Interactive comment on “ A geophysical analysis of hydro-geomorphic controls within a headwater wetland in a granitic landscape , through ERI and IP ”

3) Prior to the complete data inversion the raw electrical resistivity data was assessed for its quality according to the methods of Loke (2004), where bad data points are removed from a preliminary inversion array according to high RMS error, and are therefore excluded from the final data inversion process. This is now discussed in Section 3.1 Overall approach and reiterated in Section 3.2 3-D approach.


Introduction
Wetlands in southern Africa as in most parts of the world continue to be vulnerable to anthropogenic pressures and climatic shifts which influence the hydrological and geomorphologic processes that otherwise maintain the equilibrium of the wetlands dynamic state. Consequently many wetlands are continually undergoing degradation. In recent years the drive has been to determine the wetlands structure and function 20 within the broader landscape in order for rehabilitation of these degraded systems to be sustainable. In the interests of ecosystem integrity significant emphasis is now placed on understanding a wetlands hydro-geomorphic setting to ensure the success of such intervention, whereby the causes of wetland degradation rather than the symptom are incorporated into the rehabilitation planning (Tooth and McCarthy, 2007). In are the focus of technical rehabilitation efforts. Descriptions of the hydrodynamic behaviours of the wetland phreatic surfaces prior to rehabilitation interventions revealed stark differences in event driven responses either side of a suspected clay-plug in this wetland (Riddell et al., 2007). Clay-plugs are distinct zones of fine particles in an environment that is otherwise dominated by coarse particles, and due to the lower hydraulic 15 conductivities of this medium, they are thought to exert significant controls on lateral sub-surface flows in the wetlands of the Sand River catchment. It is the identification and examination of these sub-surface structures that are presented here.
This paper presents results of geophysical interpretation of the wetlands hydrogeomorphology using Electrical Resistivity Imaging (ERI) in both 2 dimensional form 20 (2-D) during scoping surveys and a more detailed examination of hydro-geomorphic controls in 3 dimensional form (3-D) using ERI and Induced Polarization (IP) techniques. The 2-D surveys using ERI only were deployed at various times during the hydrological characterization of this catchment (2005)(2006)(2007)(2008)(2009) and were used in a reconnaissance fashion to assist in the interpretation of hydrological processes. The 3-D 25 surveys using both ERI and IP were used to examine the clay-plug distribution and develop concepts regarding their geomorphic evolution. The geophysical techniques used here have their roots in the mineral prospecting and other geotechnical investigations (Lowrie, 2007) but are now being increasingly used by the earth science fraternity Introduction

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion involved in water resources management and geomorphology studies, whose recognition of their utility has been widely acknowledged (Robinson et al., 2008;Dahlin, 1996). ERI for instance has successfully been used in the interpretation of hillslope flow pathways in research catchments in South Africa (Ulhenbrook et al., 2005), examining permafrost distribution in peri-glacial mountain environments in Europe (Kneisel,5 2006) and weathering and erosion processes on lateritic plateaus of south Pacific islands (Beauvais et al., 2007). ERI has been praised as a valuable, cost-effective and non-intrusive tool for conducting reconnaissance surveys in geomorphological investigations (Smith and Sjogren, 2006). Although IP has not received as much attention as resistivity methods, the IP method is proving valuable as a complementary tool to 10 resistivity mapping where the application of IP to delineate sub-surface materials in low resistivity clay rich environments has shown success in, for instance slope instability studies in Switzerland (Marescot et al., 2008). The basis of ERI surveying is the induction of an electrical current by an array of current electrodes inserted into the ground surface, with a sequence of potential elec- 15 trodes receiving the electrical signal. Hence the measurements of electrical potential at the surface are dependent on the electrical resistance of sub-surface materials, expressed in ohm metres (Ω m), which vary according to sub-surface strata, as a result of their inherent pore size distribution, variations in water content and degree of salinity (Loke, 1999). The collation of this sub-surface resistivity distribution is facilitated by 20 computerized devices known as switching units that control the sequence of pairs of current and potential electrodes, known as the quadripole. Variations in the position and sequence of current and potential electrodes along a transect facilitate measurements with different degrees of horizontal and/or vertical resolution, these computergenerated sequences (protocols) are known as arrays. Inversion algorithms are used 25 to create a model of measured and apparent (calculated) resistivity values. Using Jacobian matrix calculations and forward modeling procedures produces values of the true resistivity, which when plotted in 2-D or 3-D are known as pseudosections. The Introduction

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion law governing apparent resistivity (p a ) is written as follows: and where: k is the geometric factor and varies according to the quadripole array, R is the resistance, V is the voltage and I is the current.

