act of 1973. Congress wanted property owners to purchase flood insurance to (1) provide them with financial relief should they suffer losses in a flood and (2) lessen the financial burden on federal, state, and local governments to provide grants and low-interest loans to cover the losses of uninsured property owners. These acts also sought to reduce damage from moderate-sized floods by encouraging construction of levees and other flood damage reduction structures. To achieve these goals, Congress offered incentives to communities whose flood damage reduction structures were

certified to prevent damage from a moderate-sized flood. Levee certification allows communities and structures to be removed from “Special Flood Hazard Areas” (defined as areas subject to inundation during a 100-year flood), thereby removing mandatory flood insurance purchase regulations (the 1973 Flood Disaster Protection Act requires those who are buying, building, or improving property in special flood hazard areas within NFIP communities to purchase flood insurance as a prerequisite for federal financial assistance (e.g., loan, grant, disaster assistance) when the building or personal property is the subject of or security for such assistance). Levee certification could thus exempt a community from flood insurance purchase requirements and could also remove some local land use restrictions. As certification could exempt a community from thousands, perhaps millions, of dollars of flood insurance premiums, this certification procedure (and the risk analysis therein) has great local economic and public policy significance. Unfortunately, Congress gave only vague guidance as to what size event the levees were to withstand.¹ The National Flood Insurance Act of 1968, as amended, and the Flood Disaster Protection Act of 1973, as amended, do not contain detailed statements regarding the hydrological or statistical significance of the 100-year flood, and contain but one significant reference to the 100-year flood: “Notwithstanding any other provision of law, any community that has made adequate progress, acceptable to the Director, on the construction of a flood protection system which will afford flood protection for the one-hundred year frequency flood as determined by the Director, shall be eligible for flood insurance under this chapter . . . . at premium rates not exceeding those which would be applicable under this section if such flood protection system had been completed” (42 U.S.C. 4001 et. seq; the “Director” is the Director of FEMA).

Congress was less than explicit in providing direction to the Corps concerning the desired level of flood protection for the nation. Congress directed the Corps to map 100-year floodplains as the areas that needed protection (Box 7.1 describes the adoption of the 100-year flood for

floodplain management and insurance purposes). Congress could have specified the 200- or 500-year floodplains. The committee infers that Congress desired protection against the 100-year flood. If they had wanted protection against the 200- or 500-year floods, they would have directed the Corps to map the 200- or 500-year floodplains.

Confusion crept in, however, concerning whether levees were to protect against the 100-year flood or whether they were to lower the likelihood of flooding to 1/100 per year. FEMA focused on protecting against the 100-year flood and (for certification purposes) required that 3 feet of freeboard be added to the stage (height) of the 100-year flood, to be confident that the levees could pass this flow. In the absence of risk analysis, freeboard was sensible: levees built only to the elevation of the 100-year flood would not necessarily survive it (because of waves, wind, and other uncertainties). Adding 3 feet of freeboard to levees, however, turns out to be equivalent (on average) to reducing the likelihood of flooding to roughly 1/230 per year, a much more stringent standard than Congress apparently intended.

The Corps's use of risk analysis resulted in a qualitative change in both the theory and practice of certifying levee systems. Risk analysis permits the Corps and FEMA to address problems resulting from uneven safety levels and from expenses inherent in the criterion of having the levee sized to the 100-year flood plus 3 feet of freeboard. As shown in Column 2 of Table 7.1, the 3-feet-of-freeboard criterion resulted in very different levels of protection in different communities. For example, a levee satisfying this criterion would have only a 45.3 percent chance of passing a 100-year flood in East Peoria, Illinois, but would have a 99.9 percent chance of passing a 100-year flood in West Sacramento, California; Portage, Wisconsin; and Hamburg, Iowa. When the Corps and FEMA used 3 feet of freeboard to certify a levee, they were using a criterion that provided less protection in some communities than in others. Similarly, the criterion resulted in excessive expenditures on some levees.

Higher levees may provide greater levels of flood protection, but levee heights are limited by the additional costs of higher levees and by their ecological and aesthetic effects on water bodies and floodplains. In particular, the three communities with more than a 99.9 percent chance of passing a 100-year flood (see Table 7.1) have flood damage reduction systems that, all other things being equal, are too expensive and impose high ecological and aesthetic costs. Communities whose levees have only a 45.3 percent chance of passing the 100-year flood have a much lower degree of flood protection. A community forced to upgrade its levees to the 99.9 percent level would likely find levee certification very expensive. A community that had its levees certified despite having only a 45.3 percent chance of passing the 100-year flood would be lured into complacency, as they are likely to experience a costly flood resulting in losses to individuals and in losses to the NFIP.

