Geomorphology 40 Ž2001. 145–161 www.elsevier.comrlocatergeomorph

Development of joint-controlled rock basins in Bohus granite, SW Sweden Magnus Johansson a,) , Piotr Migon b, Mats Olvmo c a

b

Department of Earth Science and Geography, Karlstad UniÕersity, 651-88 Karlstad, Sweden Department of Geography, UniÕersity of Wroclaw, pl. Uniwersytecki 1, 50-137 Wroclaw, Poland c Department of Physical Geography, Goteborg UniÕersity, Box 460, 405-30 Goteborg, Sweden ¨ ¨ Received 8 March 2000; received in revised form 17 January 2001; accepted 28 January 2001

Abstract The granite area of Bohuslan ¨ along the Swedish west-coast is characterised by an abundance of rock basins of different sizes. Within 163 km2 large part of the area over 400 basins occur, ranging in size from less than 0.005 to almost 2 km2. These basins were measured, described and classified from air photos according to their length, width, area, altitude, shape, boundaries, openness and relation to joint configuration. The depth of basins was collected from digital elevation data. Their distributions and shapes are clearly associated with the visible jointing patterns, as shown by their location along joints or joint intersections, by joint-guided enlargement and similarity between directions of long axis of basins and principal tectonic lineaments in the Bohus granite. The patterns and characteristics of the basins suggest that deep selective weathering had been the primary process involved in the origin and evolution of the basins. Later, they have been reshaped and possibly stripped of saprolites by glacial processes in the Pleistocene. Dating of deep weathering responsible for basin initiation and origin is difficult, although the complex story of Mesozoic weathering, Late Cretaceous burial and Late Tertiary exhumation is not without support. q 2001 Published by Elsevier Science B.V. Keywords: Rock basin; Granite; Deep weathering; Etching; Joint control

1. Introduction Granite landscapes have a long history of geomorphologic inquiry focused on the identification of structural influences upon relief development. Much of this research has been summarised by Twidale

) Corresponding author. Tel.: q46-54-700-18-82; fax: q46-54700-14-62. E-mail address: [email protected] ŽM. Johansson..

Ž1982. and Gerrard Ž1988., and by Rudberg Ž1973. concerning the Nordic literature. These reviews show, however, that a marked discrepancy exists in our knowledge of controls on the evolution of positive Žhills. and negative landforms Žbasins., with the latter being a rather neglected subject. While various relationships between lithology and structure, and inselberg landscapes and tors have been recognised, including association of hills and tors with massive rock compartments Žcf. Linton, 1955; Thomas, 1965; Selby, 1982. or with potassium-rich granite Žcf. Pye, 1986., or with coarse-grained granite Žcf. Lageat and

0169-555Xr01r$ - see front matter q 2001 Published by Elsevier Science B.V. PII: S 0 1 6 9 - 5 5 5 X Ž 0 1 . 0 0 0 4 2 - 3

146

M. Johansson et al.r Geomorphology 40 (2001) 145–161

Robb, 1984., much less is known about basin development. Detailed studies are rare and, significantly, no separate chapter is devoted to large and medium size basins in Twidale’s Ž1982. summary of granite landscapes. A partial explanation of this imbalance is the fundamental difficulty to assess properly the influence of rock control upon the development of erosional landforms, which is present at all scales, from weathering pits to large intramontane basins. One of the earlier studies on structure-controlled basins, particularly worth mentioning, is that by Thorp Ž1967. from Nigeria. He showed that fracture intersections are favourable places for basins to develop, whereas valleys evolve along master joints. According to Thorp Ž1967., selective deep weathering adjusted to fracture patterns would be the major preparatory process in basin evolution. Later on, the importance of basins has been emphasised by Thomas Ž1974., who distinguished multi-concave Žbasin-andtor. landscapes in his classification of granite landform systems. However, Thomas is rather vague as far as the origin of basins is concerned, saying only that A . . . wbasinsx appear to be influenced by the properties and structure of the graniteB Žp. 22.. Further examples of joint-controlled basins may be quoted from Connemara, Ireland ŽCoude, 1985., and Slavkovsky´ les, Czech Republic ŽIvan, 1982., but both papers lack supportive data to demonstrate joint influence on morphology and the basins in both areas are few anyway. Consequently, major gaps in the existing knowledge regarding the understanding of the basins are the following. Fracture control is occasionally inferred but difficult to prove and often pointed out at a very general level. Processes involved in the initiation and excavation of basins and directions of basin development over time are poorly known. Selecting the coastal area north of Goteborg in SW Sweden ¨ ŽBohuslan ¨ ., where basins are abundant ŽFig. 1., as the study area, we hope to enlarge our understanding of medium size concave landforms within crystalline areas as far as their geological control, origin, and development are concerned. Given the regional context of the basins, outlined below, we also aim to discriminate between glacial and non-glacial component in rock basin development and to demonstrate the degree of rock control upon morphology in the formerly glaciated area.

