Case Study #3: Assessment of infiltration at a rain garden

B.C. Asleson, R.S. Nestingen , J.S. Gulliver, R.M. Hozalski and J.L. Nieber at the University of Minnesota.

The University of Minnesota, St. Paul campus rain garden is located on Gortner Avenue and Commonwealth in the Mississippi River watershed. There are five rain gardens located along Gortner Avenue, and three of them are in series. Basins C and B serve as overflow basins and are connected to basin D by two drop structures consisting of bricks. The assessment was conducted on the basin D rain garden. The rain gardens were designed by Barr Engineering and installed in October of 2004. A thorough assessment of basin D was conducted in the summer of 2006.

Basin D rain garden is approximately 716 square feet in size with a ponding depth of 0.5 feet (the design plans indicate 960 square feet with a ponding depth of 2 feet). It is designed to provide storage for the maximum amount of water the space would allow. Stormwater runoff is directed to the rain garden using two inlets, a curb cut-off of Gortner Avenue located along the northwest corner of the rain garden, and an inlet pipe located at the center of the north border of the rain garden, which is connected to the stormwater sewer system. The storm sewer inlet pipe has a 5 inch by 12 inch subgrade of Fond Du Lac wall stone to prevent erosion. The native soil was excavated and filled with a sand trench to a depth of 3-4 feet and a width of 3 feet in the center of the basin. Clean sand with only 5% passing through a 200 micron sieve was used for the sand trench. Basin D rain garden was designed to infiltrate the maximum storage volume within 24 hours at an estimated infiltration rate of 0.5 inches/hour. The basin was then filled with planting topsoil to a depth of 8 inches and planted with selected vegetation. The plant design plan is shown in figure 1.

Assessment Goals

The purpose of the assessment was to determine if the rain garden had the ability to infiltrate stormwater runoff at the appropriate rate. Rain gardens are typically designed to drain within 48 hours after a storm event. Three of the four levels of assessment as described in Developing an Assessment Program were conducted: visual inspection, permeability tests, and a synthetic runoff test.

Visual Inspection

Assessment Techniques

The visual inspection of rain gardens consists of two components: a vegetation analysis and an inspection of the soil. The vegetation analysis examines the species of vegetation present in comparison with the design plans, apparent health of the plants, percent cover of vegetation, and presence of invasive weeds and/or wetland plants. The original plant design was used along with a plant field guide to identify the species present. The leaf color, height, and width of the plants were examined and described as poor, fair, or good. The site was examined for bare spots, and a percent of the vegetation cover was determined. Several photographs of the plants were taken to serve as a record of the vegetation.

The inspection of the soil was conducted by evaluating several soil properties, soil texture and color, soil moisture, and bulk density. These procedures can be found in the Soil Science Society of America Book Series: 5, Methods of Soil Analysis, Part 1-Physical and Mineralogical Methods (Klute 1986). Soil texture can be determined from a sample using sedimentation procedures or in the field using a field guide. The textural flow chart was used to determine the texture of the soil. Soil color was determined using a Munsell® soil color chart (Munsell 2007) and was done for each new layer (signified by a change in texture or color).

Figure 2 is a photograph of the texture and color of the soil being determined in the field using the USDA textural triangle, textural flow chart, and the Munsell® soil color chart. There are several methods available for measuring soil moisture, or a general wetness can be described. When making general wetness statements, the terms dry, moist, saturated, and inundated are typical descriptions. The soil moisture was measured at this site using two methods: the gravimetric method and with the use of a capacitance probe. Bulk density using the core method was the final soil property measured as part of the assessment. Bulk density can be used to convert gravimetric water content to volumetric water content, calculate porosity, and void ratio when particle density is known and is a useful index of the degree of compaction. Additional observations were made regarding channelization, sediment accumulation, erosion, and condition of inlet structures.

Assessment Results

The plants in the rain garden matched fairly well to the design plans, although there appeared to have been modifications made to the original design plan. Most of the plants appeared to be healthy with the exception of geraniums along the west edge. They were not filling in the area as they should and their growth was not as lush as expected. There was a fairly large bare area northeast of the center of the basin where the anemones and chelone come together.

