IAP-26-038

How confident can we be in coastal change? Uncertainty-aware satellite monitoring of Scotland’s diverse coasts to support coastal adaptation planning.

Background
Scotland’s coast is one of the most physically diverse in Europe, and it is changing faster than before. Around one fifth of the coastline is soft or erodible, and since the 1970s the extent of erosion on that coast has grown by 39%, erosion rates have doubled and the extent of accretion has fallen by 22% (Hansom et al., 2017). The second national assessment found erosion affecting about 46% of the soft coast, with around £1.2 billion of assets at risk by 2050 under high emissions and around £814 million even under low emissions (Rennie et al., 2021). Local authorities are now preparing Coastal Change Adaptation Plans, and these decisions rest on knowing not only how fast a coast is changing, but how confident we can be in that rate.
The research gap
Open-source tools such as CoastSat (Vos et al., 2019) and VedgeSat (Muir et al., 2024) now make it possible to track shoreline and vegetation-edge change across whole coastlines from the Landsat and Sentinel-2 archives. These tools were developed and validated mainly on relatively uniform sandy coasts, yet Scotland’s coast includes hard-rock and soft cliffs, salt marshes, shore platforms, machair, and sand, shingle and mixed beaches. In these settings the tracked indicator (cliff top, marsh edge, vegetation line) is harder to define and detect, while tides, wave run-up, cloud, shadow and seasonal vegetation change add noise to each detection. These errors pass directly into change rates, yet change rates derived with these tools are usually reported without uncertainty. We therefore do not know systematically how well existing methods perform in each coastal environment, how much confidence their rates deserve, or whether reporting that confidence would change the decisions coastal managers make.
Aims
This PhD will produce the first environment-specific, uncertainty-aware record of satellite-derived coastal change for Scotland, explain why change differs between coastal types, and test how uncertainty information alters coastal adaptation decisions. It addresses four objectives: (1) benchmark satellite methods across coastal environments; (2) reduce and quantify uncertainty in change rates; (3) explain the controls on coastal change using the confidence-weighted record; and (4) evaluate, with coastal managers, how uncertainty-aware change rates affect adaptation planning.

The project combines geomorphology, remote sensing, statistics and applied coastal management, giving the student an interdisciplinary training with a clear route to impact.
Objective 1: Environment-aware benchmarking. A labelled validation dataset will be built for 4–6 representative sites covering Scotland’s main coastal types (for example, cliffed, salt-marsh, shore-platform, sandy and mixed-beach coasts, including the long-studied coast near Aberdeen). Ground truth will come from Dynamic Coast shoreline records, airborne LiDAR, aerial imagery and repeat RTK-GNSS and LiDAR drone surveys. VedgeSat and CoastSat will be benchmarked against this dataset to measure where, and why, detection fails in each environment.
Objective 2: Reducing and quantifying uncertainty. The project will also check whether seasonal composites, built from stacks of images within a season, reduce noise compared with single-image detections, and whether that benefit differs between coastal types. Positional errors measured in Objective 1 will be propagated into change rates, so that every rate in a multi-decadal record carries a confidence interval.
Objective 3: Explaining coastal change. The confidence-weighted record will be related to wave exposure, storminess, relative sea-level change, geology and sediment supply using explainable statistical models, to identify why some coasts erode faster than others and where apparent trends are within the noise.
Objective 4: From uncertain rates to adaptation decisions (management chapter). Working with the CASE partner and local authorities preparing Coastal Change Adaptation Plans, erosion-risk zones will be mapped with and without uncertainty bounds at the study sites. Structured workshops and interviews with coastal managers will test how confidence information changes the areas they prioritise and the actions they choose. Findings will be co-produced into practical guidance on reporting confidence in coastal change for adaptation planning.

Year 1

Literature review and training and uncertainty methods (M1–3); site selection and survey design (M4–6); first field surveys and assembly of Dynamic Coast, LiDAR and aerial imagery ground truth (M6–10); initial benchmarking of VedgeSat and CoastSat (M10–12). Attendance at the British Society for Geomorphology postgraduate workshop.

Year 2

Completing of benchmarking across all environments and paper 1 (Obj 1) (M13–18); seasonal-composite experiments and uncertainty propagation to build the confidence-weighted record (Obj 2) (M16–24); repeat field surveys (M18–20); ethics approval and planning of manager workshops with the CASE partner (Obj 4) (M20–24). Presentation at the BSG Annual Conference.

Year 3

Analysis of controls on coastal change (Obj 3) and paper 2 (M25–32); workshops and interviews with coastal managers (Obj 4) (M28–34); placement with the CASE partner. Presentation at an international conference (e.g. EGU General Assembly).

Year 3.5

Paper 3 and co-produced guidance on reporting confidence in coastal change (Obj 4); then writing and submitting thesis (M37–42).

The student will be trained by the supervisory team to become an independent researcher in coastal geomorphology and remote sensing. Specific training will cover field surveying (RTK-GNSS, drone photogrammetry), satellite image processing in Python and Google Earth Engine, uncertainty quantification and statistical modelling, and the qualitative methods and research ethics needed for the management chapter. This will be complemented by Iapetus training in scientific and transferable skills (writing, presentation, data management and science communication). Through interviews with practitioners, the student will learn how coastal adaptation decisions are made in practice and how to co-produce outputs with end users, drawing on the lead supervisor’s extensive knowledge-exchange experience.