2. Study area The study area is located ca. 130 km north of Goteborg in SW Sweden and presents a bare granite ¨ upland at the altitudes below 140 m a.s.l., into which a large number of basins of different size, shape and location are incised. The area is lithologically fairly homogeneous and built of the Bohus granite ŽFig. 2.. The Bohus granite constitutes a number of monzogranitic intrusions of slightly different petrography and chemistry ŽEliasson, 1993.. The intrusions occurred during the final stages of the Sveconorwegian orogeny, 920 Ma ago ŽEliasson and Schoberg, 1991., ¨ and into mainly supracrustal gneisses with an age of ˚ ¨ and Daly, 1989.. Structural fracture 1760 Ma ŽAhall patterns are due to tectonic development and closely connected to the Caledonian, Variscan and Alpine orogenic phases and enclosed compressional forces within the Teisseyre–Tornquist zone ŽHillefors et al., 1993; Zheng, 1996., which constitutes the SW border of the East European Platform. Three sets of parallel lineaments and joints can be recognised in the southern part of the Bohus granite area; N–S, NNE–SSW and WNW–ESE ŽLjungner, 1927; Asklund, 1947; Zheng, 1996. Žcf. Fig. 1.. Minor deviations in these strikes occur for the sets in the northern part of the Bohus granite. The area is subdivided into fault blocks, and the Sotenaset ¨ peninsula ŽFig. 2. constitutes such a fault block, dipping slightly Ž0.6%. towards SW ŽJohansson, 1999.. The general morphology of the study area is characterised by plateau-like granite massifs and asymmetric or dome-shaped granite hills, bounded by high cliffs or scarps, interspersed with flat-floored basins, often with an infill of Late Weichselian glaciomarine silt and clay ŽFig. 3.. The Quaternary cover is up to 40-m-thick within the basin-like forms, according to soil-depth data from wells provided by the Swedish Geological Survey, but is thin or more frequently absent on top of the hills. The relative relief reaches a maximum of 135 m ŽJohansson, 1999.. A feature to note is the general accordance of summits of many of the granite hills and the general tilt of the projected summit surface towards the W and SW. This accordance has been interpreted as the remnant of the former level of the sub-Cambrian

M. Johansson et al.r Geomorphology 40 (2001) 145–161

147

likened basins of Bohuslan ¨ with valleys in eastern Skane and Blekinge, inferring similar origin for these concave landforms. The etching hypothesis as applied for Bohuslan ¨ has recently received further corroboration from the documentation of narrow fissures with relicts of clayey saprolites ŽLidmar-Bergstrom ¨ et al., 1999; Johansson et al., 2000. and numerous small scale etch forms, including corestones, flared slopes and deep narrow clefts in granite up to 20-m-deep and 1–1.5-m-wide ŽOlvmo et al., 1999.. The area has also been subjected to several glaciations during the Pleistocene ŽKleman et al., 1997., and glacial landforms of small size, e.g. roches moutonnees, ´ truncated hilltops and stepped lee-sides of hills due to plucking, striae and minor p-forms are ubiquitous ŽLjungner, 1930; Johnsson, 1956; Olvmo et al., 1999.. Most of these landmarks have been left by the Late Weichselian ice flow that was advancing from NE to ENE ŽLjungner, 1930.; older possible ice Fig. 1. Air-photo Ž1:30 000. of the central parts of the Sotenaset ¨ peninsula. The structural framework of the Bohus granite is well exploited by exogenic processes. The basin floors are covered by Late Weichselian glaciomarine silts and clays.

peneplain, well preserved more inland in SW Sweden ŽJohansson, 1999.. This surface is regarded as the primary palaeosurface within the Precambrian basement, at the expense of which, all younger relief of the Baltic Shield in southern Sweden has evolved during the Phanerozoic, but within different time spans ŽElvhage and Lidmar-Bergstrom, ¨ 1987.. Consequently, Lidmar-Bergstrom ¨ Ž1995. considers Bohuslan ¨ as an example of etched, totally stripped ‘joint valley landscape’ that was developing from the sub-Cambrian peneplain in the Mesozoic, chiefly in the Jurassic and Cretaceous, and then buried under Late Cretaceous sediments. Although no Cretaceous cover rocks have been found onshore, there is a continuous Mesozoic cover ca. 50 km off the coast ŽJensen and Michelsen, 1992; Norling, 1994., beneath which the crystalline basement dips. Furthermore, Lidmar-Bergstrom ¨ Ž1995. emphasises the similarity between the landscape of Bohuslan ¨ and a well dated exhumed sub-Cretaceous joint valley landscape in southeastern Sweden. Apparently, she

Fig. 2. Simplified geological map after Eliasson and Schoberg ¨ Ž1991.. Ž1. Tonalitic–granodioritic orthogneisses. Ž2. Supracrustal gneisses with screens of orthogneisses. Ž3. Bohus granite Žsee also the black area in the over-view map.. Ž4. Bohus granite rich in gneissic xenoliths. ŽS. The peninsula Sotenaset. Basins were ¨ described and measured within the marked area around Sotenaset. ¨

148

M. Johansson et al.r Geomorphology 40 (2001) 145–161

Fig. 3. Photo of basins and hills in the coastal joint valley landscape north of Goteborg. The photo was taken from the peninsula Sotenaset ¨ ¨ northwards across the fjord Bottnefjorden.

flows were from between N and NW ŽKleman et al., 1997.. Overdeepening of some sea inlets has been recognised, e.g. the Gulmarn fjord ŽFloden, ´ 1973; Hillefors et al., 1993..

3. Methods In this study, basins have been defined as concave and at least partially enclosed landforms incised into the general upland surface of Bohuslan. ¨ To discriminate between basins and valley-like forms, which are concavities with a very strong directional component, a lengthrwidth ŽLrW. ratio of 10 has been taken as an arbitrary boundary value, unless the width is larger than 1 km. For the same reason, small basins with long axis shorter than 1 km have been excluded if the LrW ratio is above 5. Although the presence or absence of a stream is not decisive, the majority of basins are not occupied by streams of any larger size, hence, present-day fluvial action seems negligible in basin development. Furthermore, we set another somewhat arbitrary lower size limit for a basin to be considered as such in this study, i.e. 120 and 90 m for length and width, respectively. This translates into 4- and 3-mm-long lines on air photos we used, which is the effective limit of safe identification of landforms. Smaller forms were difficult to identify because of building vegetation up in

some areas. Narrow depressions interconnecting the basins were also excluded from the data set. Basin boundaries in the analysis are represented by the floorrslope intersection, hence, our data in fact refer to basin floors. Such a procedure is justified by the fact that the basins are bounded by very steep, convex upward rock slopes, often vertical towards the base, and the footslopes are much easier to define than any slopercrest intersection. Moreover, given the steepness of marginal slopes, their outline below the infill is likely to be similar to one at the present-day surface. The basins were measured and descriptively classified from air photos Žblack and white, 1:30 000., according to several criteria, all with the purpose of contributing information about how structural control of basins can be manifested. Altogether, 434 basins have been used in this study. The existence of a fracture Žfracture zone. is implied by the occurrence of a straight or slightly curved discontinuity within the granite surface as shown by air photos ŽFig. 1.. Usually, they can be traced from one side of the basin to the other one, hence, we assume their continuation below the basin floor even if the fracture is not visible. Given the resolution of an air photo and geological reality, some of the joints or fractures we are referring to are in fact narrow zones of closely spaced individual discontinuities.