The overall percent plant cover of the basin was approximately 70%. There was a large bare area near the inlet and up the side slope next to the curb inlet. Vegetation was sparse in the center of the basin with several large bare areas between plants. There did not appear to be a large number of weeds present, and there was no sign of wetland vegetation. Based on the visual inspection of the vegetation, there appears to be some limitation in plant growth.

The inspection of the soil included soil texture and color, soil moisture, and bulk density. The soil texture and color was as follows:

  • 0 – 8 inches: Sandy Loam, 10YR 2/2
  • 8 – 19 inches: Silt Loam, 10YR 2/1
  • 19 – 47 inches: Sand (non-native), 10YR 6/4
  • 47 - ? inches: Silt Loam with coarse Sand, 2.5YR 3/3

The sandy loam topsoil is typical for rain gardens in Minnesota; however, the silt loam layer below is of concern. When comparing the two soils on the USDA Soil Textural Triangle, there is a much higher percentage of silt, which is finer than sand and has a lower Ksat than the sandier layer above and below it. The silt loam layer therefore restricts water from flowing downward until the entire soil profile is saturated. The original design plans indicate 3 to 4 feet of the non-native sand directly below the 8 inch layer of topsoil. The soil moisture of the basin was near saturation most of the time. This indicates that there is sufficient water for plant growth with adequate drainage. The mulch layer and canopy cover over the soil surface are likely contributing to the retained moisture during the dry season. The bulk density of the site varied spatially, with an average of 1.18 grams per cubic centimeter. These observations are lower compared to the typical 1.3 grams per cubic centimeter for most mineral soils; however, with the high organic matter content due to the mulch and plant roots, lower-than-expected bulk densities appear to be normal for rain gardens. There were no signs of hydric conditions such as gleying or the presence of mottles. Based on inspection of the soil properties, the infiltration appears to be adequate; however, the restrictive layer of silt loam may pose problems for long-term operation by retaining too much water during large storm events. No signs of erosion or channelization were present near the inlet structure, and both inlet structures were in good condition.

Capacity Testing

Assessment Techniques

The permeability of the soil was measured to determine the rain garden’s capacity for infiltrating water. At this site several devices were used to measure the saturated hydraulic conductivity (Ksat) of the soil in order to establish the technique. The three devices used to measure Ksat were the Double Ring Infiltrometer, Minidisk Infiltrometer, and the Modified Philip-Dunne Infiltrometer. Locations where point measurements of Ksat were to be made were distributed evenly throughout the entire rain garden and marked using orange utility flags. These locations varied in their proximity to the vegetation but were never placed directly over the base of the plant. Additional locations were marked at the low point of the site to better represent the frequently occurring small runoff events. Figure 3 is a photograph of the rain garden with orange utility flags marking test locations.

A total of 40 locations were marked in this site to evaluate the spatial variability of Ksat within the basin. The coordinates of each location as well as the perimeter of the rain garden was logged using a GPS device. At 38 of the test locations a measurement was made using the Modified Philip-Dunne Infiltrometer, and another measurement was made using the Minidisk Infiltrometer. At the two remaining locations, two measurements were made using the Modified Philip-Dunne Infiltrometer and two measurements were made using the Minidisk Infiltrometer. The Double Ring Infiltrometer was only used to make measurements at two of the locations due to its bulkiness and lengthy time and water requirements. Each location was allowed to dry out between measurements.

The Double-Ring Infiltrometer is a constant head infiltrometer and requires two sources of water, one for the inner ring and one for the outer ring. The inner ring had a diameter of 8 inches and the outer ring diameter was 16 inches. Constant head was maintained in the inner ring of the double-ring using a Mariotte system. The system used in the field is shown in figure 4. Water levels inside the plastic container (figure 4) and time measurements were recorded once steady state was achieved. For detailed instructions on Double-Ring Infiltrometer procedures see Soil Science Society of America Book Series: 5, Methods of Soil Analysis, Part 1-Physical and Mineralogical Methods (Klute 1986).

The Minidisk Infiltrometer was purchased through Decagon Devices and is a compact disc infiltrometer. This is a transient flow device in which water is delivered to the soil surface through a porous disc at a negative pressure. This technique is used to prevent water from flowing through large macropores and results in a Ksat value representative of the soil matrix itself. This particular device required change in water volume with time measurements to be recorded. These data were then input into a Microsoft Excel ® spreadsheet provided by the manufacturer.

