2.1 Study areas and fieldwork

Four sloping fen areas located in alpine or peri-alpine regions of Switzerland (Table 1) were identified for this study. All areas are situated in protected fen areas, and their selection was based on two main criteria:

1. the subsurface water flow must occur only in the topsoil layer and as runoff (as described in the introduction), and

2. roads constructed with either a no-excavation, an L-drain, or a wood-log structure must be present.

To fulfill the first criteria, soil profiles were analyzed to ensure that each area with different road types had the comparable soil stratigraphy: it had to be composed of organic soil on top of a layer of impermeable clay and similar hydraulic regimes (e.g., runoff and subsurface flow occurring only in the topsoil layer). In addition, to ensure that subsurface water is forced to cross the road instead of flowing parallel to the road (and thus not being directly affected by the road), another important criterion for the selection of the study areas was that the subsurface flow was perpendicular to the road.

Table 1Field site locations and features.

Download Print Version | Download XLSX

To evaluate the hydraulic connection provided by the roadbed structures, tracer tests were carried out. As illustrated schematically in Fig. 3, the upslope area was irrigated with a saline solution and the occurrence of the tracer was monitored downslope of the road. In the absence of surface runoff, the occurrence of a tracer downslope demonstrates the hydrogeological connection through the road. Furthermore, the spatial distribution of the tracer front reflects the heterogeneity of the flow paths.

Figure 3Schematic view of the sites analyzed during fieldwork.

Download

At each field site, an area of an 8 m × 20 m rectangle that included a 2.5 to 3.5 m wide road segment was selected. A network of approximately 30 mini-piezometers was installed on both sides of the road (Fig. 3) to monitor the hydraulic heads and was used to obtain samples for the tracer test.

The mini-piezometers are high-density polyethylene (HDPE) tubes no longer than 1.5 m (i.d. of 24 mm). Each tube was screened with 0.4 mm slots from the bottom end to 5 cm below ground level. They were inserted into the soil after extracting a core with a manual auger (diameter of 4–6 cm). The gap between the tube and the soil was filled with fine gravel and sealed on the top with a 4 cm thick layer of bentonite or local clay. Hydraulic heads were measured using a manual water level meter (±0.3 cm). At each point, the terrain and the top of the piezometer were leveled using a level (±0.3 cm), whereas the horizontal position was measured with a tape measure (±5 cm).

The tracer tests were conducted using two oscillating sprinklers designed to reproduce a 30 mm rain event over 2–3 h. This is equivalent to an intense rain event. Prior to the experiment, the sprinklers were activated for 15–60 min to wet the soil surface. Sodium chloride was added to the irrigated solution to obtain an electrical conductivity of 5–10 mS cm⁻¹ which is approximately 10 times higher than the natural electrical conductivity of the groundwater. Subsequently, the area (60 m²) upslope of the road (upslope injection area of Fig. 3) was irrigated with the salt solution using the two sprinklers. The electrical conductivity (EC) of soil water was manually measured using a conductivity meter in all mini-piezometers prior to the experiment, immediately after the experiment, and 24 h after the experiment. An increase in EC in the piezometers located in the downslope area indicates that the injected saltwater flowed from the upslope area to the downslope area below the road and clearly shows a hydraulic connection. Conversely, if no changes in EC are observed in the piezometers, the hydraulic connection between the upslope and downslope of the road is affected.

2.2 Numerical modeling

The modeling approach was structured in three steps. First, a 3-D base case model representing surface and subsurface water flow in a sloping fen was elaborated. Subsequently, the base case model was modified to represent the three different types of road structures. For each model, various slopes, soil, and road drain hydraulic conductivities were implemented to produce a sensitivity analysis and explore their sensitivities in the sloping fen flow dynamics (see Sect. 2.2.3 for details).

2.2.2 Conceptual models and model implementation

Figure 4 illustrates the conceptual model of each case. Existing engineering sketches were used as a basis for the implementation of the drain and road. Geometry, topography, and slopes are based on the conditions in the field. In each model, the soil layer has a thickness of 0.4 m and the surface and subsurface water originate from precipitation only. The upslope boundary is the catchment boundary (water divide) and the downslope boundary represents the outlet of the model. Finally, it is assumed that the layer beneath the soil is impermeable (as observed in the field). One Neumann (constant flux) boundary condition was used on the top face for simulating precipitation. A constant head (Dirichlet-type) boundary condition equal to the ground surface elevation (2 m) was used on the lowest cells of the slope (x=76 m in Fig. 5a) allowing groundwater to flow out of the model. Finally, a critical depth boundary condition which allows surface water to flow out of the model domain was implemented on the top nodes located at x=76 m. All other faces are no-flow boundary conditions.