5
These plotted pseudosections are then used to examine sub-surface resistivity heterogeneities based on the differences of equipotential surfaces where they hit the ground surface at potential electrodes; the interpretation of this data is based on known resistivity ranges of geologic materials. Following on from this procedure, IP surveys measure the decay of the electrical pulse from the current electrodes as the 10 sub-surface returns to electrical equilibrium, since particular geologic materials are polarizable, some parts of the sub-surface domain essentially act as capacitors. This is particularly pertinent in the case of clays which have a high propensity for membrane polarization as result of ion accumulations on the particles surface and hence within pore spaces (Kiberu, 2002). In essence then the IP survey is measuring the charge-15 ability (M) of sub-surface materials and this is written in the form: Where V 0 is voltage at time (t) 0, where subsequent time-voltage values are an expression of the decay of the steady-state voltage.
2 Field setting 20 The study site ( Fig. 1)    with a similar evapotranspiration demand. Rainfalls in this escarpment region are large and intense following the arrival of warm humid air from the south-east 10 during the wet season (October to March), the magnitude of these events drops rapidly eastward away from the escarpment (Fig. 1). Due to the dominant granitic geology, soils within the catchment are medium to coarse grained sands, and significantly, the clay concentration does not increase towards the valley bottom, hence the wetland soils are essentially sandy (Pollard et al., 15 2005), except where fine sediment clay-plugs have formed.
According to the South African vegetation type classification system (Acocks, 1988) this area falls under the lowveld sour bushveld. Land use within the Manalana is essentially densely populated rural housing of the communal land tenure typical of the area including a dense network of roads and pathways. Dry upland and wetland small holder 20 subsistence cultivation characterizes the area, where, in particular, extensive wetland cultivation (ridge and furrow systems) have altered the natural micro-topography and changed the balance of hydro-geomorphic processes within the system. This may take the form of disaggregating material through tillage practices, as well as increasing the level of channelization within the system by furrowing, amongst several other factors 25 for example. This is believed to be a causative agent facilitating the extensive gullying characterizing the Manalana catchment.

Overall approach
Since the commencement of the catchment hydrological monitoring program (September 2005) various 2-D-ERI surveys were undertaken along transects through the wetland and its contributing catchment. These were followed in July-August 2008 with a 5 detailed 3D-ERI and IP approach to ascertain the distribution of the sub-surface clayplugs and allow for some speculation on their development. In both cases use was made of an ABEM TM SAS1000 single channel Lund imaging system (Terrameter and switcher unit) with 64 electrode take-out over four cables. Electrode spacing and array types varied with each survey according to information requirements, a summary of 10 these are given in Table 1. Orientation of these surveys is displayed in Fig. 2. Inversion of 2-D and 3-D datasets were conducted using the RES2DINV and RES3DINV software packages respectively (Loke, 2005a, b). Inversion data were georeferenced with correct elevations using a Trimble® Pro-XRS differential Global Position System. 15 The first part of the 3-D study sought to examine the location and extent of the clayplugs within the wetland. For this purpose three parallel transects were placed at longitudinal orientation in the two arms of the wetland (as shown in Fig. 2), forming a juncture adjacent to the erosion gully head. Wenner-α arrays were used in each of the parallel transects using a single time interval IP sequence which also records 20 the sub-surface resistivity distribution at the same time, a default setting in the ABEM system. The second part of the 3-D approach sought to examine the interface between the hillslope and valley bottom (wetland) in order to determine whether these clay-plug zones are contiguous with the hillslope, i.e. at the interface of the wetland with the Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version Interactive Discussion it was hoped that any layered lithographic effects within the wetland substrate could also be discerned. Due to the small localized setting of this particular investigation and due to considerations in terms of the data acquisition time, new protocols were written such that Wenner-β arrays could be deployed using a single cable with a 21 electrode take-out (NB. The system consists of 4 cables, each of 21 take-out positions). This allowed for 3 subsequent transects to be set-up whilst the ABEM system was logging another. Referring to Fig. 3, the survey took the form of a gridded system made up of two orientations of 11 parallel transects, in