A coupled cosmogenic-modelling approach to quantifying Holocene erosion of Ireland’s hard-rock coastal cliffs
Abstract. Hard rock cliffs and shore platform are a key feature of the coastal environments in Ireland. Being highly influenced by marine driven erosion, the relationship between changes in relative sea level (RSL) and cliff retreat rates are of major interest when analysing coastal evolution. At the western coast of Ireland, the rates of RSL rise in the Holocene has varied and some locations experienced marine driven erosion earlier than others. Therefore, for understanding coastal hard-rock erosion and the influences of long-term changes, such as RSL changes, studying multiple at locations is crucial.
In this study, we investigated cliff erosion rates on millennial scales at two locations on the western Irish coast – Muckros Head and Ballard Bay – using cosmogenic beryllium-10 concentrations. Our findings indicate that the shore platforms at both locations were formed entirely during the Holocene, and that each site has experienced a long-term deceleration in cliff retreat, reflecting declining rates of RSL rise. The cliff retreat rate decelerated from 14 to 1 mm yr-1 at Muckros Head within the past ~9500 yrs, and from 25 to 5 mm yr-1 at Ballard Bay over the last ~5000 yrs. The best-fit models suggest shore platform widening at Muckros Head and steady-state erosion at Ballard Bay. Whereas the pronounced deceleration in cliff retreat observed at Muckros Head, relative to Ballard Bay, is explained by the elevated early-Holocene RSL at the former, differences in contemporary retreat rates between our sites likely reflect contrasting geomorphological and geological conditions and marine environments.
The authors use a coupled cosmogenic-modelling approach to investigate (i) whether the two shore platforms at Muckros Head and Ballard Bay on the west coast of Ireland are products of ongoing Holocene erosion or relics form earlier sea-level high stands, and (ii) the influences of local parameters (e.g., lithology, RSL, wave climate) on long-term cliff retreat rates. The dataset is novel and represent a valuable contribution to “rocky coast” and “cosmogenic radionuclides” community. The dataset is potentially valuable, but the interpretation in its current form is not fully supported by the evidence presented. I believe the manuscript has merit but requires major revisions before it can be considered further.
My main concern is that the number of samples (n = 4 at MH site, n = 6 at BB site except background sample) is too small to robustly constrain the model, which makes it difficult to support the main conclusions. Comparable studies that applied similar models typically analyzed a considerably larger number of samples (e.g., n = 19 in Swirad et al. (2020); n = 14 in Jeong et al. (2024)). If feasible, the author could either increase the sample size, or I suggest the authors at least tone down their conclusion, given the limited number of samples.
Also, the boundary conditions of the model are not clearly described. Previous studies have explicitly accounted for the tidal range and RSL uncertainty when modeling shore platform development.
For example,
Swirad et al. (2020):
“For all scenarios, we set a boundary condition that at any point, platform elevation could not exceed the highest astronomical tide (HAT = 3.2 m) relative to the RSL at that point.”
Jeong et al. (2024):
“The total time required for the formation of the present shore platform was calculated by determining the time taken for the sea level curve to reach the lowest part of the platform, taking into account the HAT (0.775 m) and RSL curve error (1.5733 m) (Song et al., 2018).”
However, only tidal range information is provided in the manuscript (MH: 3.2 m, BB: 3.7 m in Table 1). Please include data from the nearest tidal gauge such as Highest astronomical tide (HAT), Mean high water level of spring tides (MHWS), Mean high water level of neap tides (MHWN), Mean low water level of neap tides (MLWN), Mean low water level of spring tides (MLWS), and Lowest astronomical tide (LAT). And how did the authors set boundary condition?
Also, it is unclear that how the authors calculated the total time required for the formation of the present shore platform. It seems that the authors assigned a value of 9.5 kyr to the MH site, and 5 kyr to the BB site, but the rationale for this choice is not clearly explained. In other words, the authors should consider when the water reach the lowest point of the shore platforms.
Since this paper aims to establish the influences of local parameters (e.g., lithology, RSL, wave climate) on long-term cliff retreat rates, I suggest that the authors further investigate how the lithology at each study site influences the modeled cliff retreat rates. The cliff at the MH site is predominantly composed of sandstone, whereas the cliff at the BB site is primarily composed of siltstone. Given that previous studies (e.g., Premaillon et al., 2018) have compiled historical and contemporary cliff retreat rates for various rock types, the authors could compare their modeled retreat rates with the reported rates for the same lithologies. This comparison would help establish whether the modeled rates are relatively high or low compared with previously reported rates for similar rock types.