M. Johansson et al.r Geomorphology 40 (2001) 145–161

3.1. QuantitatiÕe data 3.1.1. Length and width of basins The length of long axis of basins and its orientation were measured from the air photos. Some basins Ž26 of 434. have such irregular shapes that no single straight axis could be drawn through them. Instead, two or three broken axes were used, and the sum was

149

taken to represent the length of a basin. The width is defined as the maximum width perpendicular to the long axis.

3.1.2. Orientation The azimuth of the formerly defined long axis has been measured on a grid overlay and expressed in

Fig. 4. Map with basins Ž n s 434. at the Sotenaset ¨ peninsula. The map covers 162.9 km2 and basins constitute 25.6% of this area.

150

M. Johansson et al.r Geomorphology 40 (2001) 145–161

degrees from the north. If an axis is compound, the orientation of each section has been recorded separately and all were included in the analytical treatment. 3.1.3. Area The size of each basin was measured from a map with basins ŽFig. 4., interpreted from the air photos, and following the above-given definition of a basin and its boundaries. 3.1.4. RelatiÕe relief The maximum depth of each basin below the adjacent summit surface was calculated from DEM data Ž50-m grid. by calculating the maximum relative relief for the central pixel within a migratory square of 0.16 km2 . The procedure has been fully described and evaluated in Johansson Ž1999.. The depth of a few very small basins has been calculated from ordinary topographical maps. It needs to be noted that the depth is measured to the topographic basin floor and not necessarily to the rock floor, which may be buried beneath Pleistocene clays. Hence, the relative relief given has to be taken as the minimum. 3.2. DescriptiÕe data 3.2.1. Altitudinal position The relative position of the basins against the regional landscape was determined, and the basins were classified into three groups according to the location of their floors. The basins can be located on the summit surface ŽAuplandB ., have their floors near sea level ŽAlowlandB . or be positioned in between ŽAintermediateB .. Illustrations of the descriptive classification criteria mentioned below are found in Fig. 5. 3.2.2. Openness The basins were classified according to whether they are open and interconnected with adjacent basins or closed, having no or a very narrow outlet, less than 1r5 of basin width. 3.2.3. Shape Five different shapes of basins were named after an analysis of the boundaries as defined above:

Fig. 5. Illustrations of the descriptive classification criteria. The different classes within the criteria openness, shape, boundaries and relation to joints are illustrated as principal sketches of the basin border as seen from above. The sketches of relief of basin floor contain dashed lines representing the unknown features of the bedrock surface beneath the late Weichselian clay.

circular, rectangular, triangular, elongated and irregular. Trying to avoid overlap between classes, these are described as follows. Circular basins have circular perimeter and their lengths, measured in different directions, do not differ by more than 10%. Rectangular basins are defined by four straight lines, parallel in pairs. Triangular basins have three straight boundaries. Elongated basins differ from the rectangular ones by having no clear straight lines on either end, which are rather funnel-shaped; perimeters are usually curved lines. The remaining basins have been classified as irregular. 3.2.4. Boundaries The boundaries of the basins, as defined before, have been described as straight, sinuous, indented or compound, if opposite boundaries markedly differ in plan. For the purpose of further analysis, the identification of straight boundaries, or straight component within a compound boundary, has been most important, as these boundaries are likely to reflect strong joint control. 3.2.5. Relief of basin floor The majority of basins are filled with Late Weichselian clay and silt, and have thus very flat bottom topography. Nevertheless, some basins contain minor hills and knobs protruding from the floor, and if there are at least few of these, the relief is said to be

M. Johansson et al.r Geomorphology 40 (2001) 145–161

151

rough. The true character of the rock floor of the basins is not known. 3.2.6. Relation to joint configuration The basins were classified according to their relation to major fractures as identified on air photos on adjacent upland surfaces. Six relations within three major groups were identified. These are basins formed along one or more linear fractures, at an intersection of two or more fractures, and those having no evident relation to major fractures. In more detail, the basins can be formed along one joint, along two parallel joints, along crossing parallel joints, at joint intersection or at intersections of several joints.

4. Basin characteristics The analysis of basin characteristics is first considered with respect to individual form attributes, followed by their mutual relationships, and finally by discussion of inferential relationships. 4.1. IndiÕidual characteristics 4.1.1. Size The size of basins varies from as little as 0.003 to 1.86 km2 . Most basins are small, not exceeding 0.05 km2 Ž60% of the population., while there are four basins only which are larger than 1 km2 . These four, however, make 14% of the total basin area, while the 60% of the small ones accounts for a mere 14% of the total basin area. Almost half of the total basin area corresponds to less than 10% of the basins. 4.1.2. Spatial coÕerage Basins are not distributed evenly across SW Bohuslan. ¨ Basin coverage in 3 = 3-km squares varies between 4% and 41% ŽFig. 6., with much higher density in the eastern part. In the two western columns, there is only one case with basin coverage higher than 20%. 4.1.3. Shape and length to width ratio The majority of basins have an irregular shape Ž52%., followed by an elongated one Ž27%.. The other shapes account for 21% of the total population

Fig. 6. Basin coverage and maximum relative relief within 9 km2 squares. The proportion of basin area is calculated for onshore parts within each square.

