The interplay of bedrock fractures and glacial erosion in defining the present‐day land surface topography in mesoscopically isotropic crystalline rocks

This paper addresses the effect of fractures within crystalline bedrock on glacial erosion processes in fast flowing hard bed glacier environments. In particular, we examine (i) whether the fracture type is critical for the capability of a glacier to erode the bedrock through quarrying/plucking processes and (ii) whether we can recognize specific fracture‐controlled erosion signatures from bedrock surface morphologies. We conducted an investigation within the northern part of the Åland Islands, southern Finland, where the ice‐flow direction (N–S) has remained constant through Late Pleistocene glaciations and where the bedrock is characterized by a lack of any mesoscopic anisotropies (such as foliation) and hence provides an optimal target to recognize the relationships between fractures and erosional morphologies. We characterized the fracture systems within the bedrock using both UAV‐acquired orthophotographs and standard field approaches and extrapolated the results to larger scales using LiDAR‐based digital elevation models. Our findings indicate that individual joints or shear fractures are associated with the development of minor vertical breaks along the bedrock surface. However, they do not provide sufficient mechanical weakness zones in the bedrock to allow effective glacial quarrying, even though their lengths can be relatively large (>50 m). By contrast, the linkage of several parallel shear fractures or the presence of larger faults with gouge‐bearing cores and well‐developed damage zones leads to localized disintegration of the rock material and the subsequent development of distinct topographic depressions along the bedrock surface. Consequently, the results allow predictions to be made about the bedrock features underlying the observed topographic signatures along the bedrock surface. Applied to the area of this investigation, abrasion associated with N–S‐directed glacial flows is responsible for the N–S‐oriented elongate but smooth fjord‐like megagrooves, whereas the more abrupt topographic breaks were generated by quarrying controlled by sub‐vertical, E–W‐trending zones of localized brittle deformation.


| INTRODUCTION
Glacial erosion has had a major impact on the present-day landscapes in areas affected by repeated glaciations during the Quaternary period (2.6 Ma), particularly in areas formerly covered by major continental ice sheets, such as the Antarctic (Thomson et al., 2013), Greenland , North American (Sugden, 1978) and Eurasian (Patton et al., 2016) ice sheets. The bedrock within areas of areal scouring displays irregular rock surfaces resulting from the processes of glacial erosion, including the mechanisms of abrasion, quarrying and glacial ripping (Glasser et al., 2020;Glasser & Bennett, 2004;Iverson, 1991;Krabbendam et al., 2022). Glacial abrasion refers to wearing of the bedrock surface caused by the rock fragments at the glacier base and may involve scoring and polishing, associated with larger (>1 cm) and finer fragment sizes, respectively. Quarrying (or plucking) refers to the expansion of fractures within the bedrock due to glacial stresses and the subsequent removal of >1 cm rock fragments. Similarly to quarrying, glacial ripping uses the bedrock fractures but removes a multitude of fragments in the same event. The capacity and style of glacial erosion are largely linked to the character and dynamics of the ice sheet, such as ice thickness (Anderson, 2014), iceflow velocity (Herman et al., 2015;Yanites & Ehlers, 2016), the temperature and hydrogeology at the base of the ice (Clarke, 2005;Lai & Anders, 2021;Neal Iverson, 2012;Sugden, 1974) and glacier bed morphology (e.g., Sugden, 1978). However, a growing body of evidence indicates that the properties of the bedrock underlying the glaciers also substantially affect the erosion processes and the resulting bedrock surface morphologies (e.g., Krabbendam & Glasser, 2011).
Lithological control over glacial erosion may relate to the (i) porosity of the bedrock, with slower flow rates and less erosion in porous rocks (Sugden, 1978); (ii) bedrock homogeneity or heterogeneity, which contributes to the meso-scale morphological character of the bedrock surface (Glasser et al., 1998); and (iii) the variable capacity of different lithologies to resist glacial abrasion (Harbor, 1995;Lane et al., 2015). Of particular significance are the contrasts between hard crystalline and softer sedimentary rocks, which have resulted in contrasting landscapes. Examples include transitions from rounded whalebacks to linear mega-scale glacial forms in the Canadian Shield  or sub-glacially overdeepened basins that are compartmentalized due to the presence of erosion-resistant rock units (Gegg et al., 2021;Woodard et al., 2021).
