Ground penetrating radar images of selected fluvial deposits in the Netherlands

Ground penetrating radar images of selected fluvial deposits in the Netherlands

ELSEVIER Sedimentary Geology 128 (1999) 245–270 Ground penetrating radar images of selected fluvial deposits in the Netherlands J. Vandenberghe a,Ł ...

4MB Sizes 0 Downloads 29 Views

ELSEVIER

Sedimentary Geology 128 (1999) 245–270

Ground penetrating radar images of selected fluvial deposits in the Netherlands J. Vandenberghe a,Ł , R.A. van Overmeeren b b

a Vrije Universiteit Amsterdam, Faculty of Earth Sciences, De Boelelaan 1085, 1081 HV Amsterdam, Netherlands Netherlands Institute of Applied Geoscience TNO, National Geological Survey, P.O. Box 80 015, 3508 TA Utrecht, Netherlands

Received 9 October 1998; accepted 16 June 1999

Abstract Ground penetrating radar (GPR) surveys have been carried out in order to characterise reflection patterns and to assess the method’s potential for imaging palaeofluvial sediments in the Maas–Rhine former confluence area in the southern Netherlands. The results show that the deposits of meandering, braided and transitional river types produce characteristic radar facies. Representative examples of each of these river types were selected where the GPR data could be directly correlated with sedimentary information derived from exposures and detailed drilling, and geomorphological data could be supplied. Individual channels may be distinguished by the radar facies of their fills. The floodplains of the different river types also show a characteristic radar facies. In GPR data from the meander floodplain, parallel, dipping reflections represent point-bar structures, while irregular, intersecting small channel patterns that alternate with parallel continuous reflections are more typical in the braid plain. A transitional river type shows characteristics of both of these types. Typical examples of GPR sections recorded in each of the different fluvial palaeoenvironments are presented in an interpretative radar facies chart.  1999 Elsevier Science B.V. All rights reserved. Keywords: GPR (ground penetrating radar); river patterns; Late Glacial; radar facies; radar stratigraphy; (palaeo)fluvial sedimentary environments

1. Introduction Sedimentological and palaeohydrological reconstructions are often hampered by the discontinuity or scarcity of source data. Near-continuous GPR cross-sections may reveal the complete two- or three-dimensional geometry of the sediment body and enable the quantitative determination of channel depth and width. With improved detection and

Ł Corresponding

[email protected]

author. Fax: C31-20-646-2457; E-mail:

mapping of sedimentary unconformities and bedding types, better characterisation of depositional sequences and determination of their spatial extent is possible. This paper aims to differentiate between specific and well-defined fluvial sedimentary facies and interpret different fluvial subenvironments by using GPR in combination with geomorphological information and data from sedimentary cores. The literature on GPR applications in fluvial geology has grown rapidly in recent years. Van Overmeeren (1998) has given an overview of radar facies from unconsolidated sediments in the Netherlands. Huggenberger (1993) and Huggenberger et al. (1994)

0037-0738/99/$ – see front matter  1999 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 7 - 0 7 3 8 ( 9 9 ) 0 0 0 7 2 - X

246

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

have studied the reflection patterns of fluvial gravel deposits in Switzerland. Radar stratigraphic analysis of coastal and fluvial environments in North America has been reported by Jol et al. (1996). Three-dimensional GPR studies of glaciofluvial sediments have been published by Beres et al. (1995), Green et al. (1995) and Bu¨ker et al. (1996). Gawthorpe et al. (1993) and Bridge et al. (1995) have imaged point bars in meander belts. Bridge (1998) used high-frequency GPR measurements to describe very detailed, but shallow sedimentary structures in braided river deposits. Finally, Stephens (1994) collected GPR data on well exposed consolidated sandstone (Lower Jurassic) showing reflection patterns from laminated flood deposits and concave-up channel deposits. It is known that sedimentary sequences and structures in specific fluvial depositional environments produce characteristic GPR reflection patterns or ‘radar facies’. Similar to seismic facies (Roksandic, 1978; Sangree and Widmier, 1979), radar facies is defined by a range of attributes, such as reflection amplitude, reflection continuity, reflection configuration, external form (geometry), dominant frequency, abundance of reflections, reflection polarity, presence of diffractions, and degree of penetration. Radar stratigraphy aims to recognise characteristic radar facies and to correlate them with specific depositional environments. Ultimately, the GPR measurements should lead to the reconstruction of palaeofluvial architectures. The potential of GPR for detailed characterisation of fluvial environments was explored by way of specific case studies in the Maas River plain (the southern Netherlands). The deposits from the Pleistocene– Holocene transition offer a rich variety of fluvial forms and deposits as shown by detailed geomorphological and sedimentological investigations (Van den Broek and Maarleveld, 1963; Vandenberghe et al., 1994; Kasse et al., 1995; Huisink, 1997). The sites studied are located in the former Maas–Rhine confluence area where, apart from a series of Pleistocene terraces, three Late Glacial terraces are distinguished (Fig. 1). Sedimentary structures and geomorphological mapping demonstrate the existence of both braided and meandering river patterns. In particular, this paper focuses on the recognition of individual channels and channel fills in GPR images, the identification of river sedimentation patterns in floodplains

and the characterisation of one-channel (meandering) and multi-channel (braided, transitional) river deposits. Some of the GPR sections are located above vertical exposures of sediments so that direct and detailed verification was possible. The other GPR sections are verified by detailed drilling. All of the sites and sections were linked to the well-established terrace stratigraphy (ref. op. cit.).