both X and Y directions, making up a total of 22 parallel lines. Electrodes were spaced at 0.75 m intervals, and thus the total length of each individual transect was 15 m. Correct geometric factors (k) were applied to the 10 raw data in order to recalculate the apparent resistivity, prior to the inversion of the data set since the ABEM system limits the electrode spacing to metre integer values with one decimal place. Hence, the apparent resistivity distributions would be different to those that would otherwise have been calculated automatically by the ABEM system. In both 3-D approaches the resulting images are quasi-3-D as results are interpo-15 lated from 2-D transects, and the one channel SAS1000 ABEM system is limited in this regard. Also parallel transects were placed apart at two times the unit electrode spacing, within the bounds of acceptable error for a quasi-3-D approach (Loke, 1999). A brief explanation of this is warranted: up to this point discussion has focused on the vertical sensitivity of the electrodes to produce a 2-D pseudosection, meanwhile 20 electrodes also exert a radial sensitivity around themselves which becomes important where data is acquired and interpolated through inversion techniques in 3-D. The reader is referred to Loke (1999) for a more detailed explanation. In both 3-D approaches used here, data of 3-D transect "blocks" were defined by the "collate 2-D to 3-D function" within RES2DINV and inverted in RES3DINV. This function takes a series 25 of 2-D surveys (2-D data files) and through an interpolation technique based on known co-ordinates of each electrode produces a 3-D data file. 3-D images were plotted using Rockworks® 2006 software for visualizing earth surface and sub-surface data. The presence of groundwater has an effect on the resistivity range of earth materials HESSD 7,2010 A geophysical analysis of hydro-geomorphic controls  (Loke, 2004) and in order to achieve maximum clarity in terms of interpreting these 3-D data, the dry winter period is most appropriate for data collection. Therefore, these surveys were conducted during July-August 2008, during the mid winter period and height of the dry season for this region. Consequently this wetland had undergone a period of desiccation following the cessation of the rains earlier in the year, and as 5 a result, the influence of groundwater on the resistivity and IP soundings was at the minimum possibly achievable for this wetland system.

Supporting hydrometry
As part of the hydrological monitoring study undertaken within the Manalana catchment two forms of data are used to assist in the verification and interpretation of ERI and IP 10 data. These are simply shallow groundwater piezometer data, which were measured using regular (bi-/tri-weekly) dip meter readings in nested groundwater piezometers at hydrological monitoring stations within the wetland. The locations of these stations are located in Fig. 2. Moreover, these piezometers were also the subject of in-situ saturated hydraulic conductivity analysis, which took the form of bail tests using the 15 method of Bouwer and Rice (1976), which takes the form: where: K is the hydraulic conductivity of the wetland material, L is the height of the open screen portion of the piezometer at its interface with the wetland matrix (300 mm), y is the vertical distance between water level in the piezometer and that within the 20 wetland material at equilibrium at time 0 and time t. R e is the effective radius over which y is dissipated and r w is the horizontal radius between the centre of the piezometer and the aquifer (plus well casing and interfacing material, 42 mm). r c is the inside radius of the piezometer casing (27 mm). vertical felsic saprolitic protrusions from the underlying granitic bedrock. The bedrock material type is made with reference to known geology for the area and with respect to geological material ranges provided by Sharma (2008). These vertical protrusions could therefore be interpreted as sub-surface controls intersecting the unconsolidated wetland sediments which, in essence show that the wetland is characterized by what 15 appears to be a series of semi-confined aquifers with high water retention capacity due to their clay content. This is highlighted by the sharp transitions between the two material types at depths below 2 m (i.e. low resistivity clays versus high resistivity bedrock/saprolite). The significance of this observation is particularly important for the way that the wetland retains water and further hydrochemical analysis will reveal the 20 nature of the groundwater recharge processes that these wetlands facilitate, within the confines of these bedrock controls. Catchment ERI cross-sections were made perpendicular to the orientation of the wetland along two instrumented transects, transects 1 and 2 (Figs. 5 and 6). In both cases the substrate appears to be increasingly conductive towards the valley bottom Interactive Discussion than on the interfluves. In Fig. 5 it is apparent that the interfluve of the NE side of the transect is more conductive than the SW