ŽFig. 7.. LrW ratio varies from 1 to as much as 7. One-fourth of the population has LrW ratio lower than 1.5, while it is G 3 for about one-fifth ŽFig. 8.. Eight percent of the basins are close to isometric, with LrW - 1.2. 4.1.4. Orientation of long axis Orientation of the long axis has been considered for 408 out of 434 basins present in the area, with 26 basins with multiple axial orientation having been excluded. The pattern of long axis orientations for n s 408 ŽFig. 9a. is, however, almost identical to that for the total 434. If the number of basins is considered, the presence of three preferential orientations can be recognised. These are NNE–SSW Ž20– 408., N–S Ž350–108., and NW–SE Ž300–3108., al-

152

M. Johansson et al.r Geomorphology 40 (2001) 145–161

Fig. 7. Ža. Shape of basins. Žb. Relation to joints. Žc. Boundaries of basins. Žd. Relief of basin floor Ž n s 434..

though the latter is less pronounced. The former two, if supplemented by the intermediate class 10–208 account for about 35% of the total, as opposed to theoretical 27.5% if the frequency is equal. Underrepresentation of W–E axes is evident. However, if the sum of length axis of a given orientation ŽFig. 9b. is taken into account, the predominance of NNE–SSW direction is even clearer, while the NW– SE one loses its importance. This means that there are relatively many basins elongated along this direction, but they are rather small.

Fig. 8. Distribution of lengthrwidth ratio in classes of 0.5 Ž ns 434..

4.1.5. Boundaries No evident domination by any class has been found, and all of them are represented in similar proportions ŽFig. 7.. 4.1.6. Floor topography Fifty-one percent of basins are entirely flatfloored, and in another 41% cases, isolated hills of various heights protrude from flat floors ŽFig. 7.. Less than 10% of the basins have rough bottom topography. However, flatness of most basin floors is not the result of lateral erosion in bedrock but due to accumulation of glaciomarine clays; hence, interpretation of this part of the data set has to be exercised with special care. The percentage of basins with rough floor relief could be much higher. 4.1.7. Relation to joint configuration Most basins are associated with closely located joint intersections, as almost half of the population is crossed by more than two joints. This is followed by their location along a joint, or at joint intersection. Only less than 10% does not show any relation to

M. Johansson et al.r Geomorphology 40 (2001) 145–161

153

1990. and delimits the peninsula Sotenaset ¨ to the north ŽFig. 4.. 4.2.2. Basin size in relation to joint configuration (Fig. 11) The most obvious characteristic is that while various types of relations to fractures can be recognised for small basins Ž- 0.05 km2 ., larger basins tend to be associated with the class of multiple fracture intersections, with the biggest ones Ž) 0.25 km2 . located almost entirely in such places. No clear relation to fracture patterns can be seen only for the sub-population of the smallest basins ŽFig. 11.. 4.2.3. Basin size Õs. location Upland basins are the smallest Žmean size 0.023 km2 ., the intermediate ones being three times larger Žmean size 0.078 km2 ., and the lowland ones nine times larger Ž0.19 km2 . ŽTable 1.. Mean depth for these three classes consistently increases from 24 m for upland basins through 36–44 m for lowland basins. Lowland basins are more elongated than the intermediate ones Ž2.5 as compared to 2.17., and the upland basins are most isometric, with mean LrW being 1.95 ŽTable 1.. Fig. 9. Ža. Long axis orientations. Žb. Sum of long axis lengths. Ž ns 408.. Intervals of 108 are used in both cases.

4.2. Relationships between characteristics

4.2.4. Shape and relation to joint configuration (Fig. 12a) The correlation between the shape of basins and their relation to fractures shows that the shape has a tendency to be controlled by the configuration of major fractures ŽFig. 12a.. Basins formed along a single fracture tend to be elongated, while basins

4.2.1. Basin size and long axis orientations (Fig. 10) Fig. 10 shows the sum of basin sizes within 108 classes of long axis orientation of the basins and reveals that the major part of the basin area is stretched in the same direction as the three dominating joint sets in the Bohus granite; N–S, NNE–SSW and WNW–ESE. Basins with their long axis in the 350–408 sector constitute 54.2% of the total basin area Žfor n s 408.. There is one obvious exception from this pattern. Basins with long axis oriented ENE–WSW are quite few ŽFig. 9a., but their total size is remarkably large. This anomaly is caused by the giant basins within the prominent fracture zone that strikes ENE–WSW ŽSamuelsson and Lundqvist,

Fig. 10. The sum of basin sizes within long axis orientations sorted in intervals of 108 Ž ns 408..

jointing pattern, which means not crossed or bounded by any joint line ŽFig. 7..

M. Johansson et al.r Geomorphology 40 (2001) 145–161

154

Fig. 11. Distribution of basins according to their size and the frequency within each class of their relation to joints Ž n s 434..

along two parallel fractures are elongated or irregular. Rectangular shape dominates if basins occur along crossing parallels. Basins that develop at joint intersections are mostly irregular. Basins with no evident relationship to major joints are irregular or circular. 4.2.5. L r W ratio Õs. basin size (Fig. 13) LrW ratios plotted against basin sizes are shown in Fig. 13a. There is no trend with high r 2 value in the plot, but the values 0.4 km2 in size and LrW ratio of 3 are worth emphasising. All basins larger Table 1 Mean values of lengthrwidth ratio, basin size and relative relief for upland-, intermediate- and lowland basins

Upland basins Intermediate basins Lowland basins

LrW ratio

Size Žkm2 .

Relative relief Žm.

