PROBLEMS WITH USING MATHEMATICAL MODELS TO PREDICT BEACH BEHAVIOR FOR COASTAL ENGINEERING

Coastal engineering models such as GENESIS and SBEACH have become important tools for investigating coastal processes and for coastal engineering design. They are also used as the basis for making public policy decisions such as the feasibility of nourishing beaches. However, the generalized, numerical models currently being applied to solving real-world coastal problems are grossly oversimplified, of indeterminable accuracy, and generally invalid. All of the models suffer from the following shortcomings: 1) poor assumptions and important omissions in model formulation; 2) use of relationships of questionable validity to predict coastal change; 3) lack of hindsighting and objective monitoring; 4) the incorrect use of model calibration and verification; 5) inability of the models to quantify the uncertainty involved thus producing results of unknown accuracy and unknown usefulness; and 6) poor quality data for setting model initial conditions, model input parameters, and model calibration. The result is models with inherent inability to predict coastal evolution quantitatively at the engineering and planning time and space scales relevant to society. Models are useful in addressing questions of mechanism such as how, why, and what if; but they cannot predict specifics such as where, when, and how much. Unfortunately, these are precisely the questions to which society demands answers--where will a beach erode, when will it erode (life span of a nourished beach), and at what rate will it erode (how much sand is needed and what will the cost be). Current research efforts would be better focused on detailed field monitoring and data collection, and ultimately, the development of academic, probabilistic models. In the meantime, it is likely that coastal experience (e.g., nearby replenished beach histories) is still the best way to evaluate potential engineering projects.