2. GPR data acquisition and processing Most of the GPR sections were recorded with a pulseEKKO 100 system (Davis and Annan, 1989). The 200, 100 or 50 MHz antennas were placed in a cart and towed behind a small 4-wheel drive allterrain vehicle. This vehicle contains a ‘field office’ and travels at the required low velocity of about 2 km=h (Van Overmeeren, 1994). Measurements were triggered at constant spacings by an odometer wheel. Some of the sections near pit walls were recorded by moving the antennas manually. Common mid point (CMP) soundings were made along most GPR sections in order to determine the propagation velocity of the radar waves. Standard data processing was applied to most of the sections, using pulseEKKO software. This included a ‘dewow’ filter to remove low-frequency induction effects and filtering by trace averaging (commonly set to 3) and sample averaging (commonly set to 3). The sections were corrected for topography where necessary. Depending on the targets and the propagation media, two types of gain were applied: spreading and exponential compensation (SEC) and automatic gain control (AGC). SEC attempts to compensate for spherical spreading losses and exponential attenuation of energy. Relative amplitude variations are thereby preserved. SEC gain is useful to highlight the major high-amplitude reflections. AGC increases amplitudes by a factor that is inversely proportional to the signal strength. Amplitude information is thereby lost. AGC is more effective for recognising low-amplitude reflections. AGC was applied to almost all of the presented sections, because sedimentary structures commonly produce only low-amplitude reflections. Some fills of abandoned meandering channels, however, exhibited large permittivity contrasts and produced reflections

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

247

Fig. 1. Survey area in the southern Netherlands (see inset), locations of the GPR sections on the Pleistocene terraces and Holocene floodplain and location of the maps in Fig. 16 (A, B). The terrace configuration is after Vandenberghe et al. (1994) and Kasse et al. (1995).

248

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

with moderate amplitudes. In such cases, a SEC gain was more appropriate (see Section 3.2.2: Lottum). Migration (two-dimensional in the frequency– wavenumber (F–K) domain; Yilmaz, 1987) was applied to several sections. Migration is a processing routine that corrects the dips of reflections and collapses any hyperbolae, which are produced by diffractions. Diffraction hyperbolae are characteristic of GPR data from highly heterogeneous sediments (i.e. braided river deposits). They distort the geological image and make it difficult distinguishing truly dipping reflections from hyperbolic branches. Migration is most helpful for such data and allows interpretation with more confidence. An example is shown in Section 3.1.3: Mill. All GPR sections displayed in this paper have vertical scales measured in nanoseconds (ns) two-way travel time. The corresponding depth (or elevation) scales are based on velocity analyses and are twice the horizontal scales.

3. Characterisation of channels and fluvial depositional patterns The recognition of individual palaeochannels and the determination of their dimensions are necessary for reconstructing the characteristics of the former river system. The channel width=depth ratio is generally related to energy conditions and thus gives information on sediment transport processes (Schumm, 1960). The channel frequency at a given time, as well as the composition of the floodplain sediments exhibit characteristic patterns that are also related to the energy conditions. Observations of such phenomena by GPR images are intended to distinguish between laterally accretionary one-channel rivers, multi-channel rivers and rivers that are intermediate. Each of these three river types is represented in one of the studied Late Glacial terraces and appears distinctly from the mapped surface morphology, so that there is no doubt about the character of the river deposits. Furthermore, the sedimentary characteristics may also be indicative for distinguishing deposits of meandering and braided channels (Williams and Rust, 1969; Allen, 1970; Cant and Walker, 1976; Miall, 1996), although this may not be conclusive in all cases (Bridge, 1985, 1993). The specific argu-

ments on which the interpretation of the river patterns of the different terraces is based are documented in a large number of sediment–petrological publications and reports, with numerous descriptions of exposures and drilling logs, and geomorphological maps (e.g. Pons, 1954; Doppert and Zonneveld, 1955; Van den Broek and Maarleveld, 1963; Boersma et al., 1981; Vandenberghe et al., 1994; Kasse et al., 1995; Van de Berg, 1996; Huisink, 1997). 3.1. Braided rivers 3.1.1. Channels in a sandy–gravelly braided river deposit: the Laumans pit at Tegelen The Pleistocene Maas terrace deposits consist predominantly of sand and gravel and are composed of extensive (sub)horizontal sheets, a few dm thick, and channel fills consisting of trough cross-strata. These deposits are very heterogeneous and contain locally very large blocks (several m3 ). They are the result of temporarily high-energy flow that is characteristic for braided rivers and resemble the deep, gravel-bed to sand-bed braided river deposits in the terminology of Miall (1996). An example from the Maas valley has been described in detail by Vandenberghe et al. (1993). Another example is the Sterksel Formation which consists of gravels and coarse sands deposited in very wide and shallow channels of the braided Rhine River during one of the glacial periods in the mid-Pleistocene (Zonneveld, 1958). These characteristics appear in numerous exposures in the Rhine valley. Some well-defined channels in these deposits have been selected to test the accuracy of imaging them with the GPR method. GPR measurements were made along exposed faces of the Sterksel Formation, in the Laumans pit at Tegelen (Fig. 1). The geological section in Fig. 2 shows two channels about 20 m wide at their top and reaching a depth of 3 to 4 m below the GPR survey level. The channels are filled with trough cross-bedded, medium-grained sand. The surrounding and underlying sediments generally have more planar stratification. The GPR survey line is situated at 2 m distance from the vertical pit face. GPR measurements were made with antenna-frequencies of 50, 100 and 200 MHz. The high resolution of the 200 MHz antennas produced the best image of the channels observed in the pit face (Fig. 3a,b). The depth scale of the

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

Fig. 2. Channels in sandy–gravelly braided river deposits of the Sterksel Formation (Laumans gravel pit, Tegelen). See Fig. 1 for location.

249

250

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

251

Fig. 4. GPR modelling of channels in the braided river deposits (Laumans gravel pit, Fig. 3); (a) GPR F–K model, (b) synthetic GPR section.

GPR section is based on an average velocity of 12 cm=ns, deduced from CMP analysis. The maximum penetration is approximately 6 m. As shown in the interpreted section (Fig. 3c), one of the channels is very well reproduced by trough-shaped reflections between positions 25 and 43 m. The other one is less clearly visible between positions 15 and 28 m. There is a good correlation between the sedimentological record and the dimensions of the trough-shaped

reflections (approximately 20 by 4 m) and the erosional unconformity between the two channels in the GPR section. The trough cross-bedding is well expressed in the radar section and even the intersection between the two gullies fits exactly the observations on the exposed pit face. Several hyperbolic diffractions appear below the channel patterns in Fig. 3. Two-dimensional (2D) modelling in the frequency– wavenumber (F–K) domain with pulseEKKO soft-

Fig. 3. GPR section over channel structures in braided river deposits (Laumans gravel pit); (a) photograph of exposed pit wall, vertical scale 2ð exaggerated, (b) 200 MHz GPR section with trough-shaped reflections, (c) interpretation.