1.95 2.17 2.51

0.023 0.078 0.19

24 36 44

than 0.4 km2 , except one, have LrW index below 3. The majority of points, accounting for 77% of the total, clusters within LrW range 1–3 and size range 0–0.4 km2 . 4.2.6. L r W ratio Õs. orientation The rose diagram with orientations of long axis for basins with LrW ) 3 ŽFig. 13b. shows, in comparison with the rose diagram for the whole population ŽFig. 9., that the N–S striking long axis become more frequent than NNE–SSW and WNW–ESE if LrW increases above 3. Only a small part of the population Ž4.1%. is bigger than 0.4 km2 , but they constitute 32.4% of the total basin area. All but one of these big basins have LrW ratio below 3. 4.2.7. Boundaries and relation to joint configuration (Fig. 12b) The boundaries of basins that have formed along one joint or along two parallel joints are mostly straight. There is an almost equal number of basins with straight, smooth and mixed boundaries for basins

M. Johansson et al.r Geomorphology 40 (2001) 145–161

155

Fig. 12. Ža. Shape of basins and their relation to master joints. Žb. Boundaries and relation to joints Ž n s 434..

developed at joint intersections. Mixed, indented and smooth boundaries dominate for basins related to intersections of several joints. Indented boundaries are quite rare in all types of basins, except for basins at multiple fracture intersections, where indented cases form 29% of the sub-population.

4.2.8. RelatiÕe altitude position, openness and size of basins (Fig. 14) The majority of the basins have their floors at an intermediate level Ž60.4%., in between upland Ž16.6%. and lowland altitudes Ž23%.. Of the whole population, 28% of the basins are closed. The pro-

156

M. Johansson et al.r Geomorphology 40 (2001) 145–161

the size andror depth of a basin increases with time, although we are not in the position to infer rates of enlargement. The other assumption is tied to the concept well established in the regional geomorphology of Sweden that landscapes of differential erosion have developed at the expense of the ‘primary’ sub-Cambrian peneplain ŽLidmar-Bergstrom ¨ 1995.. It has been demonstrated that the ‘summit level’ in Bohuslan ¨ lies only slightly below the level of the sub-Cambrian peneplain ŽJohansson, 1999., hence it is likely that basins are initiated within the upland surface and may with time become ‘lowland’ basins as defined earlier in this paper. 5.1. Joints and joint intersections as points of basin initiation

Fig. 13. Ža. LrW ratio plotted against basin size Ž ns 434.. Žb. Long axis orientations of basins with LrW ratio above 3 Ž ns82..

portion of closed basins decreases from uplands, where they form 72% of the group, via intermediate settings Ž24%. to lowlands Ž7%.. There is marked discrepancy between the number and the total area occupied by basins in each altitude position. Although upland basins form 16.6% of the population, they cover only 4% of the total basin area. On the other hand, 23% of lowland basins make almost half of the basin area ŽFig. 14..

There is no doubt that basins are preferentially located along fractures; therefore, a genetic relationship between the existence of joints and basin development may be inferred. It is a mere 10% of the total population that does not show any relation to fractures, and the vast majority of these basins are very small ŽFig. 11.. These basins may be due to subtle petrologic and structural variations within the Bohus granite or localised glacial erosion, especially due to plucking on upland surfaces. The clear manifestation of initial fracture control is the location of most small basins along a single fracture zone or at fracture intersections, whereas very few basins larger than 0.05 km2 are located along just one fracture or

5. Joint control on basin development—an interpretation The interpretation of the data in terms of controls on basin initiation and further development follows two general and simple assumptions. One holds that, in common with other erosional hollow landforms,

Fig. 14. The relative position of altitude and openness of basins. The small circle-diagram shows the proportion of total basin area at upland-, intermediate- and lowland levels Ž ns 434..

M. Johansson et al.r Geomorphology 40 (2001) 145–161

at an intersection of two. As basins coalesce, more and more individual fracture zones criss-cross one basin. 5.2. Enhancement of joint control with time The development and enlargement of basins are guided by the joint pattern. Joints are responsible for preferential growth of basins in certain directions, and this is clear from the rose diagram of long axis orientation ŽFig. 9.. All three dominant orientations, i.e. N–S, NNE–SSW and WNW–ESE, are identical with principal tectonic lineaments present in the Bohus granite ŽLjungner, 1927; Zheng, 1996.. More specifically, while quite a significant proportion of small basins has a very low LrW index, - 1.5 ŽFig. 13., such low values are very uncommon for basins bigger than 0.2 km2 . Instead, they tend to be around 2. Also, the very few highly elongated basins ŽLrW ) 5. are larger than 0.1 km2 . However, elongation does not increase indefinitely, and none of the very big basins show LrW ratio higher than 3. This is either because lateral enlargement along perpendicular joints starts to act concurrently with exploitation of a master joint or the result of coalescence of number of basins adjusted to joints of different strike. Indented boundaries, with indentations following joints, are also taken as examples of basin development along lines of structural weakness. Moreover, increase in LrW ratio from upland basins to lowland basins shows directed growth in certain directions, facilitated by the existence of major joints. 5.3. Master joints as basin boundaries It is difficult to explain the rather common occurrence of straight boundaries of basins by factors other than joint guidance Žrecent faulting seems to be ruled out in the case of Bohuslan ¨ .. The majority of basins developed along a single joint has either one or both sides straight, and so do basins bounded on two joints parallel to each other. Master joints often separate rheologically different rock masses, hence, most geomorphic activity is concentrated on the weaker side of a joint or within inter-joint area.

157

5.4. Increase in basin integration and relatiÕe relief If we compare distribution of open and closed basins over space, it becomes evident that the degree of integration increases from upland to lowland. The 10-fold increase is best explained by simultaneous growth of adjacent basins in opposite directions towards each other, their coalescence and the progressive widening of narrow joint-guided defiles. Similarly, mean depth of basins increases from upland to lowlands.

