Implementation of Storm
The process required for application of a long-wave hydrodynamic numeri-
cal model at a particular site includes grid generation, model calibration, model
validation, and production runs. Accuracy of model results is greatly influenced
by the accuracy of boundary and forcing conditions, representation of the
geometry of the study area (i.e., bottom topography and land/water interface),
and to a lesser degree, the values of certain calibration parameters. Model
calibration involves adjustment of the calibration parameters to maximize
agreement between model results and measurements.
Upon completion of calibration, the model is subject to a validation that
consists of applying the model over a different segment of time from that of the
calibration and where no changes have been made to the calibration parameters.
The model is considered validated if results agree with measurements within an
acceptable degree of error. The validation procedure provides confidence that
the model can accurately simulate hydrodynamic processes in the study area.
The strategy for calibrating and validating the storm surge model requires
that the model accurately simulate both tidal propagation and storm surge in the
study area. The model was first tested for simulation of tidal motion, then tested
with the typhoon wind and pressure fields. Procedures applied in conducting
model testing and the results of these tests are presented in the following
Calibration of Storm Surge Model
The model was calibrated for tidal propagation and subsequently verified for
storms. Water-level measurements were available from one gauge within the
study area located in Apra Harbor at the entrance to Sumay Cove. The gauge is
operated by the U.S. National Ocean Service (NOS). The gauge location is
13.4435 N, 144.6566 E. Tide gauge data for the month of January 1997 illustrate
tidal behavior at Apra Harbor (Figure 10). Mean tide range is only 0.5 m (1.7 ft).
The tide is characterized by a large diurnal inequality. High tides are generally
relatively close to msl and lower low tides typically drop well below msl. The
probability distribution of hourly tide levels during the year 1991, a relatively
Implementation of Storm Surge Model