An additional difficulty with the 3-feet-of-freeboard certification criterion is that it could impose unnecessarily large costs on communities. Congress instructed the Corps to build water projects only if the benefits exceed the costs. In its flood damage reduction studies, the

Corps executes these instructions by estimating the heights of levees that will maximize economic benefits and are consistent with protecting the nation's environment (the National Economic Development (NED) alternative). By law, the Corps will build a levee to this height, with the federal government paying its share of the construction costs (according to the cost-sharing guidelines of the Water Resources Development Acts of 1986 and 1996). But if the community would like a levee higher than the NED levee, they must pay for all the additional construction costs. As Table 7.1 shows, the FEMA 3-feet-of-freeboard certification criterion required levee elevations greater than the NED elevation in 4 of the 11 communities in this group. If a community's levees were not certified, that community (if located in a zone subject to inundation by the 100-year flood) would remain in a Special Flood Hazard Area and would therefore be subjected to mandatory flood insurance purchase regulations, as well as relevant local land use regulations.

The 3-feet-of-freeboard concept was used as a design parameter to account for uncertainties associated with hydrologic and hydraulic analysis (Huffman and Eiker, 1991). If these uncertainties were accounted for, exceptions to the 3-feet-of-freeboard requirement were granted. One way to account for these uncertainties was to follow the Corps 's risk analysis method. As documented in Appendix B, this gave rise in 1996 to a subsequent Corps–FEMA proposal for levee certification, which combined the 3-feet-of-freeboard requirement when risk analysis was not performed with a probabilistic estimate of protection using annual exceedance probability (AEP) when risk analysis was performed. The annual exceedance probability is calculated by determining the probabilities of flooding for flows of all possible exceedance probabilities (e.g., the 10-year flood, 50-year flood, 100-year flood, 200-year flood), then integrating over them to obtain a total probability of flooding from all floods. An expected value (i.e., the mean value) of the annual exceedance probability of one percent was selected as a decision criterion: if a levee had an expected annual exceedance probability of less than 1 percent, it was certified; otherwise, it was not.

In principle, the annual exceedance probability approach to levee certification enjoyed many benefits. It allowed a broader spectrum of uncertainties to enter the analysis than did the 3-feet-of-freeboard criterion. The data in Table 7.1 show that the NED elevation is greater than the 1 percent expected annual exceedance probability elevation for all communities in this group. If the Corps thus built to the NED elevation, all areas would be certified as satisfying the 1 percent annual exceedance probability criterion. No community would have to spend tens or hun-

dreds of millions of dollars raising the elevation of their levees in order to be certified. Finally, the 1 percent expected annual exceedance probability criterion was simple to understand and to communicate.

But floodplain managers feared that the new annual exceedance probability criterion would not provide adequate protection against flooding. FEMA concluded, “for the 12 USACE projects [studied] the simulation exceedance (true) probability standard of 0.01 referenced in our April 1993 letter produced levee designs with only 0.1 to 1.5 feet of freeboard and contained the FEMA base flood with a reliability of between only 50 and 75%” (Krimm, 1996; see Appendix B). Thus, after three years, FEMA and the Corps rescinded the 1 percent expected annual exceedance probability criterion. In effect, they interpreted this criterion as something that should be satisfied with a high degree of assurance on each project, rather than just on the average over many projects.

Unfortunately, the risk analysis procedure at that time was embryonic. The annual exceedance probability itself has a range of uncertainty arising from uncertainties in the methods and data used to calculate it. By choosing the expected value of the annual exceedance probability, this range of uncertainty was not directly incorporated into the decision process. If the Corps had accounted for the various sources of uncertainty, as well as quantifying reasons for levee failure other than overtopping, the levels of freeboard and probability of containing a base flood would have been greater. In the committee 's judgement, the major reasons for the difficulties with this 1 percent expected annual exceedance probability criterion were the Corps' s method, the procedures that were followed to implement it, and the Corps's lack of experience (at that time) in using risk analysis.