Figure 4(a) Base case, (b) no-excavation, (c) L-drain, and (d) wood-log structure conceptual models. BC refers to boundary conditions.

Download

Figure 5Model development: (a) base case model, (b) base case model cross-section between 61 m <x and x<66 m, (c) no-excavation model between 61 m <x and x<66 m, (d) L-drain model between 61 m <x and x<66 m, (e) L-drain model between 61 m <x and x<66 m along the transversal drain, and (f) wood-log model between 61 m <x and x<66 m.

Download

A 3-D finite element mesh was developed (Fig. 5a). The mesh was 76 m long in the x direction, 20 m in the y direction, and the mesh thickness was 1.2 m. The top elevation was fixed at 2 m on the right side (x=76 m) and varied from 9.6 to 24.8 m on the left side (x=0) according to the slope of the model. The mesh was composed of 24 layers, 127 200 nodes, and 118 440 rectangular prism elements. To guarantee numerical stability, mesh refinements were implemented. The element size varied between 2 and 0.1 m horizontally (in the x and y directions) and 0.09 and 0.06 m vertically.

The base case model and the three other models representing different road types have the same boundary conditions and finite element meshes; however, modifications were made between coordinates $\mathrm{61}<x<\mathrm{66}$ for the implementation of the different road types. Figure 5 depicts the differences between the base case model (Fig. 5a, b) and models with roads (Fig. 5c, d, e, f). In models simulating a road, the mesh and the material properties were adjusted. The fine spatial discretization of the mesh created between the coordinates $\mathrm{61}<x<\mathrm{66}$ allows a more accurate representation of the simulated processes where high hydraulic gradients are expected (near roads and drains).

2.2.3 Model application

The model application consists of the variation of model properties to assess their effect on the groundwater dynamics. The following parameters were analyzed: fen slope, soil hydraulic conductivities, and road drain hydraulic conductivities. These parameters were selected because according to Darcy's law (Eq. 1) they control the groundwater flow dynamics:

$$\begin{array}{}\text{(1)}& q=K\cdot \mathrm{\nabla }H,\end{array}$$

where K is the hydraulic conductivity of the soil and the drain, and ∇H is the hydraulic gradient of subsurface water in the fen, which itself strongly influenced by the topographical slope.

For each property varied in the sensitivity analysis, three different values were chosen (Table 2): a low, intermediate, and high value. For the soil hydraulic conductivities (KS), values presented in Chambers (2003) were used and varied between 8.64 and 0.0864 m d⁻¹. This corresponds to a soil composed of gravely organic matter (as observed for example at the Sankt-Antönien site) or loamy organic matter (as observed for example at the Schöniseischwand site). The van Genuchten parameters (α and β), as well as the residual water content, were not varied. The road drains (KD) which are made of coarse or very coarse gravel were assigned a hydraulic conductivity between 8640 and 86.4 m d⁻¹ (Fetter, 2001), with their van Genuchten parameters corresponding to gravel. The slopes were fixed at 10 %, 20 %, and 30 %, as observed during fieldwork. The hydraulic conductivities of the wood-log (W-L) drain were assumed to be 10 times more conductive and more porous than the gravel drain. The road concrete is almost impermeable; thus, it was conceptualized with a very low hydraulic conductivity, with its van Genuchten parameters corresponding to fine material. The road base is constructed using highly compacted fine material (sand and loam); thus, it was implemented with low hydraulic conductivity, with the van Genuchten parameters corresponding to fine material. Finally, the implemented soil and road surface flow properties correspond to a wetland and urban cover (Li et al., 2008).

Table 2Subsurface and surface flow parameters.

Download Print Version | Download XLSX

In order to simulate each parameter combination, a total of 90 models were developed (27 models for each road structure and 9 models for natural conditions). Models were run for 10 000 d (about 27 years) with a constant flux equal to 380 mm yr⁻¹ on the top, representing the rainfall to reach a steady state. Subsequently, subsurface flow rates in the soil layer were extracted at each section with an area of 0.4 m² (1 m wide times the soil thickness) presented in Fig. 6. Changes in subsurface flow rates indicate a perturbation of flow dynamics; therefore, a comparison of flow rates between each model was made to present the effect of each road structure and sloping fen properties on the dynamics.

Figure 6Location of observation sections in the models.

Download

3.1 Fieldwork

Based on the observations, all sites show a continuous saturated zone before the experiment, both upslope and downslope of the road, with the hydraulic gradients being similar to the terrain slope (Fig. 7, first column). In contrast, the EC maps established prior to the tracer test show a spatial variability across one to several meters (Fig. 7, second column). Within each plot, EC varies from 482 to 629 µS cm⁻¹. At the SCH site, the highest values are located downslope of the L-drain outlet which could indicate that the EC increases as water is flowing through the drain (e.g., through the dissolution of the construction material). Given that this initial distribution of EC is not uniform, the comparison of EC after the sprinkling experiment has to be made in a relative manner (Fig. 7, third column).