252 J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

Fig. 5. Channel complex from the top of the braided Lower Terrace of the Maas (exposed sedimentary section in Bosscherheide sand pit). For location see Fig. 1.

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

ware (Sensors and Software, 1996) can be very useful in interpreting complex reflection patterns and in identifying the sources of the diffractions. The only model parameter required for this type of modelling is a relative value for the reflectivity of the layer interfaces. A synthetic GPR section (Fig. 4b) produced by a model of the intersecting channels (Fig. 4a) shows a clear resemblance with the recorded section (Fig. 3c). It illustrates that the sources of the hyperbolas are point diffractors (in three dimensions: line diffractors) at intersecting channels, at channel bottoms and at the edge of a horizontal layer eroded by a channel. 3.1.2. Channels in a sandy braided river deposit: the Bosscherheide sand pit The Bosscherheide site is situated on the Lower Terrace of the Maas (Fig. 1). A channel complex dating from the end of the last glacial was exposed in a sand mining pit (Fig. 5). Its braided character has been derived from detailed sedimentary analyses of this and other vertical sections along the Maas (e.g. Kasse et al., 1995; Van de Berg, 1996; Huisink, 1997). A pattern of very shallow, multiple channels separated by bars may be recognised at the present surface of this terrace near Rijkevoort (Fig. 1). In the final stage of fluvial activity on this terrace, at the beginning of the Late Glacial, the channels became inactive (Bohncke et al., 1993). The bottom of this last channel complex corresponds with the boundary between sandy gravels of the older braided system and an overlying series of laminated, fine sands (southwest of point 55 m) or gravelly coarse sands (between positions 40 and 55 m), which represent the fill of the abandoned terminal channels. Although this boundary is not completely exposed, the GPR section in Fig. 6 shows that the channel complex consists of two individual channel forms (see below). The fine-grained portion of the channel fill shows the grain-size distribution and homogeneous structure that are typical of aeolian deposition, whereas the coarse-grained portion indicates deposition in relatively small-scale gullies within the northeastern channel (between positions 48 and 54 m in Fig. 5). Apparently, the river activity decreased considerably during this time and was limited to the northeastern channel, whereas the southwestern part was filled earlier (perhaps by windblown sand). Fi-

253

nally, the activity almost ceased and peat up to 50 cm thick formed in the deepest (northeastern) part. A humic sand or silt was deposited elsewhere in the original channel, and a soil formed on the higher levels outside the channel (Bohncke et al., 1993). At the end of the Late Glacial the last depression was filled by dune sands from 0 to several metres thick. The river system may be described as a ‘shallow, sand-bed braided river’ (Miall, 1996). A 100 MHz GPR section (Fig. 6) was recorded a few metres from the exposed pit face. The pit itself is exploited by dredging below the water table of the open pit. CMP soundings made along the profile yielded an average propagation velocity of 12 cm=ns to the water table. This yielded a depth of about 4 to 5 m, which concurred with the water table of the pit lake. The 100 MHz waves penetrate to approximately 10 m depth, but below the water table the reflection pattern is discontinuous to chaotic and no clear sedimentary images are obtained. The water table is recognised as a continuous high-amplitude reflection at about 80 to 90 ns in the northeastern half of the section (positions 0–40 m) and, less clearly, southwest of position 75 m at about 70 ns. Above the water table, subparallel continuous reflections dominate. A conspicuous trough-shaped reflection, split into two parts, is observed in the centre of the section between positions 40 and 80 m. The relation between the GPR section and the exposed sedimentological section (Figs. 5 and 7a) was not immediately evident, so modelling software (Sensors and Software, 1996) was used to help interpret the reflections. In this case, 2D ray-tracing (Yilmaz, 1987) was applied after physical properties (radar wave propagation velocity in m=ns and attenuation in dB=m) were assigned to the different layers in the model (Fig. 7b). A Ricker wavelet was used in the final convolution to the synthetic GPR section (Fig. 7c). The synthetic section shows good agreement with the field GPR section (Fig. 7d). The major components of the model were the exposed peat layer and the top of the gravel deposits. The peat layer, observed at depths between 2 and 3 m below the surface, produces the third continuous reflection between 25 and 30 ns. The double-channel-shaped reflection from the top of the gravel layer is distinct, despite the small velocity contrast between the sands (12 cm=ns) and the gravels (15 cm=ns) as-

254

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

Fig. 6. GPR section over channel complex from the top of the braided Lower Terrace of the Maas (Bosscherheide sand pit); (a) 100 MHz GPR section, (b) interpretation.

sumed in the model. The top of the gravel deposits is exposed in the pit wall at depths that correspond with the top of the trough-shaped reflections. The northeast-dipping portion in the sedimentary section at the 60 m position correlates well with the GPR reflections, but the limited exposure toward the southwest makes correlation with GPR reflections (e.g. the second trough) more difficult. The water table was included in the model at a depth of 4.25 m in accordance with the field observations. An interesting minor feature is the decrease in the amplitude of the water-table reflection beneath the peat

reflections. In the field data, noise by diffractions interferes with the water-table reflection, making its recognition more difficult. Moreover, the shape of the water-table reflection below the channel appears to mirror the overlying reflection from the channel bottom. This could be the result of a low-velocity zone in the channel fill, caused by a higher moisture content of the fine-grained parts of the channel fill. Thus, according to the interpretation, the channel has a width of about 40 m and is divided into two parts by a mid-channel high. Small gullies incise the channel fill locally. The gully observed in the

Fig. 7. Enlarged channel complex of Fig. 6; (a) exposed sedimentary section, (b) GPR ray-trace model, (c) synthetic GPR section, (d) field GPR section with interpretation.

