W. Huang et al. / Ocean Engineering 30 (2003) 22752295
2277
important to both commercial and recreational navigation. Sediment deposition in
the inlets requires costly dredging to maintain sufficient navigation depth. According
to Grosskopf et al. (2000) and Rosati et al. (1999), representative average annual
dredging requirements, in cubic meters, for the inlets are as: Shinnecock--100,000;
Fire Island--500,000; and Jones--100,000. US Army Corps of Engineers' New York
District and the State of New York have as an objective regional sediment manage-
ment program for the South Shore. This program effectively integrates operation cost
by linking dredging, sand bypassing, breach-contingency plans, and protection of
beaches vulnerable to storm erosion. Monitoring and prediction of long-term water
level variations in the coastal inlets are important elements in this long-term regional
sediment management program. For example, historic water levels are needed in the
analysis of aerial photos to determine coastal line changes, and in the boundary
conditions for coastal hydrodynamic models. Since 1999, several water level stations
have been established in Shinnecock Inlet, Moriches Coast Guard, and Fire Island
Inlet. However, long-term historic water level data are often not available in the
inlets along the South Shore except those at stations maintained by the US National
Ocean and Atmospheric (NOAA). About 60100 km away from the inlets, are two
NOAA permanent water level stations that have been operating since the 1940s, one
located at Montauk at Long Island and another at the Battery in lower Manhattan.
Water level data from the Montauk and the Battery stations starting from the 1940s
have been processed and verified by NOAA and are available online from the NOAA
web site (http://co-ops.nos.noaa.gov/coastline.shtml?region=ny). If the relationship
of water levels between NOAA stations and local stations in coastal region can be
established, valuable historical water level data at NOAA stations can be sup-
plemented for the study of long-term circulation and sediment transport in Long-
Island South Shore. In addition, the verified data at NOAA stations can be used for
the prediction of water level data at the temporary stations at the coastal inlets in
future operations so that expensive measurement instruments can be relocated to
other study areas.
Artificial Neural Networks (ANN) have been widely used in multivariate nonlinear
time series modeling in the many research areas such as electronics, aerospace, and
manufacturing engineering. ANN is capable of directly correlating the time series
of multiple input variables to the output variables through the interconnected nodes
using trainable weights and bias signal (Hagan et al., 1995). In this study, our objec-
tive was to develop a Regional Neural Network for Water Level (RNN--WL) model
to correlate the time series of water levels at a local station with water levels at a
remote NOAA station in the coastal region. The model was firstly tested in the
Southern Shore of Long Island using input data from NOAA stations at the Montauk
and the Battery. Then, the model was further tested using input data from other
remote water level stations located 60--500 km away. Because the RNN--WL
model requires only the time series inputs and outputs, it can also be applied to other
coastal areas in US. Therefore, the RNN--WL model developed in this study can
benefit researchers and engineers in the coastal engineering community.
Previous Page