The Modified Philip-Dunne Infiltrometer is a falling head infiltrometer constructed specifically for this project. The device was uniformly pounded into the soil to a depth of 2 inches. The initial soil moisture was measured at five locations around the base of the Modified Philip-Dunne Infiltrometer by two methods: the gravimetric method and using a capacitance probe. Mulch from the rain garden was placed inside the device to prevent erosion; water was then poured into the device to the desired height, which was 17 inches for this site. Two sets of change in water level with time measurements were made for additional data. The first set was the visual method, which requires an initial height of water at time zero, a time at the half way point (approximately 8 inches), and a time at empty. The second method made continuous measurements using an ultrasonic sensor. The soil moisture was then measured from directly inside the device, again at five locations.

The original Philip-Dunne equations (Philip 1993), equation 1 below, were modified and the data collected was then used to calculate Ksat. A Microsoft Excel ® spreadsheet was developed to input the measured parameters and calculate Ksat. An example of the spreadsheet can be seen in table 1; the highlighted cells indicate the necessary input parameters. Figure 5 is a photograph of the Modified Philip-Dunne Infiltrometer being used in the field with an ultra sonic sensor for continuous measurements.

Assessment Results

The 40 locations used for point measurements were positioned using GPS and input into ArcView. Figure 6 is an ArcMap of the measurements made using the Modified Philip-Dunne Infiltrometer. The results of the measurements made with the other devices as well as the Modified Philip-Dunne Infiltrometer are shown in figure 7. Both figures illustrate how Ksatvaries both spatially and among the devices. The average Ksat for the double-ring infiltrometer, minidisk infiltrometer, and the Modified Philip-Dunne Infiltrometer were: 0.999 inches/hour, 0.668 inches/hour, and 1.087 inches/hour, respectively. All of the devices used to measure Ksatare based on different theories of flow through the soil and different assumptions regarding the system. Currently, none of the devices mentioned have the ability to account for the presence of macropores or other preferential flow paths found in the soil. Detailed discussion on devices used to measure infiltration can be found in Infiltration.

As a result of an evaluation of the devices based on this and previous field work, the Modified Philip-Dunne Infiltrometer was found to be most desirable and is recommended for future assessment of infiltration/filtration practices. For more time-efficient assessment it is recommended to use multiple Modified Philip-Dunne devices. This level of assessment (i.e. level 2) was determined to be the most beneficial technique for understanding the spatial variability of the site and developing a maintenance schedule for the practice.

The time required to drain the design storage volume can be estimated using the measured saturated hydraulic conductivity value of 0.668 inches/hour as a conservative estimate of infiltration rate. With this infiltration rate and the known design depth of 6 inches, the drain time can be estimated by dividing 0.668 into 6 to get 9.0 hours.

Synthetic Runoff Test

Assessment Techniques

A synthetic runoff test was conducted at the site to measure the time required to drain the maximum storage volume. To determine if the nearby fire hydrant could provide the necessary flow, the analysis procedure detailed in example 10.1 was performed. In summary, the water quality volume of the rain garden was estimated by multiplying the surface area of 716 ft2 by the design depth of 0.5 feet to get 358 ft3. Assuming the infiltration rate measured with the Modified Philip-Dunne method of 1.087 inches/hour exists when filling the rain garden and that it is desired to fill the rain garden in 30 minutes (i.e. 1800 seconds), the required flow from the hydrant was calculated to be 0.22 ft3/s by solving the following equation:

(Qreq is the discharge the hydrant must supply if the rain garden is to be filled in 30 minutes.)

Since fire hydrants can typically provide flow up to 3 cubic feet per second (1,500 gal/min), it was determined that the required flow could be obtained from the nearby fire hydrant.

An ultrasonic sensor was positioned directly above the low point of the basin to make continuous water level measurements over time prior to flooding the site. A bare spot within the basin was chosen to provide a good reflective surface for the sound waves. The hydrant was then prepared by connecting a 2.5-inch fire hose to the hydrant using a safety valve to ensure no back flow. The fire hose discharged water into the storm sewer manhole closest to the basin until the rain garden was filled to capacity. Permission was granted by the University of Minnesota facilities management, who also assisted by providing all of the proper connectors, valves, and hoses for the fire hydrant. After the water stopped flowing into the basin, the water level was measured and the timer started. Continuous measurements using the ultrasonic sensor as well as visual measurements with a yard stick were made until the basin was completely drained.