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

255

256

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

exposure between positions 48 and 54 m (Fig. 5), however, is not visible in the GPR section, probably because of the lack of physical contrast. Surrounding and underlying the observed channel structures, subparallel reflection patterns that are characteristic for braided river deposits can be recognised (see Section 3.1.3: Mill). 3.1.3. A braided river floodplain at Mill The deposits at Mill (Fig. 1) are associated with one of the Maas terraces of middle (Late) Pleistocene age. According to the drilling logs (Huisink, 1997), the deposits consist of stacked sets of heterogeneous, coarse sediments. The individual sets are sharply bounded and between a few decimetres and one metre thick. Their basal parts consist of coarse sand and gravel, whereas the upper part is composed of medium-grained sand. The characteristics of these deposits are identical to the other Maas terraces formed during the Pleistocene glacial periods and suggest braided river deposition. GPR was applied here in order to reconstruct the lateral extent of the sets and categorising them in an architectural framework. The GPR section at Mill was recorded with 200 MHz antennas (Fig. 8a). CMP soundings made in the survey area did not yield information on velocities below the water table, which is at a depth of approximately 1.5 m. A velocity of 9 cm=ns was used for the depth scale of the section, which is a correct average at about 70 ns. The penetration of the 200 MHz radar waves was about 5 m. The undulating reflection pattern that dominates the image is discontinuous. Migration was applied to this section in order to discriminate between true reflections from sedimentary interfaces and diffractions (Fig. 8b). A program for 2D (F–K) migration (from Sensors and Software, Inc.) and an average migration velocity of 9 cm=ns were used. A remarkable improvement was achieved in some parts. The continuity of several reflections was enhanced while disturbing diffraction hyperbolae were removed. This is well-illustrated by a hyperbola at the bottom of the section (at about 100 ns) at position 297 m, as well as by an earlier (at about 50 ns) hyperbola above it. Fig. 8c presents the interpretation of the GPR section. The reflections are generally subhorizontal or gently undulating and discontinuous. Although

some appear to extend tens of metres, in fact these reflections actually intersect at very small angles. This general pattern is locally interrupted by shallow trough-shaped reflections with well-delineated channel margins (e.g. at positions 310 to 320 m). Their base represents erosion surfaces. The hummocky reflections are generally 3 to 6 m wide and about 0.5 m deep, yielding width=depth ratios of 6 to 12. In some places, basal erosion surfaces are overlain by large-scale inclined strata (discontinuous dipping reflections) passing into channel fills (e.g. at position 298–307 m). They point to slight lateral migration and filling of the channels. At other places the channels show no lateral migration (e.g. at position 308– 311). The general sedimentary structures are interpreted as alternating bar surfaces with local remnants of channels having high width=depth ratios. Fig. 8 is a representative GPR section for braided river deposits. 3.2. Meandering rivers 3.2.1. A meandering channel at Beugen The GPR images of deposits from a well-preserved river meander in one of the Maas terraces at Beugen were investigated (Figs. 1 and 9). According to pollen analysis of the meander fill, the meander was abandoned at the beginning of the Allerød period (ca. 13,500 cal. yr BP). The channel dimensions and the sedimentary characteristics of the channel fill were determined from a series of boreholes perpendicular to the meander axis (Fig. 10a; Kasse et al., 1995). The palaeomeander is incised in an older Late Glacial terrace, a remnant of which is visible on the west side of the section. The channel fill after abandonment is mainly organic (gyttja and peat) and has a total thickness of 1 to 3 m. The gyttja layer overlies coarse sand and gravel from the previously active channel. Later, the abandoned meander was briefly reactivated, resulting in the deposition of a large plug of clay and some sand. Finally peat was formed on top of the sand and clay. The GPR measurements were made with 100 and 200 MHz antennas. Best results were obtained with 100 MHz frequencies, because they penetrated better into the electrically conductive soil. CMP soundings yielded velocities of about 7 cm=ns, indicating that most of the subsurface is water saturated. The water

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

257

Fig. 8. A 200 MHz GPR section near Mill with subhorizontal and trough-shaped reflection patterns characteristic of braided river systems. For location see Fig. 1; (a) 200 MHz GPR section, (b) the same section, migrated, (c) interpretation.

258

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

Fig. 9. Geomorphology of the meander scar at Beugen showing locations of the GPR sections (1 and 2) in Figs. 10 and 11. The location of this site on a Late Glacial Maas terrace is shown in Fig. 1.

table is in fact at or near to the surface. The maximum penetration at 100 MHz is less than 3 m in the channel fill and about 5 m in the point bars. GPR section 1 at Beugen (Fig. 10b) was recorded along the borehole section of Fig. 10a. The older terrace on the west side of section 1 is characterised by subparallel, continuous reflections. The boundary between the older terrace and the palaeomeander is evident near position 30 m. The reflection at 55 ns at position 10 m is the water table (Fig. 10c). The apparent westward dip of this reflection, which is visible until position 80 m, is due to the lower topography in the meander scar. The remaining subhorizontal reflections represent stratification in the fluvial sands (see Section 3.3: Haps) but to some ex-

tent, may also be the result of cables and pipes buried parallel to the GPR profile along the road on the terrace. The fill of the meander scar exhibits several subhorizontal reflections. The reduced penetration of the radar waves can be attributed to the very high electrical conductivity of the clayey fill and a nearsurface water table. This signal attenuation prevented the base of the meander fill to be detected. However, the westernmost part of the meander shows a very clear channel structure between positions 30 and 65 m. This structure corresponds exactly with the sandy fill of the rest gully of the meander in the geological section (Fig. 10a). The bottom of the peat layer creates a significant reflection. This is illustrated by ray-trace modelling that was performed to improve the interpretation. Fig. 10d shows the model with a geometry following the lithological section (Fig. 10a) and estimated model parameters (propagation velocity and attenuation). The geometry of the peat–clay interface was slightly adjusted east of position 75 m in order to obtain better agreement with the GPR section. The major reflection in the synthetic section (Fig. 10e), convolved with a 100 MHz Ricker wavelet, is produced by the peat–clay interface. The top of the peat layer is not visible in the GPR field section because its reflection at early times is masked by the arrivals of the direct waves. The deep reflection from the clay–sand transition is only visible in the synthetic section where the clay layer is thin. In the field data, this reflection is not very clear. The attenuation of the clay is probably greater than the 10 dB=m assumed in the model for the synthetic section. GPR section 2 at Beugen (Fig. 11) was recorded in the inner part of the palaeomeander. This section shows typical radar facies and structures from this part of the meander. It is situated outside the meander scar, so that the absence of clayey deposits allowed a better signal penetration. Scroll bars are weakly expressed in the surface morphology. However, the sedimentary data from boreholes show repeating fining-upward sets of medium to coarse sand (0.5 to 2 m thick) in the uppermost metres. Slightly dipping (approximately 6º to 10º) reflections are visible

