Thumbnail Image

Modelling of low- and high-frequency sea level variability and their drivers around the southern African coast

Recent surveys indicate that marine operational forecasting is becoming increasingly important due the pressure to manage impacts associated with our changing climate. The methods with which forecasts are being produced are also changing as computational power is becoming more accessible. Nevertheless, understanding and improving the incorporation of oceanographic dynamics, underpinning hind- and forecasting models, will remain fundamental to the accurate prediction of physical ocean and coastal dynamics. Numerous recent studies have investigated current and possible future Southern Ocean dynamics. However, these dynamics are under studied in the continental shelf areas of southern Africa. The present study aims to address this knowledge gap and reports results of a methodical exploration of water level and wave dynamics in these waters. The structure of the thesis is based on the operational marine forecasting platform developed at the South African Weather Service (SAWS). As part of this thesis, I conceptualized and co-developed the operational platform presented here. The complete platform incorporates a coupled ocean model driven with atmospheric pressure and winds (simulated in the downscaled Unified Model (UM)). The oceanographic model consists of tidal, storm surge and wave dynamics. The coupled system was developed in a depth-averaged Delft3D FLOW model and Simulating Waves in the Nearshore (SWAN, a.k.a. Delft3d WAVE) spectral wave model. The study is divided into six chapters, each aligned with the physical description of a phenomenon. The hindcast development, assumptions, calibration, validation and operational deployment strategies are also presented. First the tidal characteristics of South Africa are fully investigated and mapped. In situ validation was performed with the main tidal constituents compared against those extracted from total water level signals observed at nine coastal measurement locations. Mapping of regional tidal characteristics was also performed for each constituent and compared with other regional tidal predictions (e.g. the TPXO 7.2 and 8 African computational nests). Coastal semi-diurnal tidal resonance is identified and quantified over the broad continental shelf areas (e.g. the Agulhas and Namaqua Banks). This model formed the bases of all the other models developed in the SAWS Wave and Storm Surge (SWaSS) operational platform. The storm surge dynamics around the southern African coastline were investigated in the next chapter. Model validation was performed at six coastal in situ measurement locations. The atmospheric dynamics for South Africa are summarised and independently validated. The UM and Wave Watch 3 (WW3) boundary forcing models were developed outside of the scope of the present study and thus only employed as forcing. The coupled, depth-averaged storm surge model was calibrated and validated. The validated model was used to quantify the various contributions of the drivers of storm surge. It was found that wave set-up contributed approximately 20% of the total surge signal in the southwest, while wind set-up contributed approximately 55%. Wave validation and sensitivity analysis was investigated next. The wave component of the SWaSS system was coupled online with the storm surge hydrodynamic model. This implied that the wave simulations responded to fluctuations in the changing water levels and the wave model contributed to the water level set-up in the hydrodynamic model. Reconstruction methods for spectral wave model boundary condition reconstruction methods were investigated and quantified. Both in situ (eight stations) and remotely sensed altimetry measurements were used for model validation. The accuracy of the SWaSS wave model component was also investigated with regards to whitecapping formulations. It was found that the Van der Westhuysen (2007) whitecapping formulation performed best together with fully spectral boundary conditions obtained from a global WW3 model. The results using the boundary reconstruction methods were performed adequate. Consideration must be given to each coastal location though as the combination of most appropriate models did vary depending on the dynamics of the mixed sea state (swell and wind seas). These studies are followed by a practical chapter investigating the computational efficiencies associated with deploying one of the most widely used spectral wave modelling software, Simulating Waves in the Nearshore (SWAN). The model extent and configuration were based on SWaSS. It was found that the most efficient SWAN simulation (for southern Africa and the Van der Westhuysen whitecapping formulation) is executed on six computational threads. Initial results and recommendations for future thorough investigations are made in the final chapter. Here the first results of using the new ST06, SWAN model physics are presented with a few permutations of the underlying parametrised physical models. The resulting accuracies, as compared with satellite altimetry measurements are also given.
Type of thesis
Rautenbach, C. (2020). Modelling of low- and high-frequency sea level variability and their drivers around the southern African coast (Thesis, Doctor of Philosophy (PhD)). The University of Waikato, Hamilton, New Zealand. Retrieved from https://hdl.handle.net/10289/14229
The University of Waikato
All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.