Line 30: Please provide a more specific explanation of how the differences in contemporary retreat rates between your sites likely reflect contrasting geomorphological and geological condition and marine environments.
Figure 1: Please add bathymetry to the figure.
Study sites and Table 1: Although the two study sites are geographically close to each other, their RSLs differ substantially. I consider the RSL difference to be a key parameter in the model. Although the authors have explained the RSL differences between the two sites by referring to factors such as isostatic rebound, the explanation remains somewhat limited. What explains the differences in RSL between the two study sites?
Figure 2A, Figure 3A: Please add the viewing azimuth to the photograph.
Line 126, Line 299: It seems that the platform at MH site experienced severe weathering and bioturbation. The authors could consider such impacts on downwearing of platform and cliff retreat.
Line 127: “regional shoreline transgression driven by either tectonic uplift or sea-level lowering”. The statement that shoreline transgression was driven by tectonic uplift and/or sea-level lowering appears to be reversed.
Line 163: The platform down-wearing rate of 0.17 μm yr-1 could be compared to other measured platform down-wearing rates using MEM to determine how high the rate is.
Line 227: “study(Regard et al., 2012).” should be “study (Regard et al., 2012).”
Line 249: The ‘zero platform erosion’ model … (Choi et al., 2012).
Choi et al. (2012) presented their modeled 10Be age using an erosion rate of 5 mm kyr-1, rather than assuming zero erosion.
Line 305, Line 307, Line 310 and Line 314: Please use commas to separate thousands (e.g., 5,886)
Figure 4: Should the sample numbers be labeled in the Figure? Also, the cap lines are missing from some error bars. Please improve the quality of the figure.
Table 3: The 10Be concentrations of samples MUC-1 to MUC-4 were corrected using blank B5, except for sample MUC-5, which was corrected using blank B9. Please explain why a different blank (B9 blank) was used for MUC-5.
Figure 6B caption: “For Ballard Bay, the best fit-scenarios describe decelerating cliff retreat during the Holocene, from initially 25 ± 2 mm yr-1 to a recent rate of 5.0 ± 0.4 mm yr-1, coincident with shore platform widening (MSD =1.22*106 (atoms g-1)2).”
In the graph, the open circle representing ‘steady-state erosion’ near scenario number 45 has the lowest MSD. Also, in abstract and conclusion, the authors argued that “the modelling suggests that shore platforms developed via parallel to the cliff retreat (steady-state) at Ballard Bay.”
Please explain why the best fit model for site BB corresponds to ‘shore platform model’ in the figure caption.
Line 386: “both field” should be “both field.” And please format “s” as a subscript in Hs.
Lines 445-452 and Lines 506-508: “At Ballard Bay, the comparatively smooth shore platform allows a wide swash zone, and broken waves can reach the cliff base regularly. The absence of talus slopes at the cliff base and smaller clasts on the upper shore platform is further indication that waves regularly reach relatively high elevations; such wave action mobilises sediment and ultimately drives clast abrasion of the platform surface.”
The cliff base for site BB is ~10 m, and the significant wave heights on Ireland’s west coast are typically 3–3.5 m and can surpass 5 m during winter months. How do the authors demonstrate that upper shore platform and cliff base were “regularly” exposed to wave erosion?
Lines 470-476: The paragraph explaining the presence of the high cliff base elevation in terms of the marine and terrestrial processes is a good approach, but the interpretation appears speculative. More evidence is needed to support the authors’ interpretation.
Lines 483-484: “This peak, which most runs failed to simulate, probably results from tide-induced variability in in situ 10Be production (Hurst et al., 2017).”
This explanation is too general and should be more specific. The authors should explain how tide-induced variability affects the model results.
Lines 503-506: “Despite the location of cliff/shore platform junction at both sites extending several metres above modern sea level, the backing cliffs at each site are still impacted by marine driven erosion, the severity of which is influenced in by the shore platform morphology, lithology and exposed vs protected location.”
I think this is the answer for the authors’ research objective (ii): “this study aims to establish the influences of local parameters (e.g., lithology, RSL, wave climate) on long-term cliff retreat rates.” However, again, the answer is too general and should be more specific. How did the shore platform morphology, lithology and exposed vs protected location affect 10Be concentrations?