Missing values imputation in ocean buoy time series data

Absence of data or gaps in ocean wave data sequences can result in inaccurate statistical analysis as well as pose problems for forecasting purposes, which is essential for the analysis of sudden oceanic wave formations. Such analyses are critical for maritime traffic and offshore structures. In this article, missing data imputation approaches have been investigated by using field data obtained from ocean buoys. To achieve this, the wave surface elevation is first decomposed into slow varying amplitudes through the use of a wave model. Non-linear datadriven methods as well as linear methods are then used for imputation of continuous gaps of varying lengths in these time sequences. The capabilities of the different models in the gap filling tasks are then demonstrated by using different metrics for diverse wave scenarios.