In March 1997 the Corps issued a new guidance circular (USACE, 1997e) based on the estimated probability that a levee would be able to pass a 100-year flood (the conditional nonexceedance probability; the circular is reproduced in Appendix B). The Corps considered three elevations for the levee: the elevation with a 90 percent conditional nonexceedance probability, a 95 percent conditional nonexceedance probability, and the elevation based on the old criterion of 3 feet of freeboard above the 100-year flood. The Corps and FEMA agreed to certify a levee if its elevation was at least (1) at the 90 percent conditional nonex-

ceedance probability elevation, if the old criterion resulted in a conditional nonexceedance probability of less than 90 percent, (2) at the 95 percent conditional nonexceedance elevation, if the old criterion resulted in a conditional non-exceedance probability greater than 95 percent, or (3) at the old criterion's elevation, if the old criterion resulted in a conditional nonexceedance probability between 90 percent and 95 percent.

Figure 7.1 illustrates the levee certification decision tree. The FEMA level (FL) denotes the 100-year flood plus 3 feet of freeboard, and Corps Level 90 percent (CL90) and Corps Level 95 percent (CL95) denote the levels that have assurance (conditional nonexceedance probability) of 90 percent and 95 percent, respectively, of passing the 100-year flood. This mixture of the former 3-feet-of-freeboard approach and the reliability (90% or 95%) of passing the 100-year flood reflects a standard in most engineering designs—that is, a desired performance (lower limit) and avoidance of overdesign (upper limit).

This new certification criterion would require greater levee freeboard for East Peoria and other sites that formerly had lower likelihoods of passing a 100-year flood, and it would require less levee freeboard for West Sacramento and other sites that formerly had higher likelihoods of passing a 100-year flood. Like the criterion of 3 feet of freeboard, for some communities, the levee elevation required to satisfy the criterion is above the National Economic Development height. These communities would thus have to pay the entire cost of raising the levees to the certification height. In addition, the present criterion is awkward and confusing to the public.

FIGURE 7.1 Current Corps–FEMA levee certification procedure.

Data for 13 flood damage reduction projects in nine states were assembled in 1996 by the Corps's Hydrologic Engineering Center (HEC) in Davis, California (Table 7.1). No special effort was made to select particular kinds of projects or regions of the country in preparing this data set. The “base elevation ” in Column (1) is the water surface elevation above mean sea level of the median 100-year flood discharge. For each levee sizing alternative, three data values are shown: (1) the levee elevation, (2) the corresponding freeboard above the base elevation, and (3) the inverse of the expected annual exceedance probability for this elevation, which is a measure in years of the expected return period between failures of the levee. This value is sometimes called the levee's “level of protection.” To acknowledge that this quantity is a random variable with a range of variation, it is called here the “expected level of protection.”

Data for six methods of levee sizing are presented in Table 7.1 : (1) the National Economic Development (NED) plan level, (2) the 1 percent annual exceedance probability level (the 1993–1996 certification criterion), (3) the 90 percent conditional nonexceedance probability level (CL90 in Figure 7.1), (4) the FEMA standard level (base elevation plus 3 feet; FL in Figure 7.1), (5) the 95 percent conditional nonexceedance probability level (CL 95 in Figure 7.1), and (6) the current Corps–FEMA certification level combining the CL 90, FL, and CL95 values (90%-3ft-95%). Summary statistics of the freeboard and level of protection provided by each alternative are provided at the bottom of Table 7.1.

The freeboard required for levees having an expected annual exceedance probability of 1 percent averages 0.9 feet and ranges from 0.1 to 3.2 feet (Column 7). The discrepancy between these values and the traditional three feet of freeboard led the Corps and FEMA in 1996 to abandon the mean 1 percent expected annual exceedance probability criterion.

Columns (9), (12), and (15) of Table 7.1 show the three calculated levee elevations associated with the new criterion for the 13 communities. Column (9) indicates the elevation with a 90 percent likelihood of passing the 100-year flood. Column (12) indicates the 100-year flood elevation plus 3 feet of freeboard. Column (15) indicates the elevation with a 95 percent likelihood of passing the 100-year flood. In each case, bold type indicates the levee height required for certification, shown in Column (18). The combined criterion agreed to by the Corps and FEMA results in a certification level determined by 3 feet of freeboard for three communities, by a 90 percent chance of passing the 100-year flood for

five communities, and by a 95 percent chance of passing the 100-year flood for five communities. Thus, for five communities, the likelihood of the levees passing the 100-year flood is 90 percent, for three communities, it is between 90 percent and 95 percent, and for five communities, it is 95 percent. However, the seemingly close agreement among the three criteria concerning levee height masks large cost differences in modifying levees to attain these heights.

