In recent years, considerable effort has been made in order to validate different methods that aim at estimating the wave spectra from the motions recorded on a ship or on an offshore platform. For more than ten years now, the University of São Paulo has been working on a wave inference method for moored oceanic systems, such as Floating Production Storage and Offloading (FPSO) vessels.
This paper brings the first results from an ongoing field campaign, started in December 2014, for the estimation of wave statistics by means of this system, which is based on a Bayesian inference approach. The performance of the motion-based method is checked against the wave estimations provided by a commercial marine radar system. The radar is installed in a fixed platform close to the FPSO that is being monitored, which is moored in a turret configuration. Comparison between both systems allows one not only to evaluate the performance of the method but also to evaluate the inherent limitations that exist when the estimations are based on a very large vessel, especially one whose heading may be subjected to fast variations.
In this work, a new algorithm for the partition of the wave spectrum is used, differing from the previous works. This technique allows the identification and combination of energy peaks of the directional spectrum into wave systems, providing better modal wave statistics and less noise in the final spectrum.
Previously validation of the wave inference method was made through numerical and small-scale experimental analysis. Also, in previous studies, a 9-month field campaign was also used to validate the Bayesian method using a spread-moored FPSO, but no comparison to other measurement technique was made, instead, numerical forecasts were adopted. In this work this objective is achieved: so far, 10 months of estimates from the Bayesian inference algorithm have already been compared to data supplied by a radar system.
Comparisons with the marine radar readings attest an adequate identification of mean wave directions, with the Bayesian model estimating mean wave directions that are in close agreement to those provided by the radar system most of the time. There are, however, a few cases when the discrepancy is large and the reasons for this are still under investigation.
The comparison between both sources of data provides some insightful results, which are discussed in this paper. First, they confirm and quantify the main biases of the applied method, which are related to the filtering of the energy from high-frequency waves. On the other hand, they attest that the performance of the method for more severe sea states is good overall, showing that the method is able to provide good estimates of wave height and direction and also to capture situations when wave conditions are changing fast, such as the ones that occur with the incoming of cold fronts from South.