Risk Analysis and Uncertainty in Flood Damage Reduction Studies (2000) / Chapter Skim
Currently Skimming:
Evaluation and Proposed Improvements
Evaluation and Proposed Improvements
Pages 114-138
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 114... ...
This chapter recommends improvements the Corps should make in its risk analysis methods for flood damage reduction studies to overcome the shortcomings identified in previous chapters. This chapter addresses three broad issues: (1)
|
From page 115... ...
The current framework does not span the full range of uncertainties important to flood risk and flood damage reduction, it does not clearly differentiate between natural variability and knowledge uncertainty or the relative roles of these uncertainties, and it does not recognize the importance of spatial structure and correlation among uncertainties. A clear a priori articulation of the goals of the risk analysis approach would help illuminate the needed structure of the analysis and help identify conceptual gaps.
|
From page 116... ...
. nglneerlng Performance Economic Performance Natural Variability Performance Measure Range of flood frequency Annual exceedance probability, Conditional nonexceedance probability Uncertainty of the No specification Performance Measure Range of flood frequency Expected annual damage, Expected annual benefit Percentile values of annual damage and benefit Variance of expected annual damage, and of expected annual benefit benefits of the project plan.
|
From page 117... ...
Such a distribution computed for an arbitrary location in the Beargrass Creek study is shown in Figure 6.1. The vertical axis is the annual exceedance probability of the target stage at a damage reach.
|
From page 118... ...
The committee recommends that the Corps standardize the annual exceedance probability as its principal engineering performance measure for decision making in flood damage reduction studies. The range of variation of this measure resulting from knowledge uncertainty should be specified by a table of percentile values in the same way as is currently
|
From page 119... ...
This allows consistent comparisons between risk analysis studies and the historical approach. Additional uncertainty enters the calculation of flood damage due to knowledge uncertainty regarding the choice of mode]
|
From page 120... ...
120 Risk Analysis and Uncertainty in Flood Damage Reduction Studies uncertainty of all three parameters of the distnbution. In the committee's judgment, use of the LP3 distnbution is reasonable, but a procedure should be developed that more adequately captures the true posterior distnbution of the LP3 distribution parameters in a Bayesian sense (Bobee and Ashkar, 1991; Chow~hury and Stedinger, 1991; Stedinger, 1 983a)
|
From page 121... ...
Errors in Flood Frequency Curves Derived from Rainfall-Runoff Modeling Although classical flood-frequency analysis is based on analysis of observed discharge data, practical requirements in Corps flood damage reduction project planning often lead to the use of rainfall-runoff modeling to synthesize the flood-frequency curves used in planning studies. In its risk analysis, the Corps treats these synthetic flood-frequency curves as being of equivalent accuracy to flood-frequency curves derived from a graphical fit to a set of observed flood discharge data.
|
From page 122... ...
GEOTECHNICAL RELIABILITY Geotechnical reliability is an important consideration in flood damage reduction projects involving levees. The reliability computation accounts for the potential of the levee to breach through soil failure even when the water surface elevation is not sufficiently high to overtop the levee.
|
From page 123... ...
Levee failures caused by seepage under or through the levee and even slope instabilities, which are influenced by pore pressures internal to the levee~epend not just on water height, but also on the duration of flooding. Flood duration is not considered in the current geotechnical reliability model, and indeed may be difficult to accommodate in the risk analysis method at all.
|
From page 124... ...
The three main scales are the following: ried out. · project scale at which all the economic analysis is summarized, · damage reach scale used for most analysis in HEC-FDA, and · structure scale where the assessment of damage to structures is carThe single project is subdivided into 21 damage reaches, which contain 2051 structures.
|
From page 125... ...
The Engineering Regulation that guides the evaluation process for flood damage reduction projects (USAGE, 1996b) calls for a table showing the probability distnbution of the reduction in expected annual damage due to the project plan, with percentiles at PRO, 25%, 50%, 75%, 95°/O, and it calls for a similar table for the percentile distribution of project plan net benefits.
|
From page 127... ...
These issues are not addressed in the current implementation of HEC-FDA. It appears that apart from treating the interdependence and the interaction among the three main variables at each damage reach, the current version of the risk analysis procedure assumes that all the random variables are statistically independent.
|
From page 128... ...
, but correlation length sufficiently describes the degree of spatial correlation for purposes of this discussion. In the Beargrass Creek study, for example, the first-floor elevations of the structures were estimated from 1-inch-to-100-feet topographic mapping, using a contour interval of 2 feet.
|
From page 129... ...
This figure shows water surface profiles for four different flood severities, with the locations of the structures superimposed. The river reach of approximately 12 miles in length is divided into 15 damage reaches for statistical analysis, averaging 0.8 miles in length.
|
From page 130... ...
Similarly, the laws of hydraulics impose a significant degree of continuity of water surface profiles through damage reaches. The backwater effect of the bridge constriction at the upper end of damage reach SF-1 propagates some distance upstream, perhaps through damage reaches SF2 and SF-3 but probably not any farther.
|
From page 131... ...
131 .` 2 ~— . 9-L-3S bL -- C o pe UhAO7SpJe~ m .aouanl~uoo }e Ia43an~ ~n .
|
From page 132... ...
Indeed, the Monte CarIo simulations for each reach are based on the assumption that the reaches are independent. There is thus an internal contradiction in the method to determine variability in project reduced damage and net benefits, the summation made over values at the damage reaches relies on the assumption that reach damages are perfectly correlated from reach to reach; however, the computational procedure used to create the damage values treats each reach completely independently of its neighboring reaches upstream and downstream.
|
From page 133... ...
Unfortunately, the rigors of treatment of combinations of random variables allow only a limited number of standard operations, far fewer than if the uncertainty in the variables is ignored. In effect, the current risk analysis procedure evolved from an earlier deterministic procedure by randomizing uncertain variables, solving equations for each randomization, and then averaging over the results to get expected values.
|
From page 134... ...
Similarly, there is a Tong history of generating correlated random variables to describe stream flow in a basin or region (Sales, 19931. Randomize Structures Jointly The damage reach mainly functions as a place of aggregation of structures to an index location, and this should be considered in the context of other structures.
|
From page 135... ...
For each project plan and cycle of randomization, the issue of prime concern is the difference between the damage with the plan compared to the damage without it. That calculation is presently conducted by determining the flood damage for all structures in a damage reach and then aggregating over the damage reaches to get the expected annual damage for the project.
|
From page 136... ...
dX o (6.4) If the project benefit is determined for a set of floods, p, then the expected annual benefit (EAB)
|
From page 137... ...
By summing across the whole project the damage reduced at each structure, Monte Cario replicates for benefits for the whole project can be found, and their mean and percentile distribution can be determined once the Monte Cario simulations are complete. This is a valid statistical procedure not confounded by the problem of aggregating probability `distributions from each damage reach as is now done in HEC-FDA.
|
From page 138... ...
The committee recommends that the computational procedure for risk analysis of economic performance focus on the uncertainty in the project benefits rather than on the uncertainty in the project flood damage. Statistically Compare Net Benefits from Alternative Plans If an accurate estimate of the variance of the damage reduced by a project plan can be found, it could be used to determine whether one project plan has expected annual net benefits that are better than those of another plan in a statistical sense.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.