Item

Abstract

Increasing demand for coastal and offshore marine applications calls for accurate forecast of instantaneous sea level, storm surges and better understanding of marine processes. Previously one of the major limitations in accomplishing this has been and incompatible vertical reference datum and access to a method that intelligently integrates and analyzes all the different sources. With respect to linking the different data sources the use a high-resolution geoid now allows the determination of dynamic topography which represents a more accurate term for sea level. Thus this study presents a proposed methodology, that uses machine learning strategies to forecast and better understand the marine dynamics.The methodology proposed shall utilize mathematical, statistical and machine learning strategies (e.g. neural networks and inter-technique solutions) along with various relevant data sources (e.g. marine geoid, tide gauges, hydrodynamic models, satellite altimetry etc.) to forecasts sea level in the absolute sense along with their uncertainties. The results is expected to: (i) forecast dynamic topography, storm surges up to 7 day ahead, (ii) identify and predict oceanographic patterns and processes (currents, eddies, etc.), and (ii) determine realistic under keel clearance. The developed method shall be performed in the study area of the Baltic Sea that has an intense activity of shipping and maritime activities.