Fig. 10. Palaeomeander at Beugen (for location see Figs. 1 and 9); (a) drilling section (after Kasse et al., 1995), (b) 100 MHz GPR section 1 with trough-shaped reflections from the channel, (c) interpretation, (d) GPR ray-trace model, (e) synthetic GPR section.

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

259

260

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

Fig. 11. Palaeomeander at Beugen (for location see Figs. 1 and 9); (a) 100 MHz GPR section 2 with west-dipping reflections from point bars, (b) interpretation.

throughout most of section 2. The inclination to the west, i.e. towards the palaeomeander, represents fining-upward sequences of point bars. The base of this pattern is at a depth of approximately 4.5 m. This reflection pattern terminates at 640 m, after which a more subparallel reflection pattern is visible. In more easterly GPR sections (i.e. farther from the meander scar; see Fig. 9), the same subparallel pattern continues and has a few dipping point-bar reflections only within the uppermost 1 to 2 m. This subparallel reflection pattern strongly resembles the older Late Glacial terrace on the west side of section 1. The boundary between the two types of reflection patterns is sharp and strongly inclined, and could correspond to an erosional contact. 3.2.2. The meandering river at Lottum The site at Lottum (Figs. 1 and 12) is located some 20 km south of the sections at Beugen. At the Lottum site, the Late Glacial Maas has incised in the Weichselian Pleniglacial terrace and has formed a lower terrace comprising a belt of large meanders. The Pleniglacial terrace is composed of generally

planar sand and gravel beds deposited by a braided river (detailed sedimentological descriptions of this terrace are given by Kasse et al., 1995 and Van de Berg, 1996). The surface of the Late Glacial terrace is approximately 3 m lower than the Pleniglacial terrace and belongs to the same Late Glacial terrace as the channel at Beugen. The aim of the GPR survey here is to characterise the radar facies of the different elements of the meander floodplain and to compare them with the radar facies of the braided river deposits of the Pleniglacial terrace. Deep (maximum 16 m) drilling on the Late Glacial terrace reveals the existence of long, up to 7.5 m thick fining-upward sequences of coarse gravelly sand and gravel at the base to fine silty sand at the top (detailed descriptions in Kasse et al., 1995). The lower 1 to 1.5 m are interpreted as channel bed deposits while the overlying sediments of this fining-up sequence are interpreted as point bars in a meandering river. The morphological expression of a large meander bend at the top of the terrace, and long curved ridges within and parallel to the meander bend (accretion topography) confirm that interpretation (Fig. 12).

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

261

Fig. 12. Topographic map of the meander near Lottum showing locations of the GPR sections in Figs. 13–15 (after Kasse et al., 1995).

This palaeomeander marks the boundary with the Pleniglacial terrace in the west. Like the palaeomeander at Beugen, it was active during the Allerød period. Low (aeolian) dunes are locally present on top of the point-bar sediments. The complete GPR section at Lottum is approximately 5 km long and perpendicular to the axis

of the Maas valley (Fig. 12). Its extreme west end is situated on the Weichselian Pleniglacial terrace. The individual GPR sections were recorded with 50 MHz and 100 MHz antennas and were corrected for topography. In general, the best images of the sedimentation patterns in this area were obtained with the 50 MHz antennas. The maximum penetration

262 J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270 Fig. 13. GPR section 1 near Lottum, perpendicular to the axis of the Maas valley, with transition (at position 385 m) from braided river deposits with subhorizontal reflection patterns to meandering river deposits with (west-)dipping reflections from point bars; (a) 50 MHz section, (b) interpretation. For location see Fig. 12.

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

with this frequency was approximately 10 m. No CMP soundings were made along the profile. The water table at this site was less than 1 m deep. As most sedimentary structures of interest are situated below that level, a velocity of 8 cm=ns was used for the elevation scale of the radar sections. A change in reflection pattern occurs between the western and the eastern parts of the section Lottum 1 (Fig. 13). The western part is characterised by subhorizontal and slightly undulating or hummocky reflections that are typical of braided channel deposits like those at the Mill site (Section 3.1.3). In the eastern part of the section, there is a twofold subdivision of the vertical sequence. The upper part shows rather continuous reflections that systematically dip about 6º to the west. These reflections are approximately 1.5 m vertically apart. They most likely correspond to the point-bar sequences interpreted from the sedimentological and geomorphological data. This point-bar sequence occurs east of position 480 m. The reflection pattern of the lower part, the top of which is about 6 m deep (elevation 12 to 13 m), is quite similar to the subhorizontal pattern in the west. According to the data of borehole 2 (Fig. 12), the erosional contact between the channel

263

bed deposits of the meandering river and the underlying horizontally bedded gravels is at an elevation of 11 m and the boundary between the channel beds and the overlying point bars at ca. 12.5 m. Therefore, the lower reflection pattern in the eastern part of section Lottum 1 may represent both the meander channel deposits and the underlying braided river deposits. The section between positions 384 and 480 m mainly corresponds with the position of the last active channel that, after abandonment, was filled with subplanar sand beds, which are shown as subhorizontal reflections. This reflection pattern is unlike the one in the meander scar of Beugen, because of the different fill (electrically conductive clay-rich sediments at Beugen). This part of the section is also disrupted by a diffraction from a culvert at position 446 m and by ‘ringing effects’ from metallic objects in the ground at positions 454 and 479 m. From the geomorphology (Fig. 12) it appears that a more recent meander, 400 m to the east, cut off the meander channel described above. The position of this later meander channel (portions of which remain filled with water) is imaged in the GPR section Lottum 2 (Fig. 14). The best image of the final 30-

Fig. 14. GPR section 2 near Lottum with trough-shaped reflections from a slightly younger palaeomeander channel; (a) 100 MHz section, (b) interpretation. For location see Fig. 12.

