Author(s): Rao Raghuraj; Vladan Babovic
Linked Author(s): Vladan Babovic
Keywords: Hydroinformatics; Sea level anomaly; Genetic programming
Abstract: It is often observed in the shallow water regions around Singapore that the water levels and currents deviate significantly from their regular tidal behavior. Such deviations seriously affect the coastal planning and navigation activities. Hence, accurate prediction of these sea level anomalies (SLAs) becomes essential during oceanographic modeling. The tide predicting models or the numerical models tuned for the regional tidal variation are generally unable to capture the local meteorological driven non-periodic anomalies. Complete understanding of SLAs and their prediction is vital for completeness of any real time sea level forecasting system. In this paper we undertake such an exercise to understand the SLAs in the Singapore region and establish models to generalize their spatial and temporal characteristics. Genetic Programming (GP) based data driven models are implemented as an alternate SLA forecast tool. Ten years of observed SLA data is used to build the GP models describing the temporal patterns at six locations in the region. SLA forecast models are then tested as data assimilation tools for correcting the tide predictions during the real time sea surface forecasting. The results provide a clear understanding of SLA dynamics in the region and establish the utility of GP based SLA forecast models as data assimilation tools.