side. This is due to the intersection of the catchment's northern edge by a doleritic dyke, noted through in-situ observations, whereas the SW is a slope of granitic material. This is seen again in Fig. 6 for the S-N transect, the doleritic dyke is more apparent between 50-100 m with a vertical 5 structure of <150 Ω m extending to the base of the pseudosection. Further to the north of this transect in the upslope region the granitic geology reappears. These initial 2-D ERI surveys reveal the interplay of bedrock and regolith that characterizes this particular catchment. In particular there is an obvious contrast between the two geological substrates by way of the predominant granitic material and flanking 10 of the catchments by dolerite dykes. This interplay of geologies has important ramifications for the hydrological processes operating within this catchment, particularly since soil hydraulics will be influenced by the different soil textures and porosities yielded from soils on these different geologies. Most notably granites weather to coarse structured soils and dolerites to fine clay rich soils (Schaetzl and Anderson, 2005), and this 15 was reflected by the contrasting conductivities of the materials in Figs. 5 and 6. Hence the lower resistance of the doleritic material is explained by its higher clay content than the opposing granites. The observed hydrological responses reflecting these differing hillslope geologies to precipitation events and antecedent soil moisture conditions, and the consequent variability in the way that water is delivered to the valley bottom wetland 20 have been characterized for this catchment through hydrometric observations and hydropedological interpretation of soil form (Riddell, unpublished data, University of Free State internal report).
As expected, increasingly conductive material is observed within the longitudinal pseudosection displayed in Fig. 7, in particular the <100 Ω m material on the northern Interactive Discussion wetland topography has been mechanically altered by subsistence agricultural practices remembering the context of which this study was undertaken in a wetland that is characterized by ridge and furrow systems. This consequently disaggregates the wetland substrate, reducing its bulk density and hence inducing higher relative apparent resistivity. A significant observation from this pseudosection is the apparent vertical 5 protrusions of high conductivity material at 60 m and a similar possible structure at 45 m (contrasting with the high resistant protrusions of Fig. 4). At the extreme SE of this image a large area of high resistant material occurs and this forms the eroded and deposited sediments of the erosion gully head, corresponding to −40 m in Fig. 4. The fact that the longitudinal wetland survey (Fig. 7) revealed vertical low resistance 10 structures toward the wetland toe added support to the speculation that sub-surface flows of water through this highly conductive sandy system were buffered by clay-plugs. Furthermore, these were deemed to explain the lateral differences in hydrodynamics of the phreatic surface responses at different longitudinal reaches within the system (Riddell et al., 2007). This was noted because the hydrodynamics of the wetland ap-15 peared to be partly de-coupled between longitudinal reaches of these wetlands where these clay-plugs had formed sub-surface hydrodynamic breaks in an otherwise sandy and hydraulically conductive system. Similar responses to this were also noted during the 2007-2008 hydrological season where this hydrodynamic break was observed between groundwater levels as recorded at the hydrometry stations. Figure 8 reveals the 20 hydrodynamic responses in paired piezometers to rainfall over this season. It is quite clear that the ground water table has a short lived elevation at the most head-ward reach of the wetland at T1 3, whilst moving downstream to T2 2 the perpetuity of the groundwater level is more sustained. This is to be expected since the contributing area of the catchment and the wetland to this point is considerably greater than at T1 3. 25 However as one moves downstream this expected perpetuity of groundwater level elevation at T2 3 is rescinded earlier than at the upstream T2 2, and in fact the water table here declines prior to that at T2 2. This is attributed to the location of T2 3 adjacent to the erosion gully head, which facilitates a drawdown of water from the wetland at this HESSD Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version Interactive Discussion location, and this may be noted by the fact that the water levels in both piezometers at T2 3 are considerably lower than at T2 2. It would however also be expected that this erosion gully would facilitate a hydraulic drawdown that would also be felt at the nearby T2 2. Despite this the water table elevations at T2 2 remain quite close to the surface long after the piezometers at T2 3 have begun to dry out. This supports the notion that 5 the presence of the clay-plug as seen in Fig. 7, buffers the lateral sub-surface flow of water between T2 2 and T2 3.