The statistics summarized at the bottom of Table 7.1 show that:

• the FEMA standard of 3 feet of freeboard provides a median expected level of protection of approximately 230 years, with a range of <100 years to >10,000 years,

• the Corps–FEMA 90%-3ft-95% criterion provides an average of 3.3 feet of freeboard and yields a median expected level of protection of approximately 250 years, with a range of 190 to 10,000 years,

• the 90 percent conditional nonexceedance probability provides an average of 3.0 feet of freeboard, and a median expected level of protection of approximately 230 years, with a range of 170 to 5,000 years, and,

• the 95 percent conditional nonexceedance probability provides an average of 4.0 feet of freeboard and a median expected level of protection of approximately 370 years, with a range of 210 to 10,000 years.

These data indicate that over the range of these 13 projects, the current Corps–FEMA criterion is slightly more conservative than the FEMA standard 3-feet criterion, and that the 95 percent conditional nonexceedance probability level is significantly more conservative. Levees designed to the 90 percent conditional nonexceedance probability level have approximately the same median level of safety as the traditional standard of 3 feet of freeboard.²

The range in the freeboard required for the Corps–FEMA combined criterion (90%-3ft-95%) compared to the FEMA standard 3-feet criterion

is illustrated in Figure 7.2. It can be seen that the current Corps-FEMA combined criterion requires freeboard ranging from 1.6 feet in Hamburg, Iowa, to 5.8 feet in Sny Island Levee Drainage District (LDD), Illinois. For most projects on the right of the figure, the 90 percent conditional nonexceedance probability level defines the certified elevation. On the left of the figure, there are several projects where either the 3-feet or 95 percent criterion prevails, adding approximately 0.6 feet of freeboard to the 90 percent level. It is clear from Figure 7.2 that except for American River, California, the range in required freeboard between the 90 percent and 95 percent values is small when compared to the variation in these elevations from project to project. To reiterate, however, “small” differences in levee height can translate into large differences in the cost of the levees.

Figure 7.2 is an impressive demonstration of the value of risk analysis in assessing the required height of levees. It shows that to meet a consistent national standard of levee safety, as little as 2 feet of freeboard may be required in one location, while as much as 6 feet may be required in another. This raises serious questions regarding the degree of safety being provided by the current FEMA standard freeboard of 3 feet required for non-Corps projects. It shows that the traditional 3 feet of freeboard masks a significant degree of variation of risk of failure in levees built to this standard for the citizens protected by these levees. This variation in risk of failure can be quantified by the Corps's risk analysis procedure.

Figure 7.3 compares the elevation required for the National Economic Development plan with that for the three levee certification criteria for 11 of the 13 levee projects in Table 7.1 for which an National Economic Development plan elevation exists. On average, the National Economic Development plan provides approximately 0.5 feet of additional freeboard beyond the Corps-FEMA 90%-3ft-95% criterion, but in 4 of the 11 projects (Portage, Wisconsin; Pender, Nebraska; Guadalupe River, Texas; White River, Indiana), the National Economic Development plan would not provide sufficient elevation for the levee to be certifiable. In these cases, the Corps's local partners would be required to pay the entire cost of raising the levee to the certifiable level. Few communities would likely be able or willing to do this. As a result, the levees will not be certified, part or all of the community will remain in the Special Flood Hazard Area (SFHA) and will thus be subject to mandatory flood insurance purchase guidelines—and to generally higher flood insurance premiums than communities located outside of SFHAs (loca-

FIGURE 7.2 Current Levee Certification Criterion (90%-3ft-95%) and the FEMA Standard Three Feet.

tion within an SFHA may also subject a community to more stringent land use regulations). With higher premiums, fewer people are likely to purchase flood insurance and, when the inevitable flood occurs, they may face financial ruin. The federal government will, consequently, have little choice but to offer grants and low-interest loans, which is precisely what Congress wanted to eliminate by establishing the National Flood Insurance Program.

