UPDATES: March 2015

UPDATES is an email newsletter on stormwater management, assessment (including monitoring), and maintenance research at St. Anthony Falls Laboratory and the University of Minnesota.

Infiltration into Roadside Drainage Ditches

March 2015 (volume 10 - issue 2)

Contributed by Maria Garcia-Serrana, Ph.D. student, St. Anthony Falls Laboratory, Department of Civil, Environmental and Geo-Engineering, University of Minnesota; Advisors: John S. Gulliver, Professor, Department of Civil, Environmental and Geo-Engineering, University of Minnesota; and John L. Nieber, Professor, Department of Bioproducts and Biosystems Engineering, University of Minnesota.

Funded by Minnesota Department of Transportation, Minnesota Local Road Research Board


Most of our roadside drainage ditches act as a dry grassed swale, where water is infiltrated and solids are settled as the water flows over the side of the ditch and down the center. Documentation is needed, however, about the amount of water infiltration that occurs within these ditches before they can be given full pollution prevention credit. This research project is designed to document infiltration performance and establish methods by which the roadside drainage ditches can be assigned this credit.

Infiltration rates are currently calculated by assuming that water flows as sheet flow over the side slope of swales. However, water flow, when limited, occurs in narrow and shallow micro-channels and concentrates in depressions rather than flowing as sheet flow. Only a fraction of the soil surface is covered with water.

No previous studies have analyzed the effect of fractional coverage of water on infiltration rates. Nevertheless, the non-uniform distribution of water along a hillslope directly affects infiltration. The objective of this study is to quantify the wetted contact surface and determine how partially wetted areas affect the amount of volume infiltrated. If water infiltrated in a one-dimensional vertical direction due to gravity, the rate of infiltration would be directly proportional to the wetted area. Alternatively, if water infiltrates vertically due to gravity and laterally due to soil capillary suction, the infiltration rate will be higher than the wetted area would indicate.

Both laboratory and field experiments have been conducted to investigate the effective contact surface for infiltration modeling. Results will be used to develop an overland flow model that accounts for partially wetted areas in shallow concentrated flows.


First, a full-scale model of a road and side slopes was built at the St. Anthony Falls Laboratory. The goal was to determine the impact that fractional wetting of the slope surface has on infiltration rates. A number of tests were performed with different initial conditions related to the micro-topography of the surface. To relate overland flow processes on hillslopes with micro-topographic conditions, the surface of the side slope was monitored through the use of a high resolution laser system to obtain micro-topographic data. To document locations with water on the surface, the water was mixed with titanium dioxide (TiO2), which allows the laser light to be reflected on the water surface (Legout et al., 2012).

A flume representing the cross section of a swale 1:6 slope was used to perform the laboratory experiments. The flume consisted of impervious surfaces representing two road lanes of 12 ft (3.66 m) each, a 2% slope impervious surface representing a shoulder of 6 ft (1.83 m) and a side slope composed of soil. The width of the flume was approximately 3 ft (0.91 m).  On the side slope, the soil depth was 1 ft (0.3 m). A series of 8 perforated drainpipes, encased in a permeable geotextile, were located below the soil layer and oriented slightly in the downslope direction. The soil was loamy sand consisting of a mixture of 70% C-33 sand and 30% organic compost mix, compacted before each experiment. The initial soil moisture content was approximately 12% in all the tests.

Water supply was regulated with a valve and connected to a rectangular constant head tank located on the road lane closest to the side slope. A 1.1-in 24-hr storm exceeds 90% of rainfall events in the Twin Cities metropolitan area of Minnesota, USA. The appropriate intensity curve to use in this region of the country corresponds to the SCS, Type II storm. For that type of storm 90% of the precipitation occurs in one hour in the middle of the storm event. In light of this, the water applied to the slope was equivalent to a rainfall with constant 1 in/h (2.54 cm/h) intensity over the road. Every experiment lasted for one hour and the total input of water volume was 234 L (61.82 gal) for each experiment.

Three different initial surface roughnesses were tested. One initially smooth surface where the surface was smoothed with a plastering trowel. The second scenario involved making three incisions on the surface with a 1.8 in (4.6 cm) diameter steel pipe. The outcome was three parallel, semi-circular channels that were approximately 0.9 in (2.3 cm) in depth and represented rills (Figure 1). The third scenario consisted of five parallel rills, made with a similar technique.