3-D surveys
3-D geophysical surveys where used to analyze the distribution of clay bodies within the Manalana wetland and IP surveys were of particular use. 3-D pseudosections are 10 displayed in Fig. 9 for the series of parallel transects orientated SE-NW longitudinally through the wetland. Here one observes the relatively low chargeability of the material within the vicinity of the pseudosection. However within this material three other factors are apparent. First, the chargeability time increases with distance away from the origin, particularly at a distance beyond ∼55 m, and this increases significantly towards 15 the surface of the profile beyond this point. The second observation to consider is the sinuous curve made by the material with a chargeability of between 1-2 m s at around 55 m. The final and rather striking observation is the arch-like appearance of extremely high capacitance (∼0.01-0.03 m s) material from the origin to ∼50 m. The vertical sections of this arch-like structure correspond to the vertical low resistivity material 20 mentioned in the discussion of Fig. 7. Furthermore, the chargeability distribution does not appear to change within the 4 m breadth of the profile, except for a slightly lower chargeability region (∼0.03 m s) at around 20 m at 4 m breadth. Figure 10 displays the same transect except in terms of the sub-surface resistivity distribution, and here similarities emerge as would be expected. These being for instance the profile in Fig. 10 25 which is dominated by conductive material generally no more than 1000 Ω m, which conforms to the high capacitance material of Fig. 9. Meanwhile a band of high resistance material (>1000 Ω m) is encountered in Fig. 10

Printer-friendly Version
Interactive Discussion the high capacitance arch is discernable in Fig. 8, approximately between 5-35 m in Fig. 10. Furthermore, higher resistance material (>1000 Ω m) is observed beyond 70 m near the surface in Fig. 10 when compared to the rest of the pseudosection, corresponding to the lower capacitance material in Fig. 9. However the major difference between the two images relates to the clearer definition of an arch-type distribution of 5 material in Fig. 9, this is not revealed in Fig. 10. The IP profile of the SW-NE orientated transect (Fig. 11) has a high capacitance material throughout the profile, seldom exceeding 0.1 m s. Hence the inference from this image is that this part of the wetland has a particularly clay rich profile, the full length of the wetland, contrasting with the larger arm of the wetland seen in Figs. 9 10 and 10. Of particular note in Fig. 10 is the very high capacitance (<0.01 m s) material seen from the point of origin to ∼25 m. This material corresponds with the vertical band of material seen near the origin in Fig. 9. Furthermore, one observes a vertical band of high capacitance material (<0.01 m s) at ∼30 m extending from the surface to 5 m deep. However, despite the inversion of the data using a robust method with optimized damp- 15 ing factors, the pseudosection was achieved with a relatively high root mean square (RMS) error, and hence this image must be interpreted with caution. Meanwhile, the satisfactory inversion of the corresponding resistivity image for this section aids in the interpretation of this anomalous IP output. This resistivity data is displayed in Fig. 12, where one observes a lateral stratification of the sub-surface media within this profile, 20 particularly a band of material of <1000 Ω m between 693-698 m a.s.l., similarly this banding is also discernable in Fig. 10 for the complementary transect in the alternative orientation. In both cases this low resistivity/high capacitance material overlies a slightly higher resistance material at depth (note the material under-arch in Fig. 10). Moreover, in Fig. 11 one observes the slight reduction in material capacitance below 25 697 m a.s.l., beyond 30 m. However, Fig. 12 does reveal the anomalous nature of the vertical band of high capacitance material at 30 m seen in Fig. 11, which does not tally with the higher resistant material observed at this location in the resistivity profile. One possible reason for this anomaly, speculatively speaking, may be due to the presence HESSD 7,2010 A geophysical analysis of hydro-geomorphic controls Interactive Discussion of cracked clay in the profile, which due to the dry winter period at the time of surveying is highly likely. Therefore, although the clays themselves will have high capacitance, if they are cracked and hence have very large inter-pore spaces may engender large resistivity readings due to greater air to pore ratios in the soil medium than under wetter conditions.

