Data-based methods for analysing single-point ocean wave measurements
Master thesis
Permanent lenke
https://hdl.handle.net/11250/3170602Utgivelsesdato
2024-11-18Metadata
Vis full innførselSamlinger
- Master theses [133]
Sammendrag
This master’s thesis aims to enhance techniques for understanding, modelling, and predicting ocean surface processes using undersampled data and incomplete information about the system’s state and physical principles.
Using in situ and remote sensing data from a recent field campaign to evaluate several traditional and two new criteria to assess the accuracy of these criteria in nearshore wave-breaking diagnostics. An integral parameter based on the temporal wave trough area and a differential parameter in terms of the maximum steepness of the crest front period are defined. The simple wave-braking detection test is based solely on a wave record and works with a single wave buoy or pressure gauge. Breaking and non-breaking waves are detected with an accuracy between 84% and 89% on the examined field data.
The second part of the study evaluates the potential of applying echo state networks (ESN) and autoregression (AR) for real-time single-point surface wave prediction for wave energy converters (WECs) under different water depths and wave conditions. This research successfully demonstrated real-time predictions for ESN and AR in water depths of up to 5 meters. It provides practical insights into the trade-offs between accuracy and practicality in real-time implementation of predictive models, crucial for optimising control algorithms in wave energy converters.
The last part of the study examines the potential of using compressed sensing (CS) to recover information from an under-sampled three-displacement buoy record. It demonstrated that it is possible to recover the wave signals in all three displacement directions with a significantly lower sampling rate than the Shannon-Nyquist sampling theorem. It is possible to reconstruct the peak frequency and direction in a 2-D wave spectrum from the recovered signals, even though the correct energy amount is not found. Still, applying compressed sensing is a valuable contribution to addressing technological difficulties such as inaccessibility of measurement locations, battery limitations, and budget constraints within surface ocean monitoring.
Beskrivelse
Revised version. Minor spelling and formatting errors corrected.