Figure 1: Laboratory flume with 3 initial rills. Left: Initial conditions. Right: Final surface after 60 min of flow. (Photo: M. Garcia-Serrana)

The following data was collected:

  • Volume of water infiltrated in each drain pipe over time,
  • Total volume of runoff water at the base of the slope,
  • Micro-topography of the surface, and
  • Wetted surface area.

Second, four highways in the Twin Cities area, with different types of soils, were selected for a series of field studies (Hwy 13, Hwy 47, Hwy 55, and Hwy 77). The process consisted of: cutting the surface vegetation, collecting surface roughness data using a pin meter, and applying a measured discharge of water at the top of the side slope. To see the flow patterns, the water was mixed with a 13 g/l concentration of kaolin. The kaolin was added to the water in a drum and mixed with a paint mixer to achieve a uniform concentration.

For each experiment water was applied to the side slope for 30 minutes. The application rate was equivalent to a rainfall with constant 2.2 in/h (5.59 cm/h) intensity over a 32.8 ft (10 m) wide road and 3 ft (0.91 m) length section. A 1.1-in 30-min storm is a 2-year event in the Twin Cities area according to NOAA Atlas 14; additionally, this storm exceeds 90% of the total volume and 99% of rainfall events in the Twin Cities area. The total volume of water for each experiment was 255.4 L (67.47 gal).

The water patterns were recorded by installing a camera with a mounting pole set in the channel of the drainage ditch. The camera was programmed to take a picture every five seconds. A 1x1 m (3.28 ft) mesh frame was installed on the swale to facilitate corrections of angle distortions in the pictures taken (Figure 2).

Figure 2: Overland flow pattern in one of the sites at Hwy 13. (Photo taken by M. Garcia-Serrana)

The water not infiltrated by the swale was collected at the bottom of the side slope. The total amount of water that was not infiltrated (runoff) was recorded, as well as the runoff rate when the system reached steady state.

The following data sets were collected:

  • Micro-topography of the surface,
  • Total volume of runoff water (water not infiltrated in the swale),
  • Intensity of runoff water, and
  • Wetted surface area over time.


In the laboratory experiments the difference between an initially smooth surface and both cases with rills was around 10% more of water infiltrated for the initially smooth surface. The disparity between the scenarios with 3 and 5 initial rills was insubstantial. It should be highlighted that all the scenarios in the laboratory experienced erosion and rills on the surfaces, even those that had no rills initially.

The swales in the field did not show evidence of any new erosion during the experiments. The lack of new erosion can be attributed to the grass roots reinforcing the erosion resistance of the soil, and the fact that the soils have been in place and covered with grass for many years, resulting in increased organic matter and aggregation. By contrast, the soil in the laboratory experiments had no vegetions or roots, and was composed of fresh-packed and smoothed sand and organic matter.

Analysis of roughness and wetted area

In order to compare the roughness from different surfaces with one parameter, we used the Random Roughness factor (RR), which is the standard deviation of surface elevations. The RR reveals the vertical variability in surface elevations (Yang et al., 2013). The RR was calculated at uniformly spaced (1 mm) cross-sections using,


where z is the elevation at a given point, zmean is the mean elevation of a cross section, and N is the number of points in a cross section. The slope was subdivided into seven equal length cells, and calculations of mean RR and wetted area was determined for each cell. The relationship between the percentage of wetted area of each cell and mean Random Roughness is illustrated in Figure 3. The plot indicates that the percentage of wetted area tends to decrease as mean Random Roughness increases. Among all the experiments, only a small part of the swale surface was covered with water, with the maximum wetted area observed to be 42% of one of the cells.

Figure 3: Mean Random Roughness of each cell versus percentage of wetted area. Each data point represents an individual run.