5
The results of IP surveys using the gridded 21 electrode method at the wetland footslope interface revealed the highly variable nature of the material here in terms of its chargeability. Figure 13 displays this data in two orientations around the origin. This variability is reflected in the stratification of chargeability layers and heterogeneities in the near surface. It must be remembered here that due to the short electrode array 10 and short spacing of these electrodes the vertical depth penetration is reduced, and hence the resulting survey provides a resolution to a depth of 3 m. Nevertheless the resulting output reveals the patchy network of low capacitance (>6 m s) material near the surface and this reflects the disturbed material at the wetland surface that has been mechanically altered to create raised bed and furrow systems used for cultivation 15 purposes. Meanwhile, one is able to discern that this unconsolidated material overlies a deeper layer of rather high capacitance material (<1 m s), which varies in its depth across the 15 m×15 m domain. Figure 14 displays the same data, except that an IP filter has been applied to display chargeability data only below 1 m s, which in essence allows for visual representation 20 of only the high capacitance material at the wetland-footslope interface. The resulting pseudosection reveals that although this high capacitance material is somewhat laminar there is an obvious increase in depth to this layer with increasing distance from the point of departure of the grid from the origin. With this origin being at the wetlandfootslope interface, one is therefore able to note that this material is protruding from the 25 hillside into the valley bottom, with a descent in the N, NE and E direction.
In order to verify the output of the IP profiles, whose characteristically high capacitance regions suggest the occurrence of clays, a rapid and random series of auger holes were dug to a depth were clay was observed and defined through a bolus test  (Tongway and Hindley, 2004), this represents a qualitative ground truthing of the survey area. The locations of the random points in relation to the parallel transects are displayed in Fig. 15. Furthermore, Fig. 15 also displays an image of the face of the actively eroding gully headcut (wet season March 2006), whereupon the saturation of the soil profile reveals two distinct horizons within the wetland substrate, a shallow sandy 5 loam horizon above a deep fine clay horizon. Since the SE-NW transect (Fig. 9) shows that from the point of origin to approximately 50 m the wetland is made up of a high capacitance material (<1 m s) and this is relatively consistent from the surface throughout the entire depth of the profile. The fact that clays were encountered within 1 m of the wetland surface at these locations support the inferences made about clay distribution 10 from the IP surveys. An analysis was also undertaken to verify IP interpretations by using measures of saturated conductivity (K ) at the hydrometric monitoring stations. This verification is considered semi-quantitative since, as will become apparent, repeated bail tests could not be undertaken due to the extremely slow recoveries of the piezometers. Hence, 15 the displayed K values of Fig. 16 are single measurements and therefore treated as estimates of K at the various piezometer depths. Figure 16 displays the closest measured IP (m s) and apparent resistivity (Ω m) value at each depth corresponding to the depth of the piezometer well at each hydrometric monitoring location (Fig. 2). In cases where K was determined at depths >2000 mm, the K is extremely low, implying that 20 water levels in the wells would take hours if not days to recover to equilibrium levels prior to conducting bail tests.
When resistivity data is plotted against log values of K there appears to be good agreement between the estimated conductivity of the material at the respective piezometer depths and the measured apparent resistivity of the same material for all 25 wells (note there is only one well at MP1). In each case there appears to be a decline in resistance (or increased conductance) with K at depth, which tapers off below a depth of 4000 mm. The situation with the IP data does not yield the same clear relationship, although trends may be observed. Firstly, there is a general decrease in K of

HESSD Introduction
Conclusions References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion the wetland material whilst capacitance increases (chargeability time decreases), up to 4000 mm depth. Second, at depths beyond 4000 mm K increases slightly at piezometer locations T2 2 and T2 3, and at MP1 capacitance and K tally. Therefore, there is a trend between these two variables of, K and chargeability, however the clarity of their relationship is concealed by the very small chargeability range that is plotted on 5 the secondary Y-axis. Obviously, more repeated K tests and precise determination of resistivity and chargeability would be ideal in order to verify the ERI and IP output. However this was negated by conditions in the field i.e. very low conductivities, and output of measured apparent resistivities, where the readings extracted from inversionmodel output would not have been in precisely the same location as the piezometer 10 wells. The addition of 3-D combined IP and resistivity surveys to the 2-D scoping surveys and supporting hydrometry data have allowed a real insight into the hydro-geomorphic controls within this wetland system and facilitated a proposed model for the development of these controls, particularly in terms of clay-plug formation. The rationale for 15 the proposed model will now be outlined.
