A multi-model ensemble to investigate uncertainty in the estimation of wave-driven longshore sediment transport patterns along a non-straight coastline

Although there have been many efforts in the literature to hindcast the patterns of longshore sediment transport (LST), they mainly disregarded uncertainty issues. Forcing datasets, wave transformation methods, and LST models are among the main sources of uncertainty in LST estimations. The combination of the aforementioned sources of uncertainty makes the estimation of LST patterns challenging for non-straight coastlines as the uncertainty ranges might vary from site to site. In this paper, a simple ensemble modeling framework was employed to investigate LST rate uncertainty at seven sites along a non-straight coastline (Gold Coast, Australia). The ensemble was formed by two different forcing datasets (i.e., integral parameters of total wave energy from two hindcast datasets of ERA5 and CAWCR), two different wave transformation methods (i.e., inclusion and exclusion of local wind effects), and eight LST models (i.e., bulk formulae and process-based models). Moreover, the relative importance of each source of uncertainty was ranked using the ANOVA-variance-based model. Finally, the weighted ensemble mean was used to investigate intra- and inter-annual variability of LST rates. The results showed that the range of uncertainty of LST rates for open coasts of Gold Coast is much higher than that of semi-sheltered coasts. On annual scale, for open coasts, 40% to 50% of total uncertainty was due to the choice of wave transformation methods, while for semi-sheltered coasts, it was 20%–30%. Moreover, almost for all sites, 30% to 50% of total uncertainty was controlled by the choice of LST models and the interaction of wave transformation methods and LST models. Although the weighted ensemble mean could provide an estimate of LST patterns along the coast, addressing the residual uncertainties (arising from other sources, discussed at the end of this paper), in future works, would help increase the certainty of the estimations.