There are two consequences of using an annual exceedance probability of 1/230 rather than 1/100 as the certification criterion. The first has to do with whether the flood insurance premiums are actuarially fair. If the premiums are based on floods with a probability of 1/100 and the actual probability is 1/230, the premiums are far too high. These premiums would thus discourage people from buying flood insurance, which is the opposite of what Congress desired. The second is that Congress gave benefits to communities whose levees are certified, benefits such as lower insurance rates and an exemption from having to buy flood

FIGURE 7.3 Comparison of freeboard requirements for the National Economic Development plan elevation and those arising from the levee certification criteria.

insurance. If the certification criterion is more stringent than Congress desired, some communities will have to pay higher insurance rates and will have to buy flood insurance that they need not have purchased.

The data for the 13 flood damage reduction projects in Table 7.1 are plotted in Figure 7.4 to illustrate the relationship between levee freeboard and the expected level of protection for a number of levee sizing criteria. The expected level of protection is the inverse of the expected value of the annual exceedance probability. As anticipated, the expected level of protection increases with the amount of freeboard required. However, the data for the Portage, Wisconsin, flood damage reduction project in the dashed box on the lower right of the diagram represent an anomaly. These values indicate that the expected level of protection provided by about 2 feet of freeboard at Portage is much higher at this location than in the other 12 projects surveyed. This anomaly may be a result of some peculiarity in the topography at Portage, or the risk analysis for this project may have been done differently for this project than for the other projects.

Part of the stimulus for this committee's establishment arose from a conflict among the Corps and state and local interests in Wisconsin con-

FIGURE 7.4 The expected level of protection provided by levee freeboard.

cerning the appropriate levee height for the Portage project. Figure 7.4 suggests that the Portage project is not representative of the general relationship between freeboard and expected level of protection provided by Corps projects. According to the Corps's risk analysis, requiring 3 feet of freeboard on the levee at Portage led to an abnormally expensive project. Assuming that the Corps's risk analysis was correct, the Corps was correct in resisting so expensive a project that went beyond the levee called for in the National Economic Development alternative.

All Corps flood damage reduction studies prepared by Corps offices across the country are reviewed at Corps Headquarters in Washington, D.C. As part of that review process, the committee recommends that the Corps maintain an inventory from past projects of the amount of freeboard provided for the base flood and of the resulting expected level of protection provided by project levees.

To summarize, the former 3-feet-of-freeboard criterion for certifying levees (1) provided quite different levels of flood protection to different communities, (2) was unnecessarily expensive in some communities, and (3) was above the National Economic Development elevation in some communities, requiring these communities to pay for all the costs of the

additional levee elevation. The new certification criterion is an improvement, as it results in a narrower range of protection (90% to 95% chance of passing a 100-year flood, rather than 45% to 99.9%). The new criterion, however, does not solve the problems associated with the former 3-feet-of-freeboard criterion.

One possible certification criterion based on risk analysis is the use of a conditional nonexceedance probability for the 100-year flood that must be at least some appropriate value, such as 0.90. This criterion focuses on the best estimate of the 100-year flood and can be used in both unprotected areas (delineation of the floodplain with some extra safety factor) and areas protected by levees. However, this measure is unnecessarily complicated and difficult to understand.

In summary, and based upon this review, the committee recommends that the Corps and FEMA promptly develop an improved levee certification procedure. The committee recommends a roughly equivalent criterion be used: the annual exceedance probability (the probability that the area will be flooded during a year). The committee recognizes that considerable resources and time will be required to modify and implement this recommendation. In this interim period, the committee therefore recommends that the Corps and FEMA adopt a single conditional nonexceedance probability for use in their joint levee certification program.

A sample of 13 projects is of limited size for drawing general conclusions about the equivalence of risk measures among different levee sizing criteria. The committee therefore also recommends that the Corps and FEMA compile comparable data to those in Table 7.1 for a significantly larger number of levee projects, with a view to replacing the current 90%-3ft-95%.levee certification rule with a rule based on the annual probability of flooding.

The current Corps–FEMA levee certification procedure is an important step toward implementing a rational approach to assessing the degree of flood protection provided by a levee system. The procedure accounts for uncertainty in assessing the frequency and severity of future flooding, and it estimates the probability that the levee system will perform as intended. The committee recognizes the difficulties the Corps

and FEMA face in attempting to establish a rational, workable procedure for certifying levee systems. The Corps and FEMA are to be commended for deciding to directly face these challenges.

