Estimates of levee system reliability can conflict with experience and intuition. For example, a very high failure probability may be computed while no evidence of failure has been observed, or a very low failure probability when signs of failure have been detected. This conflict results in skepticism about the computed failure probabilities and an (understandable) unwillingness to make important management decisions based upon them. Bayesian networks (BNs) are useful in these circumstances because they allow us to use observations to improve our reliability estimates quantitatively.
|Title of host publication||Integral Design of Multifunctional Flood Defenses|
|Subtitle of host publication||Multidisciplinary Approaches and Examples|
|Editors||Baukje Kothuis, Matthijs Kok|
|Publisher||Delft University Publishers|
|Publication status||Published - 2017|