6. Origin and age of basins in the Bohus granite The early debate about the origin of joint valley landscapes in Sweden has been summarised by Rudberg Ž1973., who introduces the general hypothesis proposing a sequence of events starting with ‘preglacial’ weathering along zones of weakness, followed by partial fluvial evacuation of weathered rock and a final evacuation by glacial erosion in the Pleistocene. While discussing glacial erosion as a possible mechanism for basin development in south-western Sweden, he also draws attention to the similarity between a glaciated granite coast and the non-glaciated granite coast of Brittany. In conclusion, Rudberg states that the landform uniformity between these areas might simply be due to the substantial difference in resistance between densely jointed parts of the bedrock and parts with low fracture density, thus, pointing clearly towards equifinality. However, since Rudberg, much evidence has accumulated that confirms the very important role of deep weathering Žetching. in the development of various facets of Swedish topography ŽLidmarBergstrom, ¨ 1995., including joint-guided valleys and basins as in southern Sweden. Moreover, it has been shown that different rock-cut topographies are of different ages and usually record a complex story of exposure, erosion, burial and exhumation ŽLidmarBergstrom, ¨ 1989.. Finally, in spite of numerous glaciations, the erosional imprint of ice sheets has been highly variable over space, including large tracts of terrains with evidence of glacial action virtually lacking or confined to minor features such

158

M. Johansson et al.r Geomorphology 40 (2001) 145–161

as potholes or striae ŽLidmar-Bergstrom, ¨ 1989, 1997; Johansson, 2000.. Consequently, the debate about the origin of rock-cut basins will inevitably focus on the recognition of glacial and non-glacial component, especially in Bohuslan, ¨ where the evidence of glacial erosion is restricted to small-scale features ŽLjungner, 1930; Hillefors et al., 1993; Olvmo et al., 1999; Johansson, 2000.. Several studies demonstrate that various small to medium scale glacial erosion forms are highly controlled by bedrock structure Že.g. Ljungner, 1930; Johnsson, 1956; Rudberg, 1973; Gordon, 1981; Rastas and Seppala, ¨ ¨ 1981; Glasser and Warren, 1990; Sugden et al., 1992.. It is thus reasonable to assume that basins might indeed form by simple glacial erosion along pre-existing zones of weakness . However, bedrock geomorphology of southwestern Sweden, including the pattern of basin distribution in Bohuslan ¨ is difficult to relate to any of the former glacial flow patterns. There is a clear under-representation of ENE long axis of basins, yet this was the direction of the Late Weichselian ice flow, which has left ubiquitous erosion forms of medium and small size ŽLjungner, 1930.. Directions of older possible ice flows from between N and NW ŽKleman et al., 1997. are similarly not entirely compatible with dominant basin orientation. On the other hand, the proximity of Cretaceous cover rocks in Skagerrak, the frequent occurrence of kaolinised fracture zones within the Bohus granite and a relief type similar to exhumed sub-Cretaceous relief in south-eastern Sweden, suggest that the gross morphology of Bohuslan ¨ is an exhumed sub-Mesozoic etch surface formed by selective deep weathering along fracture zones ŽLidmar-Bergstrom, ¨ 1988, 1995; Lidmar-Bergstrom ¨ et al., 1999.. Widespread distribution of basins suggests that there must have been many places of basin initiation and their concurrent development with each other. Weathering, operating across the entire area, explains basin distribution patterns better than glacial erosion, where a trend towards a stronger preference of orientation and a limitation to size and shape variability would perhaps be expected. Furthermore, it is difficult to visualise how eroding ice streams could have initially formed a maze-like pattern of basins, and the same applies to fluvial erosion. Basin distribution is hardly consistent with any kind of

integrated drainage pattern and, indeed, no such drainage pattern exists at present. Another line of evidence follows Mattsson’s Ž1962. remarks that preglacial saprolite remnants occur in fissures of the Bohus granite. Although he did not present any inventory or spatial distribution of the weathering remnants, the authors have recently found seven new sites of grus and clayey saprolites with smectite and kaolinite present within the Sotenaset ¨ peninsula, all exposed in road cuttings and granite-quarries ŽJohansson et al., 2000.. They occur in decimeter-wide fissures of unknown depth and their frequent occurrence confirms that deep weathering has been operating along joints and fracture zones, although at present, the age of this alteration cannot be specified with any certainty. Lidmar-Bergstrom ¨ Ž1995. suggests that joint-valley landscape of Bohuslan ¨ belong to the sub-Cretaceous landscape and remained buried beneath Late Cretaceous sedimentary rocks until late in the Tertiary, hence, the bulk of deep weathering would be Mesozoic. However, there is no reason to exclude the possibility of weathering renewal in the Tertiary, following re-exposure. As Late Tertiary weathering residuals in Scandinavia are of grus type ŽLidmarBergstrom ¨ et al., 1999., it is possible that in Bohuslan, ¨ we are dealing with two generations of saprolite remnants, the Mesozoic and the Tertiary ones. A deep weathering hypothesis may also assist in explanation of some other features of basins. Master joints often form basin boundaries. They may have separated areas of different densities of secondary joints, with an area of higher density being etched out much faster. This is, however, impossible to prove because any evidence of such densely fractured compartments has been lost due to erosion, although air photos clearly show contrasting joint densities within upland surfaces at present. With differential etching in progress, the depressions have had more regolith, which in turn retained more water from the surroundings. More active weathering attack may then have taken place, and all dimensions of depressions increased. Within some basins, central residual elevations up to 45 m high are present. They demonstrate that such basins have not developed from a single depression, but perhaps are remnants of rock compartments that once existed

M. Johansson et al.r Geomorphology 40 (2001) 145–161

between two or more adjacent linear depressions. If exposed, the rock surfaces of the threshold will have become relatively resistant to weathering because in contrast to regolith within the basin, they remain dry. Also, side attack will have been less intense because little water will flow off these spatially restricted surfaces, hence, moisture of footslopes may be insufficient. This concept is similar to ‘divergent weathering’ and the model of basin evolution as envisaged by Bremer Ž1975. and could be applied to explain larger dimensions and greater depths of lowland basins in relation to the upland ones. Although deep joint-guided etching has been very important in the development of basins, the glacial component in basin development must not be overlooked. Certainly, ice streams and meltwater contributed to modification of basins and their surrounding slopes through abrasion, plucking and potholing. It is also possible that basins with long-axis orientation in accordance with ice flows have received an increase in depth and width, as the fact of overdeepening may indicate. Future use of seismic soundings may contribute more information. Finally, it is most likely that glacial erosion and meltwater erosion evacuated most of the preglacial regolith from the basins in the Quaternary. In summary, we propose that basins had initially been formed by selective, joint-guided weathering, and after that, partially reshaped by glacial erosion. At each stage, joints provided available lines of weakness, along which, destructive action of weathering and erosion could work most effectively.