264

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

Fig. 15. GPR section 3 near Lottum across point-bar deposits of the slightly younger meander system; (a) 50 MHz section, (b) interpretation. For location see Fig. 12.

m-wide channel and channel fill was produced with the 100 MHz antennas. The trough-shaped reflection pattern contrasts strongly with the subhorizontal reflections from the older meander channel to the west. A SEC gain was applied to this section in order to distinguish high- and low-amplitude reflections. It is clear that the amplitudes of the channel fill reflections are higher than those of the neighbouring subhorizontal reflections. West of the main channel, high-amplitude reflections from a second, shallower channel can be seen. Further to the east, a new series of inclined reflections appear in the GPR images (Fig. 15). They represent point-bar structures that belong to the younger meander system. In section Lottum 3, dips of reflections in the western part are somewhat steeper (6.5º) than those in the eastern part (5.5º). At shallow depth, some contrasting dips are observed, and it is not clear whether these have a sedimentological origin or are hyperbolic diffractions. It is also possible that they represent overwash deposits.

3.3. A transitional river system at Haps The GPR section at Haps is located northeast of the Pleistocene terrace near Mill (Fig. 1). It is situated on a lower, younger terrace that dates from the beginning of the Late Glacial (Bølling) and is slightly older than the meandering river terrace of Beugen and Lottum (Vandenberghe et al., 1994; Kasse et al., 1995; Huisink, 1997). The surface morphology of this terrace is characterised by a number of relatively shallow, narrow and slightly to moderately curved channels (Fig. 16). This pattern is intermediate between a one-channel high-sinuosity meandering system and a braided system consisting of a multitude of low-sinuosity channels (Vandenberghe et al., 1994). The channels alternate with relatively narrow, linear to slightly curved ridges that are tens of metres to a hundred metres apart. The height differences between ridges and channel floors are no more than 2 m as appears from detailed borehole sections (described in Huisink, 1997). Deposits consist of stacked, fining-upward sequences of sandy

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

265

Fig. 16. Mapping of the multiple channel system (2) on the transitional Maas terrace of Late Glacial age at two different sites (for location see Fig. 1). The Late Pleniglacial braided terrace (1) is at the left side and the Younger Dryas and Holocene river plains (3) at the right side in (A). Their wider geomorphological setting is apparent from Fig. 1. 4 D urban areas; 5 D location of the GPR sections Haps 1 and 2; (A) is after Vandenberghe et al. (1994), (B) is based on data of Huisink (1997).

gravel to very fine sand, which have thicknesses ranging from 0.5 to 2.5 m. A sandy silt or clay may be present on top. The relatively thick fining-upward sets and the very low braiding index make this river type clearly different from the typical braided system (such as at Mill). It resembles to some degree the ‘low-sinuosity river with alternate bars’ described by Miall (1996), which is also an intermediate type between braided and meandering. The GPR sections at Haps were recorded with 100 MHz antennas. CMP soundings made in the survey area did not yield information on velocities below the water table. The depth to the water table is about 1.5 m; therefore, a velocity of 8 cm=ns was used to produce the depth scale of the section. The maximum penetration of the 100 MHz radar waves was approximately 5 m. In Fig. 17 the GPR section Haps 1 shows two distinct reflection configurations with their boundary at about 3 m depth. The lower one has an undu-

lating pattern with low amplitude (about 1 m) and large wavelength (about 10 to 20 m). These reflections are more curved than in a typically braided pattern. The upper configuration consists of inclined reflections within troughs that are 12 to 16 m wide. These reflections are interpreted as lateral accretion layers from channel migration. This pattern resembles that of the meandering structures at Lottum and Beugen, except that it is clearly smaller scaled, although still larger than that of the braided structures at Mill and Tegelen. Reflections from sedimentary and morphological features of an intermediate zone between braiding and meandering are interpreted in the GPR section. More specifically, moving upward from the lower to the upper reflection configurations in Fig. 17 represents the transition from a braided to a meandering river system. In the transitional system, the relationship between an undulating surface topography and underlying sedimentary structures that are imaged in the uppermost

266

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

Fig. 17. GPR section 1 northwest of Haps showing subhorizontal to undulating reflections and small trough-shaped reflection patterns that are characteristic of a transitional river system; (a) 100 MHz section, (b) interpretation. For location see Figs. 1 and 16B.

portions of the GPR sections is not always straightforward. The best concordance may be observed in the section Haps 2 (Fig. 18). Under a topographic ridge at positions 635–655 m, convex-upward reflection patterns can be seen in the GPR section. They are interpreted as channel bars that formed as islands between the individual branches of a multi-channel system (cf. Williams and Rust, 1969). Below the topographic channel depression (positions 670–700 m), oblique reflections are visible in the GPR section. To some extent, they resemble the point-bar structures of the meanders (e.g. sections from Lottum), but they are less clearly defined and have distinctly smaller dimensions. They are interpreted as lateral accretion bars formed by slight migration of the individual channel. The intermediate character of the fluvial system at Haps between one-channel (meandering) and multi-channel (braided) is consistent with this interpretation (Vandenberghe et al., 1994).