Remembering that the SE-NW orientated parallel transects (Figs. 9 and 10) situated in a larger region of the wetland (refer to Fig. 2) show that the chargeability time in the wetlands sub-surface increases with distance NW away from the origin. Meanwhile the SW-NE orientated transects within the smaller wetland region displays an 20 extremely high capacitance all the way along this part of the wetland. This provides for the first point of interest. The suggestion is that the smaller wetland region has a more uniform clay distribution in the sub-surface than the larger region, and the latter therefore is likely comprised of increasingly coarser (higher resistance, lower capacitance) material away from the juncture of these two regions (tributaries) of the wetland. 25 Furthermore, it was noted that there was a distinct down sloping of higher capacitance material with distance away from the hillslope-wetland interface. These two observations facilitate the notion that finer (clay) particles are removed from the hillslope units through the process of elluviation, which is a common process on these geological HESSD Introduction

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion terrains in semi-arid settings, particularly in the case of clay enriched sodic sites of the region (e.g. Khomo and Rogers, 2005). This elluviation allows for their deposition (or illuviation) as distinct clay rich horizons at the hillslope toes, which in this catchment have been described according to the South African soil taxonomic system as a Kroonstad soil form (Soil Classification Working Group, 1991). These soils translate to the 5 World Reference Base (FAO, 1998) system as planosols. This elluviation process has probably been more concentrated within the smaller wetland region due to the higher capacitance and more uniform distribution of this material within this region. This has effectively created a barrier across the confluence with the larger wetland region. Furthermore the confinement zone, or valley pinches that are observed at this confluence 10 has also contributed to this barrier creation, and one observes this process when interpreting Figs. 9 and 14 to their respective positions at this confluence. Indeed the arch-like structure obvious in Fig. 9 suggests an extremely dense clay formation (clayplug) that could have been created by two (or more) elluvial pathways at this juncture.
Effectively what this scenario creates is a zone of infilling by coarser clastic sediments 15 into the larger wetland region from the broader contributing catchment. Indeed the soil types in this larger region of the wetland have been described as hydromorphic soils of the Katspruit form, which accordingly translate to gleysols (FAO, 1998) which typically develop under wet conditions in unconsolidated materials. This pedological evidence lends credence to a hypothesis proposed here that val-20 ley bottom (wetland) infilling, by unconsolidated colluvial and alluvial material, has occurred behind these clay-plugs. This proposed model therefore implies that there is a lateral as well as vertical removal of finer materials in the slopes of these wetland catchments and their concentration at zones of valley confinement. The proposed model is depicted schematically in Fig. 17 and photographically in Fig. 18. 25 Based on the observations and measurements presented in this manuscript, it is known that the Manalana wetland is underlain by a deep clay material (>4000 mm) with a moderate hydraulic conductivity in relation to an overlying dense clay with extremely low hydraulic conductivity (somewhere between 4000 mm-2000 mm). At considerable HESSD Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version Interactive Discussion depth, certainly beyond 6000 mm the wetland is bounded by bedrock and saprolitic material. It is also known that the clays have a particularly high density at the confluence of the two arms of the wetland, adjacent to the clay-plug. In addition, the near sub-surface of the wetland is largely a sandy horizon with high hydraulic conductivity, and in the smaller arm (SW-NE) of the wetland this horizon is very shallow such that 5 this section is predominantly underlain by clay material, that through lateral illuvial processes, and illuviation of clays from the hillslope at valley constriction (see Fig. 18) has effectively created an illuvial barrier across the confluence with the larger wetland region (SE-NW). Furthermore, as one moves upstream along the larger wetland region, material becomes less consolidated giving way to a deeper sandy matrix. 10 Understanding of climatically induced evolution of wetlands in southern Africa is outlined in , as being intimately linked with glacial and interglacial periods of erosion dominated cycles attributed to the warmer and wetter conditions of the latter, and deposition dominated dryer cycles of the former. Furthermore, the clarity of erosion terraces seen within the Manalana catchment reveal that this is indeed the 15 case and the process of cut-and-fill valley evolution in the Manalana has been discussed by , and this reciprocates the findings discussed here, as valley bottom in-filling is noted through our interpretations presented here.