7. Concluding remarks Rock-cut basins are a rather neglected subject in structural geomorphology, although they may be fairly common in crystalline upland terrains. Most often, they are perceived as landforms of selective denudation acting upon heterogeneous bedrock Žcf. Thorp, 1967; Hall, 1991; Godard et al., 1994. and their occurrence has been ascribed to differential deep weathering, but reasons for spatial variations in the intensity of weathering remain unclear Žcf. Bremer, 1975; Jahn, 1980; Demoulin, 1995.. Joint control and petrovariance are usually inferred.

159

In the granite area of Bohuslan, ¨ there exists an abundance of rock-cut basins of various sizes, whose form, location and inferred development patterns are clearly controlled by characteristics of the jointing patterns. This is shown by preferential location of small basins along joints or joint intersections, jointguided enlargement and similarity between directions of long axis of basins and principal regional joint directions. Various processes appear to have been involved in the development of the basins, with selective etching and subsequent stripping being critical. Unfortunately, at present, we cannot date the etching with any greater precision other than simply saying it was preglacial, although the story of Mesozoic age, Late Cretaceous burial and Late Tertiary exhumation is not without support. Although stripping may have been initiated well prior to the Quaternary, the Pleistocene glacial erosion has been important in altering etch characteristics of the basins and may have contributed to ultimate evacuation of saprolites.

Acknowledgements PM wishes to thank the Department of Physical Geography, Goteborg University, for financial sup¨ port that enabled him to participate in this project. He also extends his thanks to Karna LidmarBergstrom ¨ ŽStockholm. for the introduction to the preglacial morphology of Sweden during a field trip in 1997. Karna has also critically read the early draft of this paper and offered help at various stages of this work. Both referees of the journal are thanked for their perceptive and most accurate comments on the early version of this paper.

References ˚ ¨ K.I., Daly, J.S., 1989. Age, tectonic setting and provenance Ahall, ¨ of Ostfold–Marstrand Belt Supracrustals: westward crustal growth of the Baltic Shield at 1760 Ma. Precambrian Res. 45, 45–61. Asklund, B., 1947. Gatsten och kantsten. Sver. Geol. Unders., Ser C. 479, 187 pp. Bremer, H., 1975. Intramontane ebenen, prozesse der flachenbil¨ dung. Z. Geomorphol., Suppl. Bd. 23, 26–48.

160

M. Johansson et al.r Geomorphology 40 (2001) 145–161

Coude, A., 1985. L’alveole ´ granitique de la Baie de Bertraghboy, Connemara ŽIrlande.. Physio-Geo. ´ 13, 11–18. Demoulin, A., 1995. L’Ardenne des plateaux, heritage des temps anciens: surfaces d’erosion en Ardenne. In: Demoulin, A. ŽEd.., L’Ardenne: Essai de Geographie Physique. Universite´ de Liege, Liege, pp. 68–93. Eliasson, T., 1993. Mineralogy, geochemistry and petrophysics of red coloured granite adjacent to fractures. SKB Tech. Rep. 93-06, 68 pp. Eliasson, T., Schoberg, H., 1991. U–Pb dating of the post-kine¨ matic Sveconorwegian ŽGrenvillian. Bohus granite, SW Sweden: evidence of restitic zircon. Precambrian Res. 51, 337–350. Elvhage, C., Lidmar-Bergstrom, ¨ K., 1987. Some working hypotheses on the geomorphology of Sweden in the light of a new relief map. Geogr. Ann. 69A, 343–358. Floden, ´ T., 1973. Notes on the bedrock of the eastern Skagerrak with remarks on the Pleistocene deposites. Acta Univ. Stockholm Contrib. Geol. XXIV, 5. Gerrard, J., 1988. Rocks and Landforms. Unwin and Hyman, London, 319 pp. Glasser, N.F., Warren, C.R., 1990. Medium scale landforms of glacial erosion in south Greenland; process and form. Geogr. Ann. 72A, 211–215. Godard, A., Lagasquie, J.-J., Lageat, Y. ŽEds.., 1994. Les Regions ´ de socle: Apports d’une ecole franc¸aise de Geomorphologie. ´ Faculte´ des Lettres et Sciences humaines de l’Universite´ Blaise-Pascal, Nouvelle serie, ´ Fasc., vol. 43, 324 pp. Gordon, J.E., 1981. Ice-scoured topography and its relationships to bedrock structure and ice movement in parts of northern Scotland and west Greenland. Geogr. Ann. 63A, 55–65. Hall, A.M., 1991. Pre-Quaternary landscape evolution in the Scottish Highlands. Trans. R. Soc. Edinburgh: Earth Sci. 82 1–26. ˚ Floden, ˚ Hillefors, A., P., 1993. Abyfjorden on the ´ T., Soderberg, ¨ Swedish west coast; morphology, superficial deposites and gas-bearing clays. Geol. Foeren. Stockholm Foerh. 115, 241– 261. Ivan, A., 1982. Nektere ˆ ´ morfostrukturnı´ rysy reliefu ´ na granitoidech Ceskeho ´ Masıvu. ´ Geomorfologicka´ Konference. Univerzita Karlova, Praha, pp. 53–60. Jahn, A., 1980. Main features and age of the relief of the Sudety Mountains. Geogr. Pol. 43, 5–23. Jensen, L.N., Michelsen, O., 1992. Tertiær hævning og erosion i Skagerrak, Nordjylland og Kattegatt. Dansk Geologisk Foren˚ ing Arsskrift for 1990–91, pp. 159–168. Johansson, M., 1999. Analysis of digital elevation data for palaeosurfaces in southwestern Sweden. Geomorphology 26, 279– 295. Johansson, M., 2000. The role of tectonics, structures and etch processes for the present relief in glaciated Precambrian basement rocks of SW Sweden. Department of Earth Sciences, Goteborg University, Dissertation, A53. ¨ Johansson, M., Eliasson, T., Olsson, S., 2000. Mineralogy and geochemistry of in situ weathered Bohus granite in south-west Sweden; implications for landform evolution. Abstracts, Weathering 2K Conference. School of Geography, Queen’s University, Belfast, UK.