4. Discussion and conclusions Characteristic patterns of the different fluvial radar facies are presented in a chart in Fig. 19. A threefold subdivision of the observed reflection patterns corresponds to the three depositional patterns observed in the study region: a braided, a meandering and a transitional pattern. Radar facies of both the floodplain and the channels are clearly different from each other in the braided and meandering systems. In the studied examples, the channel fill of the abandoned meander is characterised by horizontal reflections (representing a horizontal fill of clayey sediments), but the sandy channel fill of the braided channels produced trough-shaped reflections that mimic the channel base. In the floodplain of the meandering river large-scale parallel dipping reflections represent point bars, while the images of the braided floodplain are dominated by irregularly intersecting, concave-

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

267

Fig. 18. GPR section 2 northwest of Haps illustrating the correspondence between topography and GPR reflection pattern (convex-upward structures below the topographic ridge, oblique reflections below the topographic depression); (a) 100 MHz section, (b) interpretation. For location see Figs. 1 and 16B.

upward reflections with diffractions in places (representing the multiple channels of the system), or wide horizontal reflections (representing the sheetlike bar structures of the braided river plain). Obviously, these characteristics are simplified to some extent, for instance the individual channels of a braided system may migrate laterally and form accretionary bars, but they indicate the dominating patterns distinctly. The transitional system radar facies show a regular alternation of small, sand-filled channels and interfluves with a few indistinct curved structures (possibly representing bar structures); some weak lateral accretion is locally present also. The individual facies represent type-sections that have been added to an interpretive radar facies at-

las comprising diagnostic reflection patterns in all those depositional sedimentary environments in the Netherlands where radar waves can penetrate. Table 1 summarises the basic elements of the different radar facies of the studied palaeofluvial sediments. It is clear, however, that the radar facies alone cannot uniquely define a specific fluvial palaeoenvironment. Nevertheless, it is a handy guide for interpreting such environments, and should be used in combination with the traditional sedimentological and geomorphological analyses. In the cases described, the potentials of GPR measurements for the interpretation of fluvial sedimentary structures on a detailed level are demonstrated. Not only were individual sedimentary facies within

268

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

Fig. 19. Radar facies chart of characteristic reflection patterns from fluvial palaeoenvironments of the Maas valley. Channel and floodplain patterns in braided, meandering and transitional river systems are distinguished. The vertical scale is twice the horizontal scale.

the braided and meandering systems distinguished, but also the transitional system between braided and meandering was characterised. The possibility of correlating such a multitude of quickly-acquired

and voluminous GPR data with sedimentological data offers new perspectives for geomorphological and palaeogeomorphological investigations. For example, the possibility of determining accurately

Table 1 GPR reflection patterns from sediments in various fluvial palaeoenvironments in the Netherlands Braided river system

Meandering river system channel (and fill)

Transitional river system

channel (and fill)

floodplain

Reflection configuration

prograded or trough-shaped; diffractions

subhorizontal to trough-shaped with hummocky and subhorizontal fill undulating

oblique (low angle)

dipping (cross-bedded) in small troughs overlying more continuous undulating

Continuity of reflections

low–moderate

low

moderate

moderate

low

low

low

moderate

low

low

wavy

channel

cross-layered sets

Small troughs overlying wavy pattern

Reflection

amplitude a

External form (geometry) channel of facies unit

floodplain

a Almost all reflections from the observed fluvial systems have low amplitudes (hence, the sections have been displayed with AGC gain). An exception is the moderate amplitudes of reflections from meandering channels.

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

palaeochannel dimensions is useful for palaeohydrological studies where the input of realistic channel parameters in models and in sediment budget calculations is necessary. In addition, knowing the characteristic heterogeneities and anisotropies at different scales for each river system deposit is important for environmental and engineering purposes.

Acknowledgements The GPR investigations were carried out within the scope of a project financed by six Dutch water supply companies: Nuon Water BV, NV Waterleiding Maatschappij Gelderland, NV Waterleidingmaatschappij Oostelijk Gelderland, Waterleiding Maatschappij Overijssel NV and NV Waterleidingbedrijf Midden-Nederland. Additional financial support was received from the Dutch Ministry of Economic Affairs. The GPR data were acquired by J.W.T.M. Reckman and P.J. Dekker of TNO. The following students from the Vrije Universiteit Amsterdam participated in both the fieldwork and the processing of the GPR data: M. Deelder, A.I.J.M. van Dijk, O. van der Kolk and R. Rippens. The maps were drawn by H. Sion (Vrije Universiteit Amsterdam), and all other illustrations were made by W.A.J. Immers (TNO). The authors are grateful for discussions and sharing data by Dr. M. Huisink and Dr. C. Kasse and the helpful comments of Dr. M. Beres and Dr. J.S. Bridge.

References Allen, J.R.L., 1970. Studies in fluviatile sedimentation: a comparison of fining-upward cyclothems, with special reference to coarse-member composition and interpretation. J. Sediment. Petrol. 40, 298–323. Beres, M., Green, A., Huggenberger, P., Horstmeyer, H., 1995. Mapping the architecture of glaciofluvial sediments with threedimensional georadar. Geology 23, 1087–1090. Boersma, J., Van Gelder, A., De Groot, Th., Puigdefabregas, C., 1981. Formen fluviatiler Sedimentation in neogenen und jungeren Ablagerungen im Braunkohlentagebau Frechen (Niederrheinische Bucht). Fortschr. Geol. Rheinland Westfalen 29, 275–307. Bohncke, S., Vandenberghe, J., Huijzer, A.S., 1993. Periglacial palaeoenvironments during the Weichselian Late Glacial in the Maas valley, the Netherlands. Geol. Mijnbouw 72, 193–210.

