An evaluation of conditional spatial predictions of significant wave height based on the nonstationary spde model

• Main Authors: Hildeman, Anders, Mao, Wengang
• Language: unknown
• Published: 2020
• Subjects: Probability Theory and Statistics; Gaussian random field; Significant wave height; Stochastic partial differential equation; North Atlantic
• Online Access: https://research.chalmers.se/en/publication/39ff3927-7e17-4f05-8fb6-8bf6343d0ae1

The sea state plays an important role in offshore-and marine operations. It affects both direct costs as well as risks for human and/or material loss. A better understanding of the present-, near-future-, and far-future sea states will increase efficiency and safety in shipping since it allow a ship to reroute to a safer and/or more cost effective route. In the offshore industry it allows for minimizing downtime and aids in planning the construction of new offshore sites. Due to the complex nature of the sea state, its spatial distribution over a large region of ocean should be modeled using a probabilistic model. In this way, uncertainties due to lack of information and/or computing power can be quantified and decisions can be taken based on both what is known and what is not known. We analyze such a spatial probabilistic model in order to assess its ability to predict the significant wave height in the whole north Atlantic based only on measurements on a small line path, i.e., conditional prediction. This work is relevant for several applications, for instance data assimilation, oceanographic forecasting, and routing of ships.