Given the fact that the Manalana represents one of many sub-catchments at the Sand Rivers headwaters which are all in a similar state of degradation the identification 20 of hydro-geomorphic buffers/barriers, such as clay-plugs is crucial. This relates to the concept of catchment connectivity. Whereby, the connectivity of process domains from coupled hillslope-valley bottom wetlands and longitudinal through-wetland processes, as in the case of Manalana catchment and others, are increasing significantly due to degradation process such as erosion gullying. This relates to a concept described by 25 Fryirs et al. (2006) where, the identification and understanding of the hydro-geomorphic processes that buffers/barriers (be they rock outcrops or differential sediment depositions) facilitate in terms of switching on or off these coupled processes, within a fluvial network, is important for their successful management. Therefore in light of the al-

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion ready determined hydraulic function that the sub-surface clay-plugs have in terms of ameliorating lateral throughflows in the wetland, and the possible identification of hydrogeomorphic process zones that engender the development of these clay-plug buffers, one should now be able to identify zones within these catchments where they are likely to occur (cryptically). This therefore provides a means to protect these zones, in the ab-5 sence of access to technical geophysical equipment, one proposes that to reveal and delineate the location of these clay-plugs, identification of regions of valley constriction be made, which may then be analyzed qualitatively through in-situ soil analysis for their clay content. Beauvais et al. (1999) successfully used ERI to describe hydro-geomorphic pro-10 cesses on other granitic terrains in Senegal by identifying both mechanical and weathering induced drivers of hillslope geomorphology. The ERI approach here using additionally, IP, has similarly assisted in the development of a conceptual model for the evolution of hydrogeomorphic controls in the wetlands of the Sand River system. However it is interesting to note that even in a relatively recent sedimentology study (Baines 15 et al., 2002) making use of similar technology highlighted the lack of a due procedure to be followed when undertaking such geophysical surveys for geomorphic understanding. Despite this (Baines et al., 2002;Loke 1999) advocate the use of qualitative ground truthing to verify geophysical interpretations, and the verification approaches used in this study were highly valuable in this respect. This is an issue that should be devel-20 oped and refined in the future, particularly where the use of geophysical methods for low cost rapid hydro-geomorphic interpretation (outside of the fields of geological and mineral prospecting for instance) of wetlands and other landscape units is required. It is therefore advocated here that the combined use of ERI and IP be used, were possible, in cases such as wetland rehabilitation planning. This is particularly pertinent 25 to South Africa, a water scarce country, where wetland rehabilitation is common place and where hydrological and geomorphological data relating to the wetland is often lacking. Moreover, it is widely accepted that these geophysical approaches add huge value to understanding hydrological process data in research catchments, in which they aid  (Wenninger et al., 2008). Hence the data discussed here will be of real benefit for the development of a hydrological process model of the Manalana wetland and its sister catchments in the future.

Conclusions
The use of the geophysical methods of ERI and IP have shown their suitability for de-5 scribing hydro-geomorphic structures within wetland catchments. In particular these relatively rapid approaches are envisaged to have application in the future for extrapolating hydrological regimes and hydro-geomorphic controls in wetland systems, this has important hydro-ecological and economic implications for the plethora of wetland (and river) rehabilitation projects now underway in South Africa where robust conceptual 10 models of these wetland processes are required to facilitate appropriate and successful rehabilitation measures. In this regard the use of this approach allowed for the identification of crucial geomorphic processes that control catchment connectivity in this region. In particular the surveys allowed for the delineation of clay-plug sub-surface hydrological buffers and provided information on their modes of formation, albeit by 15 inferring where they will likely occur with respect to longitudinal valley position. The findings will now enable the future protection of these zones from externally derived degradation processes and it is envisaged that this methodology be suitably applied in the future, particularly for wetland rehabilitation planning. Advancing process-based watershed hydrological research using near-surface geophysics: a vision for, and review of, electrical and magnetic geophysical methods, Hydrol.