269

Bridge, J.S., 1985. Palaeochannel patterns inferred from alluvial deposits: a critical evaluation. J. Sediment. Petrol. 55, 579– 589. Bridge, J.S., 1993. Description and interpretation of fluvial deposits: a critical perspective. Sedimentology 40, 801–810. Bridge, J.S., 1998. Large-scale structure of Calamus River deposits (Nebraska, USA) revealed using ground-penetrating radar. Sedimentology 45, 977–986. Bridge, J.S., Alexander, J., Collier, R.E.L., Gawthorpe, R.L., Jarvis, J., 1995. Ground-penetrating radar and coring used to study the large-scale structure of point-bar deposits in three dimensions. Sedimentology 42, 839–852. Bu¨ker, F., Gurtner, M., Horstmeyer, H., Green, A.G., Huggenberger, P., 1996. Three-dimensional mapping of glaciofluvial and deltaic sediments in Central Switzerland using ground penetrating radar. Proc. 6th Int. Conf. Ground Penetrating Radar, Sendai, pp. 45–50. Cant, D.J., Walker, R.G., 1976. Development of a braided fluvial facies model for the Devonian Battery Point Sandstone, Quebec. Can. J. Earth Sci. 13, 102–119. Davis, J.L., Annan, A.P., 1989. Ground-penetrating radar for high-resolution mapping of soil and rock stratigraphy. Geophys. Prospect. 37, 531–552. Doppert, J.W.C., Zonneveld, J.I.S., 1955. Over de stratigrafie van het fluviatiele Pleistoceen in W. Nederland en N-Brabant — Voorlopige Mededeling. Meded. Geol. Stichting N.S. 8, 13–30. Gawthorpe, R.L., Collier, R.E.Ll., Alexander, J., Bridge, J.S., Leeder, M.R., 1993. Ground penetrating radar: application to sandbody geometry and heterogeneity studies. In: North, C.J., Prosser, J. (Eds.), Characterization of Fluvial and Aeolian Reservoirs. Geol. Soc. London, Spec. Publ. 73, 421–432. Green, A., Pugin, A., Beres, M., Lanz, E., Bu¨ker, F., Huggenberger, P., Horstmeyer, H., Grasmu¨ck, M., De Iaco, K., Maurer, H.R., 1995. 3D high-resolution seismic and georadar reflection mapping of glacial, glaciolacustrine and glaciofluvial sediments in Switzerland. Proc. Symp. Application of Geophysics to Engineering and Environmental Problems (SAGEEP), Orlando, FL, 419–434. Huggenberger, P., 1993. Radar facies: recognition of facies patterns and heterogeneity estimation (Pleistocene Rhine gravel, NE Switzerland). In: Best, J., Bristow, C. (Eds.), Braided Rivers: Form, Processes and Economic Application. Geol. Soc. London, Spec. Publ. 75, 163–176. Huggenberger, P., Meier, E., Beres, M., 1994. Three-dimensional geometry of fluvial gravel deposits from GPR reflection patterns: a comparison of results of three different antennas frequencies. Proc. 5th Int. Conf. Ground Penetrating Radar, Kitchener, ON, 805–816. Huisink, M., 1997. Late-glacial sedimentological and morphological changes in a lowland river in response to climatic change: the Maas, southern Netherlands. J. Quat. Res. 12, 209–223. Jol, H.M., Smith, D.G., Meyers, R.A., Lawton, D.C., 1996. Ground penetrating radar: high resolution stratigraphic analysis of coastal and fluvial environments. In: Pacht, J.A., Sheriff, R.E., Perkins, B.F. (Eds.), Stratigraphic Analysis utilizing

270

J. Vandenberghe, R.A. van Overmeeren / Sedimentary Geology 128 (1999) 245–270

Advanced Geophysical Wireline and Borehole Technology for Petroleum Exploration and Production. Proc. GCSSEPM Foundation 17th Annual Research Conference, Houston, TX, 153–163. Kasse, C., Vandenberghe, J., Bohncke, S., 1995. Climatic change and fluvial dynamics of the Maas during the Late Weichselian and Early Holocene. Pala¨oklimaforschung 14, 123–150. Miall, A., 1996. The Geology of Fluvial Deposits. Springer, Berlin, 582 pp. Pons, L., 1954. Het fluviatiele Laagterras van Rijn en Maas. Boor Spade VII, 97–110. Roksandic, M.M., 1978. Seismic facies analysis concepts. Geophys. Prospect. 26, 383–398. Sangree, J.B., Widmier, J.M., 1979. Interpretation of depositional facies from seismic data. Geophysics 44, 131–160. Schumm, S.A., 1960. The shape of alluvial channels in relation to sediment type. U.S. Geol. Surv. Prof. Pap. 353B, 17–30. Sensors and Software, 1996. PulseEKKO 2D Ray Trace Modelling and 2D F-K Modelling User’s Guide, version 1. Sensors and Software Inc., Mississauga, ON. Stephens, M., 1994. Architectural element analysis within the Kayenta Formation (Lower Jurassic) using ground-probing radar and sedimentological profiling, southwestern Colorado. Sediment. Geol. 90, 179–211. Van de Berg, M.W., 1996. Fluvial sequences of the Maas: a 10 Ma record of neotectonics and climate change at various time-scales. Doctors Thesis, Wageningen, 180 pp.

Vandenberghe, J., Mommersteeg, H., Edelman, D., 1993. Lithogenesis and geomorphological processes of the Pleistocene deposits at Maastricht-Belve´de`re. Meded. Rijks Geol. Dienst 47, 7–18. Vandenberghe, J., Kasse, C., Bohncke, S., Kozarski, S., 1994. Climate-related river activity at the Weichselian–Holocene transition: a comparative study of the Warta and Maas rivers. Terra Nova 6, 476–485. Van den Broek, J.M.M., Maarleveld, G.C., 1963. The Late Pleistocene terrace deposits of the Meuse. Meded. Geol. Stichting 16, 13–24. Van Overmeeren, R.A., 1994. High speed georadar data acquisition for groundwater exploration in The Netherlands. Proc. 5th Int. Conf. Ground Penetrating Radar, Kitchener, ON, 1057– 1073. Van Overmeeren, R.A., 1998. Radar facies of unconsolidated sediments in the Netherlands — a radar stratigraphy interpretation method for hydrogeology. J. Appl. Geophys. special issue 40, 1–18. Williams, P., Rust, B., 1969. The sedimentology of a braided river. J. Sediment. Petrol. 39, 649–679. ¨ ., 1987. Seismic Data Processing. Society of ExploYilmaz, O ration Geophysicists, Tulsa, OK, 526 pp. Zonneveld, J.I.S., 1958. Lithostratigrafische eenheden in het Nederlandse Pleistoceen. Meded. Geol. Stichting N.S. 12, 31– 64.