This manuscript presents a global dataset of ten wavecut scarps formed by successive Pliocene sea-level oscillations that are presently between 6 and 109 m above sea level. Correcting for glacial isostatic adjustment does not explain the observed difference in Pliocene paleoshoreline elevations around the globe. The main aim of the manuscript is to investigate whether mantle-flow-driven topography can explain these differences. A series of 27 backward advected mantle flow models is presented across which the radial viscosity and 3-D buoyancy structure are varied. The 3-D buoyancy structures were obtained from two tomographic models in the upper mantle and four tomographic models in the lower mantle. A plate correction is applied to the dynamic topography models, resulting in 135 models in total. Two criteria (either weak or stringent) are used to assess the success of dynamic topography models: the Mean Weighted Standard Deviation (MSWD) of the GIA- and DT-corrected scarp elevations, and the inferred GSML value for the dynamic topography model. The results suggest that while dynamic topography likely contributes to sea level change over these time scales, global mantle flow models do not predict dynamic topography changes at the scale of the inferred topographic oscillations along scarps (tens to hundreds of kilometres). An important contribution of the manuscript is to expose and discuss the limitations of the mantle flow models. My main recommendation is to provide more details and representations of the dynamic topography models to make the contribution accessible to readers who are not experts on this topic.

 

The results are promising in terms of deriving Pliocene paleoshoreline elevations around the globe by combining global digital elevation models with geological observations and data. The presented glacial isostatic adjustment correction includes error propagation and considers 36 radially symmetric viscosity structures. The predictions of the considered dynamic topography models significantly differ from one another, and between 1/135 and 15/135 models satisfy the stringent criteria at five of the ten considered sites. At four sites, between 3/135 and 4/135 models satisfy the weak criteria. There is one site for which none of the considered models succeeds.

 

I found the discussion to be appropriate and balanced. It mentions that other tectonic processes (flexure, deformation) may be at play and that an important limitation is that tomographic models are too smooth to resolve the topographic trends observed along GIA-corrected scarps. Despite the limited success of the dynamic topography models, it is noted that there are five scarps for which the mantle flow models predict consistent GMSL estimates. Results are emphasised for the well-dated De Hoop scarp (for which 3/135 dynamic topography models succeeded), which leads to a GIA- and DT-corrected GMSL 11.6 ± 5.2 m, a range that would require little melting of ice sheets under warm Pliocene climate conditions.

 

I appreciate that the limitations and uncertainties of the mantle flow models are exposed. Dynamic topography predictions are essential to the manuscript, however, no global maps of such predictions are shown. It would be helpful to show maps of dynamic topography and/or of the change in dynamic topography either globally and/or at some or all of the ten sites. It would also be helpful to represent the considered parameter space for the flow models graphically, and which models succeed or fail within that parameter space. This could be done across a series of X-Y plots or using some more sophisticated visualisation. 

 

While the sensitivity of the mantle flow models to viscosity and buoyancy is exposed, it would be good to remind the reader that the viscosity space is large. A figure showing the different viscosity structures considered in the study would be helpful, ideally with some context of previous work (e.g. Mao and Zhong, 2021, https://doi.

org/10.1029/2020JB021561). Showing differences in the buoyancy structure would require some visualisations of the mantle flow models, perhaps at present-day. It would be helpful to mention that the buoyancy structure depends on the conversion factor from relative seismic velocities to relative density variations. While a reasonable choice was made for the study, other choices would be possible. It might also be possible to mention the adjoint approach for mantle flow modelling, and its possible suitability to the problem at hand.

 

It is not obvious from the manuscript how GMSL (global mean sea level?) corrections are obtained from each mantle flow model. It seems that this is done at each site, which I find confusing because at one location one can infer relative sea level, not global sea level. I do not think it is sufficient to refer the reader to Hollyday et al. (2023) for this calculation.

 

I do not understand what plate motion correction is applied, and how it results in five new variants of each dynamic topography model (from 27 models to 135 models, L. 231-232). I do not think it is sufficient to refer the reader to Hollyday et al. (2023) for this correction.

 

How is the lithosphere defined in the models? It would be helpful to be explicit even though the potential role of the lithosphere is acknowledged in the discussion (see also Davies et al., 2019, https://doi.org/10.1038/s41561-019-0441-4). Could short-wavelength lithospheric thickness variations and associated flow explain some of the topographic observations along fault scarps?

 

L.582-583: 'The preferred model also predicts patterns of present-day DT consistent with observations (present-day DT varying from ~100 m to >500 m; Hoggard et al., 2017).' Should this criterion (fit to residual topography) be used to filter the predictions of all dynamic topography models? How many scarps is the preferred model consistent with? Which model is consistent with the most scarps?

 

Early in the manuscript, it seemed that scarps would only be constrained using remote sensing data and TerraceM. It became clear later on that remote sensing analysis was carried out for scarps for which geological and/or field observations are available. It would be helpful to mention this early on.

 

 

L. 574: should this be Fig. 8?

 

L. 588-591: what are GIS, WAIS and EAIS?

 

Is GMSL defined in the manuscript?

 

There should be a space between consecutive units (see rates of change units)