Figure 7Fieldwork results at the four field sites: the first column shows the measured groundwater heads before the tracer test, the second column shows the measured EC before the tracer test, and the third column shows the before and after tracer test differences in EC. The hydraulic head downslope of the road at the Stouffe site is about 25 cm, whereas that upslope of the road at the Schöniseischwand is about 225 cm (between two isolines); these values are not presented in the figure.

Download

The heterogeneity of the hydraulic conductivity of the soil is apparent from the tracer tests (Fig. 7, third column: EC 24 h after injection). At all four sites, the front of the saline solution is not uniform due to the heterogeneity of the soil hydraulic conductivity. Nevertheless, the road structures or the drains may create preferential flow paths. This clearly occurs at the SCH site, where the front follows two preferential flow paths: one related to the L drain (right path) and the other unrelated to the L drain (left path). This suggests that the latter drains only a part of the water and that the remaining water follows a natural preferential flow path. At the HMD site, the saline solution is far more concentrated on the left side of the plot; however, this is apparently not as a result of the road's structure. Rather, the soil appears more permeable on the left side of the plot, both upslope and downslope of the road. Finally, the decrease in EC observed 24 h after injection at some locations might result from the following: (1) the tracer injection induces the displacement of a small volume of local water with a lower EC, via “piston effect”; (2) the tracer injection was preceded by a period of irrigation without tracer. This could have diluted the pre-irrigation soil solution.

In each case, the irrigation experiments demonstrate the continuity of subsurface flow under the road for all structures. For the no-excavation and wood-log type, the perturbation of the flow field seems to be controlled by the natural heterogeneity of the soil and flow paths, and not by the road itself. Conversely, the field data suggest that the L drain constitutes a preferential pathway. This flow convergence can cause gully erosion.

3.2 Modeling

Figure 8a shows the results of the models with a slope of 10 %, Fig. 8b shows the results for a slope of 20 %, and Fig. 8c shows the results for a slope of 30 %. In each panel, the groundwater flow rates (always in cubic meters per day, m³ d⁻¹) are plotted using crosses for the base case model, diamonds for the no-excavation model, squares for the L-drain model, and circles for the wood-log model. In addition, the maximum flow rate capacity of the soil calculated with Darcy's Law (Eq. 1) and the flow rate induced by precipitation are also presented for the interpretation of the results. In the following paragraphs, the base case (natural conditions) results are presented and discussed, followed by the simulations of the road structures.

Figure 8Simulated groundwater flow rates 2 m downslope of each road structure and each parameter combination with a slope of (a) 10 %, (b) 20 %, and (c) 30 %. Numbers in the bottom right corner of each panel represent the ratio between the maximum and minimum groundwater flow within the L-drain transect.

Download

In the base case model, groundwater flow rates vary from 0.003 to 0.069 m³ d⁻¹ for a 10 % slope, from 0.006 to 0.069 m³ d⁻¹ for a 20 % slope, and from 0.009 to 0.069 m³ d⁻¹ for a 30 % slope. The groundwater flow rate decreases following a decrease in the hydraulic conductivities (KS) of the soil layer. The groundwater flow rates are mainly controlled by the hydraulic conductivities, and the slope plays a less important role. This is expected, as the ratios of the maximum and minimum hydraulic conductivity span 2 orders of magnitude, while slopes were multiplied by a factor of 2 (for a slope of 20 %) or 3 (for a slope of 30 %). Therefore, groundwater flow is increased by a factor 3 between the model KS3 with a slope of 10 % and model KS3 with a slope of 30 %. Concerning the formation of surface flow, the following observation can be made: for all KS2 and KS3 models, surface flow occurs, while the infiltration capacity of the KS1 models is never exceeded and, thus, no surface flow occurs.

In the no-excavation and wood-log models, the influence on flow rates caused by the presence of the road structures is quite similar. Groundwater flows vary from 0.01 to 0.069 m³ d⁻¹ for a 10 % slope, from 0.01 to 0.069 m³ d⁻¹ for a 20 % slope, and from 0.010 to 0.069 m³ d⁻¹ for a 30 % slope. Compared with the base case model, results show that the no-excavation and wood-log structures have a minimal impact on flow perturbation. The only marked difference is that groundwater flow rates are slightly higher if the soil hydraulic conductivities are low (KS3). This is due to the hydraulic conductivity of the base of the road (consisting of wood logs) being higher than the hydraulic conductivity of the soil which facilitates infiltration. Conversely, in the base case model, less water infiltrates and more surface runoff occurs. In the 20 % and 30 % slope models, the results of the no-excavation model are similar to the base case model.