In reviewing the current certification procedure, the committee identified four technical issues that should be addressed in the near term to improve the risk analysis method. First, the procedure does not use straightforward probabilistic measures, such as annual probabilities of flooding, that are easily interpreted and easily compared to program mandates. The procedure is difficult for the public, and for the informed technical community, to understand and to communicate.

Second, the current certification procedure examines the base (100-year) flood alone and not the range of floods that can occur. For example, the process does not consider the potential that the levee system could fail due to wind and waves leading to overtopping, embankment failure, or other related factors during the 90-year event. Wind, waves, debris, and ice could lead to the 90-year flood overtopping the levee, or absence of them could lead to the 110-year flood not overtopping the levee. The current procedure thus gives a probability of flooding, assuming that the 100-year flood occurs. A more relevant measure would be the probability of flooding, given the distribution of floods of all sizes.

Third, evaluation of the uncertainties in levee performance does not comprehensively address sources of knowledge uncertainty in the geotechnical evaluation of levee system reliability, river hydraulics, and foundation or embankment failure; thus, an arbitrary level of reliability is being evaluated as part of the certification process.

Fourth, the current procedure focuses on the portion of each levee that is most likely to fail, which may not provide a sufficient analysis of the performance of levees as a system. Assessment of the levee system should account for the potential for failure at any point along the levee during a flood event, considering multiple modes of levee failure—including overtopping, piping, embankment instability, foundation instability, and other geotechnical considerations. These technical corrections do not require a review of policy implications or cost. They could therefore be implemented immediately.

Beyond the improvements to the current procedure suggested above, the Corps and FEMA should work toward an alternative criterion based

on a simple “annual probability of levee system failure (annual exceedance probability) ” accounting for both natural variation and knowledge uncertainty and accounting for threats to the levee system other than just overtopping. Levee system failure is equated here with the failure to prevent inundation of any area that should be protected (i.e., the expected level of protection). Such a criterion would have direct benefits compared to the current criterion: (1) there would be a direct measure of the level of protection, (2) the criterion would be focused on protection from all floods, not just on the 100-year flood, and (3) the criterion would be easier for the public, Congress, and the engineering community to understand.

The committee believes that earlier difficulties encountered with an annual-probability-of-flooding approach can now be overcome because of the Corps's greater experience with risk analysis and because the Corps can now consider target annual probabilities in light of historical practice. The criterion for certifying a levee should be that it provides adequate protection against failure of the flood damage reduction system. Because the old criterion for levee certification produced different levels of flood protection for different communities, there is no reason to replicate its levels of protection in the new criterion.

The former certification criterion of 3 feet of freeboard provided an expected annual exceedance probability of roughly 0.00435 (return period of 230 years), while the 90 percent nonexceedance probability levee provided the same expected annual exceedance probability of 0.00435. The committee recommends that levees provide a uniform level of protection across communities. It is not obvious which level of protection should be chosen. In the committee's judgment, the criterion should aim to provide the level of protection provided to the most people in the past—the median level historically provided. Based on a small sample of Corps flood damage reduction projects, that median annual exceedance probability is roughly 1/230.

The committee also recommends that the Corps develop a table showing percentiles of variability in the annual exceedance probability or showing levels of protection similar to that used for the measures of economic performance in project planning. By choosing an appropriate percentile value in this range, a corresponding level of assurance can be obtained that the expected level of protection is at least 100 years, as required. It was the lack of allowance for this variability that led to the abandonment of the annual exceedance probability criterion during the 1990s.

A difficulty with establishing a criterion based on an expected annual

exceedance probability of 1/230 (0.00435) is that the resulting elevation often exceeds the elevation of the National Economic Development (NED) alternative levee. Few if any of these communities would elect to spend tens or hundreds of millions of dollars to raise their levees above the NED elevation to become certified. However, the judgment regarding the level at which to certify levees is a political, not a technical judgment.

The alternative criterion changes the focus from the 100-year flood to the expected probability of flooding. Although the committee judges the latter to be what Congress intended in the National Flood Insurance Act of 1968 and its later amendments, we recognize that it has not been the criterion used in federal regulations. It is therefore recognized that review and discussion will likely be required before implementing the alternative criterion. Although shifting the focus to the annual probability of flooding is the more desirable alternative, the committee does not want to delay implementation of the technical corrections. All of these corrections are required for implementing the alternative criterion. Thus, nothing is wasted by immediate implementation of the corrections. At the same time, the committee recommends the alternative criterion, recognizing that its implementation will require examination and discussion.