GPS outage recovery using wavelet and neural network models in support of multibeam hydrography

In classical hydrographic surveying, the use of GPS is limited to providing horizontal control for survey vessels. More recently, an alternative practice has evolved, which determines depth values relative to a geodetic datum and then relate them to tidal datums through a series of vertical datum transformations. Although has a number of advantages over classical hydrographic surveying, this practice requires accurate 3D positioning information. Unfortunately, accurate 3D positioning solution may not always be available as a result of communication link problems, GPS outages, or unsuccessful fixing for the ambiguity parameters. This paper examines the use of wavelet analysis to spectrally combine the GPS/INS height data series and the heave signal to bridge the height data gaps. In addition, a neural network-based model is developed to precisely predict the horizontal component of the survey vessel.