Infiltration and Wetted area

The MPD infiltrometer method (Ahmed et al., 2014) was used to obtain the saturated hydraulic conductivity (Ksat) of the laboratory soil. The saturated hydraulic conductivities obtained with this technique had a geometric mean of 5.05 in/h (12.83 cm/h), and the effective wetting front suction head was 0.22 in (0.56 cm). This mean Ksat value is higher than the ones obtained in the field from swales located in the Twin Cities area. For example, the geometric mean of the Ksat calculated for the side slope of Highway 47 was 2.56 in/h (6.5 cm/h), and for Highway 51 was 0.78 in/h (2 cm/h).

Figure 4 represents the relationship between water infiltrated and percentage of wetted area in each cell for the laboratory and field experiments. More wetted area was associated with higher infiltration. The wetted area varied from 4% to 42% in the rilled surfaces at the laboratory and from 64% to 84% in the field. The slope of the line fitting the field data is smaller than for the laboratory; this could be due to the higher intensity of rainfall applied (1-year and 2-year return period) and the lower geometric mean Ksat.

Figure 4: Dimensionless percent water infiltrated versus percent wetted area in the lab (blue) and field (red) experiments.

Figure 5 shows the results of the percentage of wetted area and infiltration of the field experiments. For the cases of equal wetted area, the percentage of water infiltrated appears to be related to the saturated hydraulic conductivity of the soil. Further research needs to be done to link the difference in water infiltrated, for equal wetted area, to the capillarity of the soil as well.

Figure 5: Dimensionless percentage water infiltrated versus percentage wetted area in the field experiments. Geometric mean values of the saturated hydraulic conductivity measurements and %Infiltration at three different highways where the same percentage of wetted area was observed.

Based on the data obtained, the infiltration rate is linked to the saturated hydraulic conductivity and fraction of the surface covered by water. With the preliminary data obtained from the field, a regression equation (standard error = 0.0746 cm/h) has been developed:

Infiltration Rate = β0 * Ksatβ1 * Fraction Coverageβ2


where, β0 = 3.19 [cm/h]1-β1

β1= 0.34

β2= 0.67


The goal of this study was to analyze the effect of fractional coverage of water on infiltration rates. All the tests showed that water on the lateral slope of a swale flows as a concentrated flow, not as sheet flow, at the typical intensities for which infiltration practices are designed and utilized to improve surface water quality.

The laboratory and field experiments demonstrate that there is a relationship between the percentage of wetted area and the amount of volume infiltrated. The trend is to have more infiltration when there is more wetted area, and the results show a proportional relationship. In addition, the trend is to have more infiltration when the saturated hydraulic conductivity is higher, as expected.

Additional work is needed to determine whether infiltration below a fractionally wetted surface is predominantly vertical and one-dimensional, or vertical with a significant lateral capillary component (multi-dimensional). Such an effect will impact the infiltration rate, with the one-dimensional flow having a lower rate than the multi-dimensional flow. It is not possible to evaluate this effect with the current data. An analysis will be conducted with a computational flow model to provide further tests on the significance of the lateral flow component in the presence of fractionally wetted surfaces.

The average percentage of water infiltrated for a 2-year event observed in the field was 85%. Further experiments will be conducted using other rain intensities. The results will be incorporated in the regression equation and a Runoff-Infiltration Model for swales will be developed. This model will be a significant contribution to the calculations related to the infiltration capacity of roadside drainage ditches.



  • Ahmed, F., Nestingen, R., Nieber, J.L., Gulliver, J.S., Hozalski, R.M. (2014). A Modified Philip–Dunne Infiltrometer for Measuring the Field-Saturated Hydraulic Conductivity of Surface Soil. Vadose Zone J. Vol. 13 No. 10. doi:10.2136/vzj2014.01.0012
  • Legout, C., Darboux, F., Nédélec, Y., Hauet, a., Esteves, M., Renaux, B., Cordier, S. (2012). High spatial resolution mapping of surface velocities and depths for shallow overland flow. Earth Surface Processes and Landforms, 37(9), 984–993. doi:10.1002/esp.3220
  • Yang, J. & Chu, X. (2013). Effects of DEM Resolution on Surface Depression Properties and Hydrologic Connectivity. Journal of Hydrologic Engineering, Vol.18:Issue.9,1157–1169. September.
  • NOAA Atlas 14 Precipitation- Frequency Atlas Volume 8, 2013.