Views
Introduction
The general purpose of this web site is to convince the hydrological and hydraulic modelling community of the advantages of using risk and uncertainty tools and to give an initial guideline on available tools.
Risk analysis has always formed a central part of the science of hydrology and hydraulics. For example, in Frontius’ legacy (translated from Latin by Herschel, 1899), which describes the water supply system of ancient Rome, the necessity of adding additional aqueducts was justified by the Risk of water shortage. Although, the concepts of what constitutes risk may have since evolved to a more detailed level, the basic understanding is rooted deeply in our civilisation.
Uncertainty analysis, on the contrary, is a very recent development in the modelling community. Most papers in the beginning of the last century did acknowledge the existence of uncertainty, but did not deal with it in an explicit way. For instance, the first research papers on the computation of discharge in pipes, point out that there is a considerable uncertainty in computing exact values and suggest that any engineer will have to expect a range of values, which could be equally likely (King, 1918). However, the solution has been to take an average and ignore any distribution around this average. This long standing tradition has survived the last century and it is not surprising that modelling results are still very rarely presented with uncertainty bounds to decision makers or even at scientific conferences.
Risk-based decision analysis is a logical consequence of estimation of risk and uncertainty. In simple terms in the flooding context, an increase in the magnitude of an event (smaller exceedence probability) will have a greater consequence (increased flood damage. A decision about infrastructure development for Flood protection must take the risk into account as well as the infrastructure costs. As noted already, the scientific estimation of risk involves uncertainty, therefore the estimation of uncertainty should be embedded into the decision making framework. As well as being based on a comparison of risk and cost, risk-based decision-making in general includes attitude to risk in the form of utility functions. Whether or not this is formally included, some reflection on the decision-makers attitudes and preferences is essential. There are a number of methodologies for risk-based decision making (see, for example, Bedford and Cooke Bedford and Cooke, 2001).
Today there is a greater appreciation of the uncertainty of model predictions and the effect that such uncertainties might have on decision making. However, there has still been relatively little use of uncertainty analysis by hydrologic and hydraulic practitioners ( see "Ignorance is Bliss - 7 reasons not to use uncertainty analysis"). This may be because their contractor did not desire them to use such tools or perhaps that they could not see any advantage. It has been also suggested, that there may be is no trust in the credibility of the tools available and that the methodologies may be over complex (Sayers et al., 2002). An alternative reason for the lack of use of these software tools may be a lack of communications skills of the hydrological risk and uncertainty community. This report seeks to provide a convincing case for the use of risk and uncertainty tools.
Justification for risk and uncertainty analysis
Current practice in flood risk management has not taken much account of risk and uncertainty in decision making. Flood frequencies are generally estimated statistically, but even when extrapolating to the 100 year or 200 year flood from short records, the uncertainty in those estimates is rarely considered. Flood risk maps, based on those 100 year or 200 year flood discharge estimates are usually modelled deterministically, without consideration of the uncertainties associated with model implementations, choice of parameter values, or the uncertain discharge estimates. Real-time flood forecasts are still being made deterministically, despite the real uncertainties in knowledge of the rainfall inputs to a catchment and in the prediction of runoff generation (there is, as yet, little consideration of uncertainty in the National Flood forecasting system commissioned by the Environment Agency).
Public discussions, policy decisions and scientific discourses, results and decisions are frequently based on such deterministic model results. We proceed as if our models are true and every prediction can be made with certainty even though, most modellers would, after a cursory reflection, acknowledge the existence of uncertainties. This reflection does not necessarily translate into the application of uncertainty methodologies and, in setting the context for risk and uncertainty analysis, it interesting to reflect on why this should be the case.
Certainly, uncertainty analysis is an additional complication that can only be incorporated at the cost of additional expense, understanding and training but to ignore uncertainties in any form of flood risk prediction carries an associated risk for the analyst of being wrong, and does not allow the decision maker to take account of different risks of potential outcomes.
Go to
Risk and Uncertainty (Description and Definition)
-
The Concept of Risk
-
Definition of Consequences
-
Definition of Probability
-
-
The concept of uncertainty
-
-
Numerical approximation [Uncertainty of model equations
-
Definition of the Flow Domain [Uncertainty
-
Definition of Boundary Conditions [Uncertainty including forcing input data
-
-
Evaluating Model Performance and Conditioning of Uncertainties as Data are made available
-
Model Performance Measures
-
Conditioning uncertainty as data are made available
-
References
Bedford, T. and Cooke, R., 2001. Probabilistic [Risk Analysis: Foundations and Methods. Cambridge University Press, Cambridge.]?
Freer, J., Wagener, T. and Zehe, E., 2004. [Uncertainty analysis techniques in environmental modeling.]?
[ http://www.hrwallingford.co.uk/downloads/projects/FD2320/ Sayers, P.B., Gouldby, B.P., Simm, J.D., Meadowcroft, I. and Hall, J., 2002. Risk, Perfromance and Uncertainty in Flood and Coastal Defence - A Review, DEFRA/Environment Agency - Flood and Coastal Defence R&D Programme, Wallingford.
Fuzzy --Wed, 28 Feb 2007 08:18:19 +0000