The bedrock structure controls glacial erosion through ductile foliations and folds, discrete fractures and fault zones. Ice-flow-parallel foliations in metamorphic rocks promote the generation of elongate hollows, whereas those in isotropic lithologies are more uniform (Glasser et al., 1998). Within otherwise coherent bedrock, discrete fractures of typically pre-glacial origin (Dühnforth et al., 2010;Hooyer et al., 2012) are key features, which control the glacial erosion and the resulting glacier bed morphology (Woodard et al., 2021). The fractures are used by glacial quarrying processes, which typically operate within the lee side of the rock drumlins or Roches moutonnées (Bennett & Glasser, 2009;Iverson, 1991;Rastas & Seppälä, 1981). Fracture intensity governs the style of glacial erosion, and quarrying is the dominant mechanism in intensely fractured areas, whereas abrasion characterizes more intact rock volumes (Briner & Swanson, 1998;Crompton et al., 2018;Dühnforth et al., 2010;Woodard et al., 2019). Similar controls have been documented for sedimentary rocks, where thin-bedded and thick-bedded units may display dense and sparse jointing, which consequently favour quarrying and abrasion, respectively (Krabbendam & Glasser, 2011). The dip of the primary bedding and associated fractures may also control the style of erosion, as demonstrated by the dominance of abraded surfaces when bedding planes dip gently of moderately in the direction of the topographic basins (Kelly et al., 2014) or the flow-direction of the glacier . By contrast, cliffs and overdeepenings characterize areas where bedding dips in the opposite direction with respect to the slope of the valley (Kelly et al., 2014), and down-ice dipping bedforms are controlled by moderate bedding dips . Elongate topographic valleys of larger lengths and lateral extents in gneiss terrains are understood as products of syn-glacial removal of preferentially weathered (Krabbendam & Bradwell, 2014;Olvmo et al., 2005;Olvmo & Johansson, 2002) or otherwise intensely fractured domains of bedrock (Dühnforth et al., 2010;Scott & Wohl, 2019;Skyttä et al., 2015).
Abrasion has been considered as the dominant erosion mechanism under fast-flowing glaciers (Herman et al., 2015;Yanites & Ehlers, 2016). However, many (recent) contributions indicate that glacial quarrying and ripping are the dominant mechanisms in (intensely) fractured crystalline bedrock (Boulton, 1987;Dühnforth et al., 2010;Glasser et al., 2020; and result in effective erosion, particularly during glacial retreat. In contrast to the overriding positive correlations between fracture intensity and erosion efficiency, some recent numerical modelling indicates that, that, in the specific case of ice-bed parallel fractures, larger fracture spacing may promote quarrying (Woodard et al., 2019). Bearing these in mind, it is evident that both the composition and the structure of the bedrock contribute to the selection of the glacial erosion mechanism and efficiency, but less is known about which types of fractures (mechanical discontinuities) are effective in promoting quarrying and the development of distinct elongate bedrock depressions.
In this paper, we show that individual joints or shear fractures, even with substantial (lateral) extents, do not provide sufficient mechanical weakness zones to lead to the generation of significant elongate depressions along the bedrock surface topography. More specifically, such fractures localize glacial plucking, leading to the characteristic smoothed slopes and steep edges on the stoss and lee side of the bedrock bumps, respectively (Krabbendam & Bradwell, 2011 We examined the topography of glacially eroded crystalline bedrock in the northern part of the Åland Islands, southern Finland tions. The study area displays exceptionally well-exposed bedrock outcrops (Figure 2b), from which we characterized the fracture systems within the bedrock using both UAV-acquired orthophotographs and geological field mapping methods. For extrapolating the study results to the semi-regional scale, we conducted structural interpretations on LiDAR-based digital elevation models (DEMs).

Reconstruction of the dynamic behaviour of the Fennoscandian
Ice Sheet (FIS) through the Pleistocene glacial cycles is based on geological evidence such as lineations, striations, till fabric and stratigraphy, glaciofluvial sediments and prominent end moraine complexes (e.g., Boulton et al., 2001;Larsen et al., 2016;Stroeven et al., 2016;Svendsen et al., 2004), as well as numerical ice sheet modelling The FIS flow patterns are generally well established for the last glacial cycle (e.g., (Boulton et al., 2001), and they probably exhibited similar characteristics during the preceding Pleistocene glaciations Svendsen et al., 2004). The BSB has thus been repeatedly subjected to glacial erosion by the FIS during the Pleistocene epoch (Boulton et al., 2001;Patton et al., 2016;Svendsen et al., 2004). In fact, one of the models suggested for the origin of the BSB is erosional over-deepening under the FIS during glaciations (Amantov et al., 2011;Puura et al., 2003). This is supported by the fact that up to 87% of the total Pleistocene sediment volume in the sink area in continental Europe (i.e., Poland) was deposited during and after Marine Isotope Stage 12, and the material originates from the BSB .
