model-to-prototype time scale for bed movement cannot be computed from 

 the linear scale relations because the interrelation of the various pro- 

 totype forces affecting movement and deposition of sediments is too com- 

 plicated for accurate definition, and consequently is much too intricate 

 to permit establishment of mathematical scale relations for each compo- 

 nent of force. The model-to-prototype time scale for bed movement is 

 therefore determined empirically during the model verification; i.e., 

 the actual time required for the model to reproduce certain changes that 

 occurred in a given period of time in the prototype is used to determine 

 the model time scale for bed movement. 



d. Model Tests . The testing phase is the fruition of the efforts ex- 

 pended in the development of the model study, and is perhaps the easiest 

 of all phases to accomplish. The model has been carefully designed and 

 built based on measurements obtained from the prototype. The model has 

 performed similarly to the prototype by responding to events to which it 

 was subjected during verification in the same manner the prototype was 

 observed to respond when similar events occurred in its history. The 

 model may now be justifiably expected to respond as the prototype would 

 respond to an event or sequence of events which has not yet occurred to 

 the prototype at the particular point being investigated, for the same 

 hydrography and operating conditions. This response of the model is termed 

 the "predictive capability" of the model, since the behavior of the proto- 

 type under similar conditions can be inferred from that response. 



The predictive capability of the model is quite important, but this 

 capability has its limitations in time and in the relative magnitude of 

 the events used in the testing program. The time limitation is imposed 

 since many natural events (winds, waves, etc.) are random in nature and 

 thus impart a probablistic pattern to the behavior of the prototype in- 

 let which cannot be described or treated by the deterministic means in- 

 volved in modeling. As the duration of the model prediction period is 

 increased, this probabilistic part grows, and at a distant point in time 

 the predictive capability of the model becomes questionable. Therefore, 

 long-term shoal and scour projections with the model is undesirable un- 

 less long-term data are available for model verification. Attempts to 

 change the values of variables by large amounts may also lead to a break- 

 down of the predictive capability of the model since the variables may 

 exert a nonlinear effect on the behavior of the model inlet. For in- 

 stance, the use of very high and steep storm-type waves in a model test 

 may give misleading results unless assurance was made during verifica- 

 tion that the model responds properly to waves of this type. 



A model test series always involves at least two separate tests. 

 The first test is a "base" test, which studies the existing inlet and 

 provides a basis for comparison with later tests that have alterations 

 to the inlet. The next test or tests in the series is the "plan" test, 

 so called because the plan or plans for improving or stabilizing the 

 inlet are installed in the model and tested. The plan tests are always 

 conducted with model conditions identical to those of the base test. 

 This test procedure allows straightforward interpretation of the test 



479 



