Lisbon — Portugal PREDICTING THE BEHAVIOR OF BEACHES: ALTERNATIVES TO MODELS

In the USA, mathematical models are heavily relied upon in the design of beach replenishment and coastal engineering projects; yet, there is a great discrepancy between the predicted beach behavior produced by die models and the reality of actual beach behavior. Some of the problem is rooted in politics but more important is the unreality of analytical and numerical models used in the design process. The typical assumptions used to simplify the model equations are often highly questionable (e.g.. the existence of closure depth), processes are generalized and incomplete (ignoring seaward flowing bottom currents), models are assumed to apply to all beaches (a shoreface with outcropping rock is treated no differently than a shoreface covered with unconsolidated sand), and adequate real world data (e.g.. wave gauge information) is generally lacking. In addition, the model approach used in USA coastal engineering design is non-probabilistic — in effect, storms are considered to be unpredictable accidents. They are not directly accounted for in most models. The track record of model use is poor and we recommend that models should be shelved for real-world applications while recognizing their potential usefulness in basic coastal science. Taking beach replenishment as an example, there are three design approaches that could be used, avoiding the use of mathematical models: (1) determining beachfill volume requirements by measuring volume loss and shape changes of an eroding shoreline over a period of years, assuming similar behavior of replenished beach in the future (imitating nature), (2), pump sand on the beach without design predictions (the Kamikaze option) and (3), design beaches based on past experience on neighboring or regional beaches (learning from the past)