Figure 9Extent of perturbations due to the L-drain road type: simulated groundwater flow rates at different distances from the road.

Download

In the L-drain model, the effect of the road is markedly different from the other road structures. The groundwater flows vary significantly with respect to the observation sections. The maximum flows are always obtained in observation section G (see Fig. 6 for the location of the sections) just downslope of the drain outlet and can be 10 times higher than the base case. Conversely, minimum flows are obtained in observation sections C and D in which flow rates can be 10 times lower. Significant differences in groundwater flow are also observed in the same transect (within the same model). To condense this information, the ratios between maximum and minimum flow rates are calculated for the L-drain structures (numbers in the bottom right corners of the panels in Fig. 8). The maximum differences are observed for the cases where the hydraulic conductivity of the soil (KS) and drain (KD) are high and vary from 0.025 to 0.150 m³ d⁻¹. Conversely, when KS and/or KD is low, the differences along the transect are smaller. Finally, the slope accentuates the groundwater flow rate differences along the transect. Therefore, an increase in the groundwater flow differences is observed for the 10 % and 30 % slope scenarios, within the same model. The impact of the L drain may be further explored by extracting groundwater flows lower than 2 m downslope of the road to assess the extent of perturbations. Figure 9 shows simulated groundwater flows for the most critical cases (i.e., KS1 with a slope of 10 %, 20 %, and 30 %) downslope of the road at 3.5 and 6.5 m, respectively, and 2.5 m upslope. At 3.5 m downslope, groundwater flow regains the upslope conditions. At 6.5 m downslope of the road, all observation sections are very similar to the upslope flows except in section G where flows are still slightly higher.

Figure 10Simulated surface flow of the KS2–KD2 model and a slope of 20 % for each road structure (min. velocity =0 m d⁻¹ and max. velocity =0.25 m d⁻¹). The results clearly indicate the increased risk caused by the L drain with respect to triggering surface runoff and, in turn, potential gully erosion and sections of the wetland drying out.

Download

In addition to the assessment of perturbation due to roads, the model results can be used to evaluate the risk of gully erosion. As presented in Fig. 8, the maximum flow rate capacity of the soil is small in comparison to precipitation. For all model scenarios except for KS1, the soil capacity is lower than the precipitation and, thus, surface runoff occurs in the models and is likely to occur naturally. However, surface runoff may be triggered by the presence of L-drain structures. To illustrate this process, the simulated surface flow velocities of each road structure downslope of the road for the model KS2–KD2 and a slope of 20 % are presented in Fig. 10. In this case, the maximum flow rate capacity of the soil is approximately equal to precipitation, and therefore runoff should not occur. However, this is not the case for the L drain. The occurrence of surface runoff is a consequence of the subsurface flow concentration. In this configuration, the infiltration capacity of the soil is too small to accommodate the concentrated flow collected upslope, thus groundwater emerges and surface flow is triggered. This constitutes an increased risk of gully erosion. In addition, the perturbation of roads upslope of the road was assessed.

Finally, the impact of road structure on the upslope road dynamics was also assessed (figure not shown) 2.5 m upslope. Upslope flows are similar to the base case model; thus, the influence of the road is, not unexpectedly, marginal for all road types.

The development of models with various combinations of parameters allowed for the exploration of a larger parameter space than fieldwork alone. For instance, the fact that the impact of an L-drain structure on the water dynamics is less marked if the hydraulic conductivity of soil is low would have been impossible to identify using just fieldwork. However, a numerical model is always a simplified reproduction of reality. The main model assumption is that the hydraulic conductivity of the soil is homogeneous – as opposed to the field conditions analyzed. However, the models are not intended to reproduce small-scale observations, i.e., the exact hydraulic head in a piezometer, but instead can be used to explore the influence of the road structures under different soil conditions (bulk hydraulic conductivities and slopes). Given that no heterogeneity-induced horizontal redistribution of the flow downslope can be simulated using homogeneous soil conditions, the models constitute a worst-case scenario. It is a worst-case scenario because we exclude the possibility that a fraction of the drained water could be horizontally redistributed downstream through natural heterogeneity, thereby potentially reducing the negative impact of the road. Therefore, the models allow for a relative ranking of the potential impact and clearly show the increased risk of surface water flow generation and, in turn, gully erosion. For the scenarios investigated, the L drain consistently shows the largest impact. Thus, the other two road structures are the preferred construction types.