Quantifying the contribution of abrasion versus. block removal to subglacial erosion on basement rocks: very-high resolution DSM analysis Forsmark, east Sweden.

Glacial erosion is generally thought to encompass two main erosion mechanisms: abrasion and
block removal (by plucking or quarrying, glacial ripping, conchoidal fracturing and other processes).
Landforms left by abrasion and block removal are distinct: abrasion results in smooth surfaces,
marked with grooves, striae and polish; block removal leaves sockets, sharp edges and characteristic
plucked faces. It is commonly assumed that erosion by block removal is faster and more important
than abrasion, but few studies have provided direct, quantitative comparisons between abrasion and
block removal.
In this study we estimate the depth of glacial erosion by abrasion and block removal by the Fennoscandian Ice Sheet during the last glaciation, the Late Weichselian, at the Forsmark site, east-central
Sweden. We combine previous erosion depth estimates from terrestrial cosmogenic nuclide (TCN)
inventories with a detailed geomorphological analysis of five selected sites at the Forsmark site: Stora
Asphällan, Stånggrundet and three sites on the Lilla Sandgrund island. The geomorphological analysis
used drone-acquired high-resolution Digital Surface Models (DSM) and orthophotos and involved
detailed geomorphological, bedrock and fracture mapping and quantitative morphometric analysis.
The geomorphological mapping separated out three types of surfaces:
• abraded surfaces, with low roughness, long topographic wavelength (> 10 m), low slope surfaces,
showing only evidence of abrasion;
• block-removal surfaces, with higher roughness, locally steep slopes, fresh fracture surfaces and
a small (< 2 m) topographic wavelength of topographic highs. These surfaces include surfaces
produced by typical lee-side and lateral plucking but also sockets and crescentic scars;
• re-abraded block-removal surfaces, with similar shape as the block-removal surfaces, except
showing signs of minor abrasion, polishing and edge rounding.
Morphometric analysis of the DSM shows systematic differences in roughness and slope between
abraded and block-removal (including re-abraded) surfaces. Field mapping showed both sharp and
rounded edges in block removal domains. The abundance of sharp edges suggest that block removal
was particularly active just prior to deglaciation. The Lilla Sandgrund sites showed an abundance
of sockets and crescentic scars, whereas Stånggrundet showed an abundance of typical lee-side and
lateral plucking features. The spatial extent of block removal surfaces (fresh and re-abraded) varies
from c 50 % at the Lilla Sandgrund sites to c 90 % at Stånggrundet; fresh block removal surfaces
vary from 6–37 % of surfaces; these were not separated for Stånggrundet.
Spatial variability of extent and depth of erosion by block removal is related to fracture networks.
Stånggrundet, with abundant lee-side plucking surfaces, has a dense, well-connected fracture pattern
with abundant gently dipping fractures that delineates small blocks. In contrast, the Lilla Sandgrund
sites, with a high proportion of abraded-only surfaces and relatively abundant sockets and crescentic
scars, has a lower fracture density, dominated by steeply inclined fractures, with fewer gently dipping
fractures. At Stora Asphällan, significant in-site variation exists in fracture networks and geomorphological features between amphibolite and fine-grained felsic gneiss.
By modelling an extrapolated abraded surface over the present-day surface, the depth of erosion by
late block removal can be spatially estimated. The estimated depth of block removal ranges from
0–1.3 m, and averaged 0.2–0.35 m where it did occur, i.e. excluding the abraded surfaces. Averaged
over the entire area including the abraded surfaces, the average depth of late block removal amounted
to 0.1–0.3 m.
These estimates can be combined with estimates of erosion depths from TCN inventories from
previous work (Hall et al. 2019a, 2023). Results from samples from summit surfaces mapped as
abraded surfaces range between 1.17–1.40 m; results from other surfaces, including re-abraded
surfaces reach 2.18 m. Based on the assumption that abraded surfaces developed solely under abrasion and without block removal, the abrasion:plucking ratios, in terms of depth of erosion, varied
from 16:1 to 3:1 across the different study sites. Estimates of erosion by abrasion from Wave Rock
6 SKB TR-23-27
summit areas, further inland, are considerably less: 0.2–0.4 m (Hall et al. 2023). The difference in
estimated abrasion depth between Wave Rock and coastal outcrops at Forsmark maybe related to
coarser rock grain size and/or higher local relief at Wave Rock. All these estimates are applicable to
relative topographic highs: the excavation of topographic lows (trenches, joint-valleys) would need
to be added to a regionally averaged erosion depth. The high abrasion:plucking ratio questions the
common assumption that block removal is faster and more important than abrasion. The results
of this study show that, at least at Forsmark, the opposite is the case.
The difference in areal extent of abraded-only surfaces – high in the Lilla Sandgrund sites; low at
Stånggrundet – suggests that abraded-only surfaces diminish in total surface area as subglacial erosion
progresses. This process starts with the development of sockets and crescentic scars, which progressively amalgamate into composite sockets, allowing lee-side plucking to proceed, locally aided by
isolated subhorizontal fractures. Overall this leads to a progressive roughening of the surface. This
implies that the development of block removal surfaces at some sites does not follow up-ice migration
of lee-side steps, as in other published models of plucking.
Compared to other glacial erosion studies (including both ice sheets and smaller glaciers), the results
from east Sweden show a very low overall erosion depth of crystalline rock, a very low erosion coefficient and a high abrasion:plucking ratio over the Late Weichselian glaciation. This may be caused
by a combination of: i) hard basement rocks, with locally fracture patterns unfavourable for block
removal; ii) a flat starting surface controlled by the Cambrian unconformity, suppressing plucking;
iii) a long period of thick-ice conditions that suppressed block-removal including plucking, compared
to a very short period of thin-ice, ablation-zone conditions during deglaciation that favoured block
removal including plucking.
Overall, the depth of erosion at Forsmark during the Late Weichselian glaciation was low. Abrasion
contributed between 0.2 to 1.5 m of erosion, in line with results from other ice-sheet settings. Block
removal by plucking and formation of sockets and crescentic scars contributed another 0–1.6 m
of erosion. Locally, affecting perhaps 10–20 % of the area, glacial ripping added a further 2–4 m
of erosion. Somewhat deeper erosion occurred in the relatively shallow trenches and valleys in the
Forsmark region