Background & Motivation: Coastal flood and erosion risk is a global issue influencing communities and critical infrastructure. To assess present and future coastal hazards due to rising sea level, storm surge, waves and beach level change, coastal managers are more commonly applying coastal vulnerability indices (e.g., Jevrejeva et al., 2020). These tools reply on numerical modelling of the nearshore conditions and their impact at the shoreline. The models used are often global or regional simulations of present and future conditions, with low resolution at a local (defence) scale. The initial beach condition is often based on the latest available survey or a beach slope approximation that may not be representative of present day. Validation data from observations are often sparse. A new approach to monitor coastal hazards over large spatial scales is required to capture localised process interactions and coastal responses that are unresolved by regional models. Such monitoring capability could provide early warning of hazardous changes in the natural system, for example, by identifying hotspots (e.g. Fig. 1) where beach lowering or erosion increases the flood hazard from wave run-up. Satellite observations provide a means to obtain seasonal information about the intertidal beach morphology (e.g. Fig.2) from the waterline position (Bell et al., 2016). New Machine Learning techniques that deliver coastal flood hazard warning indicators from satellite-derived information are vitally needed. However, to derive an accurate shoreline bathymetry a local water level record is required and may not always be available. Tidal models and nearby tide gauges provide the relevant information but a better understanding of the uncertainties they introduce into a flood hazard assessment is required. Uncertainty quantification will add value for others using satellite derived bathymetry in the maritime sector, e.g., ports and harbours. The PhD candidate will sit in the Channel Coastal Observatory to access a range of shoreline monitoring information and to align their research outputs with coastal managers’ requirements. They will have hands-on experience at fieldwork, data collection and processing, and have lots of opportunity to engage with coastal stakeholders.
Aims: This PhD will develop methods to assess seasonal changes in coastal flood hazard using numerical techniques and satellite observations.
Objective 1 will quantify and reduce the uncertainty in the intertidal bathymetry derivation from satellite due to distant water level records and modelled tides.
Objective 2 will apply satellite bathymetries within a coastal impact model to asses flood hazard due to wave run-up.
Objective 3 will use the new hazard data to develop efficient algorithms that predict the flood hazard from satellite products that coastal managers can use.
Methodology: A number of beach types around the UK will be studied to compare and contrast results for beaches with different gradients, exposure and sediment mixes. The use of satellite data will enable beach profiles to be obtained with minimal restrictions due to their global coverage. The European Space Agency Sentinel-1 satellite constellation have a C-band Synthetic Aperture Radar sensor that collects data day and night in all atmospheric conditions. The global repeat is 6 days or better at ~9m spatial resolution. This will be the primary source of satellite data. The Sentinel-2 constellation have multi-spectral optical sensors at 10m spatial resolution and can be used to supplement the SAR data if required. Using waterline analysis techniques the beach profiles will be extracted and applied within an XBeach model to simulate wave run-up for past wave and water level conditions (Lyddon et al., 2021). The bathymetry will be obtained using a range of methods to reference the local waterline elevation, e.g. tidal models, tide gauges, pressure sensors. Sites will be selected where there are nearby national networks that can provide coastal observation to force the offshore model boundary and assess the accuracy of the satellite derived beach profile. For the coastal environments chosen, a model database of wave run-up conditions will be generated and the impact of the sediment properties and bed roughness on the flood and erosion hazard explored. Convolutional neural networks will be used to extract bathymetric information from satellite imagery, and further applied to combine this with nearby coastal monitoring data (e.g. wave buoys, tide gauges, pressure sensors and beach profile surveys). This method will ultimately provide a flood hazard assessment at a seasonal time-scale (Bird et al., 2017). Bayesian methods will be used to assess uncertainty in the flood hazard index developed.
Bell et al. (2016), https://www.sciencedirect.com/science/article/pii/S037838391500157X
Bird et al. (2017), https://www.sciencedirect.com/science/article/pii/S0169555X16306493
Jevrejeva et al. (2020), https://doi.org/10.5194/nhess-20-2609-2020.
Lyddon et al. (2021), http://nora.nerc.ac.uk/id/eprint/530839/1/Lyddon_2021_NOC_internalreport.pdf