The present study area lies centrally to the Baltic Sea ice stream with ice repetitiously moving in the north-south direction, as indicated by bedrock striations (360 ± 5 ) and glacial lineations ( Figure 1b). Glacial lineations in the Geta area are mostly rock drumlins formed parallel to the ice flow with significant polishing of their stoss sides. Rock drumlins are often considered old and long-lived bedforms that were reshaped during successive glaciations, thereby steering the local ice-flow directions (Krabbendam et al., 2016). Some of the rock drumlins in the present study area have a thin till cover on their lee side. The study area was deglaciated at around 10.8 cal ka BP, during the Yoldia Sea phase in the BSB history, with the relative sea level in front of the retreating glacier lying at 150 m above the present-day level (Ojala et al., 2013;Stroeven et al., 2016). Today, roughly 90-95% of the study area is composed of bedrock outcrops with some depressions filled by postglacial clays.

| MATERIALS AND METHODS
The area of the field investigations (outcrop scale) covers an approximately 4-km-long, E-W-oriented stretch along the well-exposed 2). The outcrops are exceptionally clean due to pronounced glacial erosion and exposure to present-day wave and sea ice activity. There is typically no vegetation within a distance of 50-100 m from the coastline; further inland, outcrop becomes gradually more covered and less accessible to low-altitude aerial imaging due to vegetation ( Figure 3a,d).
We approach the research questions by characterizing and classifying the brittle structures of the bedrock and determining their effects on bedrock morphology. Furthermore, we extrapolate these findings to a semi-regional-scale interpretation, which covers the 6 by 6 km area of the Geta Island (Figures 2b and 6a), using a LiDAR elevation model (0.5 points/m 2 ; available from the National Land Survey).
We conducted the outcrop-scale structural characterization of the bedrock using drone-acquired orthophotographs as the primary dataset. These orthophotographs have a lateral resolution of 0.55 cm per pixel, and they cover a total extent of 328 000 m 2 . The first step

| Extrapolation of the outcrop-scale results to less-exposed adjoining areas (semi-regional scale)
This section documents our interpretation how the results from the detail area (Figure 2) can be used to provide a geologically motivated structural interpretation within a semi-regional scale, allowing usage of lower-resolution datasets and minimizing the time required for the structural characterization of a larger area. We emphasize that our semi-regional scale structural interpretation ( Figure 6) is based solely on the topographic signatures of the LiDAR data and did not include validation through field observations. For this reason, uncertainties and some sampling bias between the low-lying coastal and higher inland areas may be present, and the interpretation should be considered a first-pass model that can be validated and improved with focussed field observations.  (Kelly et al., 2014) or dipping towards the ice flow  have generated smoother topographic signatures characterized by abrasion. The higher fracture density and connectivity within the fault zones contribute to the fragmentation of bedrock into small block sizes and hence promote quarrying as the main mechanism of glacial erosion, as concluded by (Dühnforth et al., 2010). As the fragmented fault zones are topographically lower than surrounding abraded areas, it is evident that quarrying has been more effective than abrasion in eroding the bedrock.
This finding is in conflict with the model in which abrasion dominates over other subglacial erosional processes beneath fast-flowing glaciers (Herman et al., 2015;Yanites & Ehlers, 2016) but is in line with other studies indicating quarrying dominates over abrasion (e.g., Loso et al., 2004), with particular reference to effective subglacial quarrying during the final stages of glacial cycles, when abundant subglacial meltwater was available to aid in quarrying (Glasser et al., 2020). The most pronounced glacial erosion could be achieved by glacial ripping, involving hydraulic jacking of pre-existing bedrock fractures by overpressurized sub-glacial meltwaters , but no widespread bedrock fragmentation associated with sandy or silty fracture infills was found in the present study area. For this reason, we attribute the zones of the most pronounced glacial erosion to quarrying, localized by the presence of pre-glacial faults and fault zones.
