Sea cliff morphology and erosion rates are modulated by several factors, including rock control that reflects both lithology and rock structure. Erosion is anticipated to preferentially exploit “fractures”, broadly meant as any discontinuity in an otherwise continuous medium, where the rock mass is weakest. Unpicking the direct control of such fractures on the spatial and temporal pattern of erosion remains, however, challenging. To analyze how such fractures control erosion, we monitored the evolution of a 400 m long stretch of highly-structured sedimentary cliffs in Socoa, Basque Country, France. The rock is known as the Socoa Flysch Formation. This formation combines decimeter-thick turbidites composed of repeat triplets of medium to strong calcareous sandstone, laminated siltstones and argillaceous marls. The sequence plunges at 45° into the sea with a shore-parallel strike. The cliffs are cross-cut by two normal and reverse fault families, with 10 – 100 m alongshore spacing, with primary and secondary strata bound fractures perpendicular to the bedding, which combined delimit the cliff rock mass into discrete blocks that are exploited by erosion process.

Erosion, and sometimes plucking, of such beds and blocks on the cliff face was monitored using ground-based Structure-from-Motion (SfM) photogrammetry, over the course of 5.7 years between 2011 and 2017. To compare with longer time change, cliff-top retreat rate was assessed using SfM-orthorectified archive aerial photographs spanning 1954-2008. We show that the 13,250 m2 cliff face released 4500 blocks exceeding 1.45 ×10-3 m3, removing a total volume of 170 m3. This equates to an average cliff erosion rate of 3.4 mm/yr, which is slightly slower than the 54-year-long local cliff top retreat (10.8 ± 1.8 mm/yr).

The vertical distribution of erosion reflects the height of sea water inundation, where the maximum erosion intensity occurs ca. 2 m above high spring-tide water level. Alongshore, the distribution of rockfall scars is concentrated along bed edges bounding cross-cutting faults; the extent of block detachment is controlled by secondary tectonic joints, which may extend through several beds locally sharing similar mechanical strength; and, rockfall depth is always a multiple of bed thickness.

Over the longer-term, we explain block detachment and resultant cliff collapse as a cycle. Erosion nucleates on readily exploitable fractures but elsewhere, the sea only meets defect free medium-strong to strong rock slabs offering few morphological features for exploitation. Structurally delimited blocks are quarried, and with sufficient time, carves semi-elliptic scars reaching progressively deeper strata to be eroded. Lateral propagation of erosion is directed along mechanical weaknesses in the bedding, and large episodic collapses affect the overhanging slabs via sliding on the weak marl beds. Collapse geometry is confined to one or several triplets of turbidite beds, but never reaches deeper into the cliff than the eroded depth at the foot. We contend that this fracture-limited model of sea-cliff erosion, inferred from the Socoa site dynamics and its peculiar sets of fractures, applies more broadly to other fractured cliff contexts, albeit with site specific geometries. The initiation of erosion, the propagation of incremental block release, and the ultimate full failure of the cliff, have each been shown to be fundamentally directly controlled by structure, which remains a vital control in understanding how cliffed coasts have changed in the past and will change in the future.