Beach rotation through longshore sediment transport can alter shoreline orientation, increasing erosional risks for coastal communities at short-term (storm events) and decadal to centurial time scales (long-term evolution). Identifying, predicting and planning for embayed beach rotation requires understanding of the morphological response to changes in directional wave climates driven by atmospheric forcing. Through assessment of annual to centurial datasets, this thesis aims to improve understanding of beach rotation at local (embayment) to regional (coastline) extents. The multi-annual rotational response of a 12-km, longshore-dominated gravel embayment (Start Bay, Devon, UK) to a set of extreme and contrasting bi-directional winter wave conditions was assessed using multi-method topo-bathymetric surveys. Previously, the limited extent of intertidal measurements constrained insights into sediment pathways during rotational events; however, this study found that accounting for measurement uncertainty was critical in the calculation of robust total sediment budgets (sub-aerial to sub-tidal), allowing identification of full-embayment rotation. Application of this new methodology revealed that under extreme directional wave energy (>1:50 year southerly winter season), full-embayment rotation (6.5 x 105 m3 transport) is possible through headland bypassing (50% of total) and sub-tidal change (33% of total), providing new insights into sediment transport pathways for gravel beaches. Retrieval of beach volumes after rotational events requires sustained or equal extreme wave energy from opposing directions; however, only sub-embayment rotation was observed (under opposing easterly conditions), proposing that headland bypassing is asymmetrical, extending previously understood timescales of recovery. Analysis of hindcast modelled wave data and a 10-year record of 36 intertidal beach profiles located across the full embayment of Start Bay, revealed that interannual to decadal scale beach rotation may be predicted by a new index of the normalized wave power directional balance (WDI), suggesting that subtle variations in bi-directional wave climate drive sustained changes in beach planform, increasing erosional risk at embayment extremities. Assessment of the modelled wave climate (1980 - 2018) highlighted that the two dominant wave directions (southerly and easterly) are correlated with winter averages of two key climate indices, the North Atlantic Oscillation (NAO) and West Europe Pressure Anomaly (WEPA). This is the first robust demonstration of the relationship between negative NAO and easterly wave power, indicating that atmospheric variability significantly explains the WDI and beach rotation at this location. Spatial expansion of this analysis revealed that bi-directional wave climates are regionally comparable, extending throughout the length of the South coast of England. Similar correlations with both climate indices were exhibited, showing other rotational sites are controlled by the atmospheric influence of bi-directional wave climates. Examination of 10-15 years of intertidal beach profiles at 22 embayed South coast locations identified 11 sites exhibiting significant rotational responses, with regionally coherent common factors, including oblique shoreline orientation to bi-directional wave approach, steep slopes and coarser sediment. Beach rotation was shown to be correlated with the WDI for most sites, and significant direct correlations between beach rotation and WEPA at a number of locations, indicates the future potential skilful forecasts of atmospheric indices may have in predicting seasonal rotation at regional to basin-wide scales. To explore the significance of recent (10-year) beach rotational behaviour in Start Bay within centurial timescales, a stepwise multi-linear regression model was developed to hindcast the WDI using long-term (1906-present) sea-level pressure records of the NAO and WEPA. It was found that combining the two indices vastly improved the predictive skill of the regression model when compared to using individual indices (R2 of 0.66 between model data and index predicted values of the WDI). Qualitative validation of beach rotation in response to the >100-yr hindcast WDI timeseries was achieved from proxy records of oblique photography and topographic maps, finding that low frequency (~60 years) phases of clockwise/anticlockwise beach rotation followed positive/negative phases of the detrended cumulative hindcast WDI record, over the period 1906 to 2018. This variability reflects observed multi-decadal fluctuations in phases of NAO (~60-80 year) and WEPA (~50-60 year), demonstrating atmospheric control of directional wave climate and beach rotation over centurial timescales, explaining historical accounts of coastal settlement relocation. When examined in the context of millennial-scale proxy NAO reconstructions, the recent centurial-scale analysis does not capture the much greater magnitude and duration of past detrended cumulative variability observed over the last 3000 years, indicating that previously inferred phases of extreme coastal realignment may recur in the future, presenting a significant long-term issue for locations affected by beach rotation. This work contributes new insights into embayment rotation at different spatial and temporal scales. Application of a new total sediment budget approach improves knowledge of full embayment rotation and recovery, whilst bi-directional waves are shown to predict beach rotation, driven by atmospheric forcing at medium to longer timescales. This thesis directly contributes towards new understanding of the past and future timescales of beach rotation, as well as proposing a mechanism for season ahead forecasting based on atmospheric variability.