As discussed earlier (Sections 4.2 and 4.3), the N-S-trending fractures occurring parallel to the ice flow were locally used by glacial quarrying (Figure 4b,d,e) but did not have a major role in defining equally continuous elongate topographic depressions comparable to those associated with the E-W-trending faults. Consequently, the N-S-trending fjord-like topographic valleys (or mega scours) in the Geta Island are megagrooves (Bradwell et al., 2008;Eyles, 2012;Stokes & Clark, 2003), which are pathways for enhanced glacial flow that developed during successive glacial advances (Roberts et al., 2010).
Megagrooves typically develop in well-stratified or layered rocks (Krabbendam & Bradwell, 2011), and their orientation may be controlled by the alignment of the lithological units and contacts (Roberts et al., 2010) or fractures (Bradwell et al., 2008). As s layering or stratification is absent from the bedrock of the present study area, it is likely that bedrock fractures had the primary contribution to the development of the observed megagrooves. Consequently, we suggest that the N-S-trending ice-flow-parallel fractures contributed to the process of lateral plucking (Krabbendam & Bradwell, 2011), which at small scales provided some control over the width of the roches moutonnées (Rastas & Seppälä, 1981) and at larger scales defined the margins of the larger elongate ice-flow-parallel fjord-like topographic depressions ( Figure 6). Moreover, we infer that the sub-horizontal icebed parallel fractures-which are poorly constrained in this studyplayed a role in the effectivity of glacial quarrying (Woodard et al., 2019) and ice-bed morphology variation.

| Implications to ice flow dynamics-driven glacier erosion in landscape evolution
Bedrock along the northern coast of the Åland Islands has been subjected to prolonged streamlining during successive glaciations, with particular contribution from the fast-flowing Baltic Ice stream during early Holocene (Greenwood et al., 2017). The flat-lying polished rock surfaces, distinct glacial striae and whalebacks are compatible with basal debris-driven abrasion under hard-bed conditions as the prevailing erosion mechanism during the Late Pleistocene (Glasser & Bennett, 2004). The scarcity of short distance transport of boulders and boulder fields on Åland further indicates that glacial quarrying, jacking or ripping Krabbendam et al., 2022) were not the regionally dominant erosion mechanisms. As such, conditions of glacial erosion in Åland indicate glacial conditions of thick and fast, constantly sliding ice (e.g., Evans, 1996;Glasser & Bennett, 2004), which is likely associated with a thin clay rich 'lubrication' layer of former basinal sediments beneath the streaming glacier. Similar conditions are typical for numerous locations around the FIS and the former Laurentide Ice Sheet region (Bukhari et al., 2021;Eyles, 2012;Eyles et al., 2016;Krabbendam et al., 2016). In areas of homogeneus bedrock lithology, we conclude that the presence of pre-existing fractures and topographic depressions-as recognized in this study-are a result of localized, structurally controlled quarrying that took place under dominantly abrasive erosional conditions. This finding is in line with the conclusions that site-specific fracture patterns is the most important feature in controlling the bedrock erosion and associated bed morphology (Krabbendam et al., 2016;Woodard et al., 2021). In a regional scale, bed morphology is the main controlling factor over the slip of hard-bedded glaciers (e.g., Eyles, 2012;Kamb, 1970), as it contributes to the resistance of the ice to flow across and around basal obstacles. Bingham et al. (2017) recognized that much of the variability in the basal traction of the West Antarctica glaciers relates to topographical variation at the ice bed in the scale of individual subglacial basins, and such variation may not be adequately constrained with the available geophysical methods. For the presently exposed glacier beds in Northern America and Fennoscandia, structurally controlled bedrock ridges have been shown to slow down or temporarily even stop the glacier as shown by the deposition of end moraines and glacifluvial deltas (Crossen, 1991;Skyttä et al., 2015). Such findings that even moderate (<100 m) vertical changes in the ice-bed topography contribute to the ice flow (Skyttä et al., 2015) further highlight the need for using structural geological approaches in addressing the topographical variations of the ice bed. This applies to understanding the major deformation zones (faults, shear zones and fracture zones) but also the spatial variability in fracture densities within the bedrock, as the latter has been shown to play a role in the development of surge-type glaciers (Crompton et al., 2018). the significance of the fracture type for glacial erosion into semiregional scales allowed us to attribute the contrasting character of two orthogonal topographic profiles not just to their relationship with the ice-flow direction, but also with the contrasting type of the dominant fractures. The results of this study also potentially enable the prediction of the character of bedrock structures underlying topographic depressions and may therefore aid in focusing investigations associated with infrastructure projects.