Assessment Results

Figure 8 is a graph displaying the change in water level with time using the data collected from the ultrasonic sensor. The synthetic runoff test represents the drain time of two hours when the rain garden is filled to capacity. This is about four and one-half times shorter than the conservative estimate of 9.0 hours, which was obtained by assuming the saturated hydraulic conductivity value was equal to the infiltration rate.

Conclusions and Recommendations

The overall performance of the basin D rain garden is satisfactory. The results of the visual inspection indicate that there are some concerns for a few of the plant species. These particular species should be further evaluated to eliminate possible causes of growth limitation. Some examples of possible growth inhibitors include improper soil moisture regime, improper sun/shade location, limited oxygen in the soil, high levels of salt in the soil, the presence of invasive species, and a lack of several other plant-specific requirements. The center of the basin had near-saturated soil conditions during the dry season and therefore the plants located in this region should represent a wet meadow plant community. Despite the issues near the inlet and in the center of the basin, the majority of the vegetation was in good health and provided sufficient cover. The soil inspection indicates the potential for problems for large runoff events. The silt loam soil layer just below the topsoil and above the sand trench will result in restricted infiltration to the sand trench due to the smaller pore sizes and increased holding capacity of that particular layer. A more thorough inspection of the soil layers throughout the basin should be conducted to determine the extent of the restrictive layer. The distribution of infiltration rates also indicates that there is a problem with the soil in the center of the basin. All of the very low infiltration rates were located in the center of the basin, which is also where the soil core was taken to characterize the soil layers of the basin. This indicates that the restrictive soil layer could be causing these lower infiltration rates in the center of the basin. The side slopes of the basin had high infiltration rates. Table 2 summarizes the results from the capacity testing and synthetic runoff testing to determine whether stormwater will drain within the specified 48 hours.

To determine the time it will take for the basin to drain using the capacity testing results, first use the dimensions of the basin to calculate the surface area and the storage volume. The infiltration discharge can then calculated by multiplying the surface area by the measured Ksat.The time to drain the storage volume can be estimated by dividing the storage volume by the infiltration discharge. An alternative approach that was previously used in section 3 was to divide the design depth by the infiltration rate. As mentioned in section 2.3, each method for measuring Ksat of the soil can result in different values due to the theory of flow they are designed for and the scale of the measurements. The results from the measurements made with the Minidisk Infiltrometer would represent the minimum value for the soil, and the synthetic runoff test represents the Ksat when the basin is filled to its holding capacity. The Double-Ring Infiltrometer and the Modified Philip-Dunne Infiltrometer capture a percentage of the macropores present in the soil but cannot account for the total spatial variability of the rain garden. To estimate a conservative drain time, the results from the Minidisk Infiltrometer should be used for this site as it represents the permeability of the soil matrix itself.

The synthetic runoff test indicated very good infiltration when the basin is filled to the maximum storage volume, and the average results of the capacity tests are better estimates for typical rainfall events. Additional synthetic runoff test at varying depths could be conducted to understand how the basin drains for smaller rain events. Although there appear to be some concerns regarding the basin, drainage time is well below the designed 48 hours. The low infiltration and sparse cover of vegetation occurring in the center of the basin should be further evaluated and amended to prevent failure of the basin in the future. A maintenance schedule should be developed based on this evaluation to ensure adequate stormwater treatment efficiency.


Klute, A. 1986. Methods of Soil Analysis, Part I. Physical and Mineralogical Methods, 2nd edition. Soil Science Society of America, Inc. Publisher, Madison.

Munoz-Carpena, R., C. M. Regalado, J. Alvarez-Benedi, and F. Bartoli. 2002. Field evaluation of the new Philip-Dunne permeameter for measuring saturated hydraulic conductivity. Soil Science 167:9-24.

Munsell. 2007. Munsell Soil Color Chart.

Philip, J. R. 1993. Approximate Analysis of Falling-Head Lined Borehole Permeameter. Water Resources Research 29:3763-3768.

Return to Case Studies.