Johnsson, G., 1956. Glacialmorfologiska studier i sodra ¨ Sverige. Medd. Lunds Univ. Geogr. Inst. Avh. 30, 407 pp. Kleman, J., Hattestrand, C., Borgstrom, ¨ ¨ I., Stroeven, A., 1997. Fennoscandian palaeoglaciology reconstructed using a glacial geological inversion model. J. Glaciol. 43 Ž144., 283–299. Lidmar-Bergstrom, ¨ K., 1988. Denudation surfaces of a shield area in south Sweden. Geogr. Ann. 70A, 337–350. Lidmar-Bergstrom, ¨ K., 1989. Exhumed Cretaceous landforms in south Sweden. Z. Geomorphol. N.F., Suppl. Bd. 72, 21–40. Lidmar-Bergstrom, ¨ K., 1995. Relief and saprolites through time on the Baltic shield. Geomorphology 12, 45–61. Lidmar-Bergstrom, ¨ K., 1997. A long-term perspective on glacial erosion. Earth Surf. Processes Landforms 22, 297–306. Lidmar-Bergstrom, ¨ K., Olsson, S., Roaldset, E., 1999. Relief features and palaeoweathering remnants in formerly glaciated Scandinavian basement areas. In: Thiry, M. ŽEd.., Palaeoweathering, palaeosurfaces and related continental deposites. Spec. Publ. Int. Assoc. Sedimentol. 27, pp. 275–301. Lageat, Y., Robb, L.S., 1984. The relationships between structural landforms, erosion surfaces and the geology of the Archaean granite basement in the Barberton region, Eastern Transvaal. Trans. Geol. Soc. S. Afr. 87, 141–159. Linton, D.L., 1955. The problem of tors. Geogr. J. 121, 470–487. Ljungner, E., 1927. Spaltentektonik und morphologie der Schwedischen Skagerrak-Kuste: Teil I. Das Grundgebirge des ¨ Gebietes und die post-archaischen Magmengesteine ¨ Westschwedens; Teil II. Spaltentektonik und bruchtopographie. Bull. Geol. Inst. Univ. Uppsala, XXI. Ljungner, E., 1930. Spaltentektonik und morphologie der Schwedischen Skagerrak-Kuste: Teil III. Die erosionsformen. ¨ Bull. Geol. Inst. Univ. Uppsala, XXI. ˚ 1962. Morphologische studien in sudschweden Mattsson, A., und ¨ auf Bornholm uber die nichtglaziale formenwelt der ¨ felsensskulptur. Lund Studies in Geography, Ser. A. Physical Geography 20, Lund, 357 pp. Norling, E., 1994. Kontinentalsockelns berggrund. In: Freden, ´ C. ŽEd.., Sveriges Nationalatlas—Berg och Jord. Bra Bocker, ¨ Hoganas, ¨ ¨ 207 pp. Olvmo, M., Lidmar-Bergstrom, ¨ K., Lindberg, G., 1999. The glacial impact on an exhumed sub-Mesozoic etchsurface in southwestern Sweden. Ann. Glaciol. 28, 153–160. Pye, K., 1986. Mineralogical and textural controls on the weathering of granitoid rocks. Catena 13, 47–57. Rastas, J., Seppala, ¨ ¨ M., 1981. Rock jointing and abrasion forms on roches moutonnees, ´ SW Finland. Ann. Glaciol. 2, 159–163. Rudberg, S., 1973. Glacial erosion forms of medium size—a discussion based on four Swedish case studies. Z. Geomorphol. Suppl. Bd. 17, 33–48. Samuelsson, L., Lundqvist, I., 1990. Bedrock map of Goteborg ¨ and Bohus County. Sveriges Geologiska Undersokning Ah 12. ¨ Selby, M.J., 1982. Form and origin of some bornhardts of the Namib Desert. Z. Geomorphol. N.F. 26, 1–15. Sugden, D.E., Glasser, N., Clapperton, C.M., 1992. Evolution of large roches moutonnees. ´ Geogr. Ann. 74A, 253–264. Thomas, M.F., 1965. Some aspects of geomorphology of domes and tors in Nigeria. Z. Geomorphol. N.F. 9, 63–82. Thomas, M.F., 1974. Granite landforms: a review of some recur-

M. Johansson et al.r Geomorphology 40 (2001) 145–161 rent problems of interpretation. In: Brown, E.H., Waters, R.S. ŽEds.., Progress in Geomorphology. Inst. Brit. Geogr., Spec. Publ. 7, pp. 13–37. Thorp, M.B., 1967. Closed basins in Younger Granite Massifs, northern Nigeria. Z. Geomorphol. N.F. 11, 459–480.

161

Twidale, C.R., 1982. Granite Landforms. Elsevier, Amsterdam, 372 pp. Zheng, F., 1996. Tectonic development of the Bohus Granite ŽSW Sweden. and its adjoining areas. Stockholm Contrib. Geol. 44, 208 pp.