Stormwater control measures (SCMs) mitigate the runoff from increased urbanization. One common SCM is a rain garden that consists of a depressed vegetated area that captures runoff from surrounding impervious area, holding and treating the non-point source pollutants in stormwater. Efforts to assess ET as part of the water balance cycle have recently increased, as it has been shown to account for volume reduction in small scale SCMs such as vegetated rain gardens.
Evapotranspiration (ET) is the combined effort of evaporation and transpiration to remove water from the soil and vegetation. Evapotranspiration is a known component of the water balance and can account up to 50% of the annual water budget for some Delaware River Basin watersheds (Sloto 2005). This parameter comprises an important component in the hydrologic cycle, but it is often neglected as a volume reduction mechanism in the design constraints in SCMs. Current design and permitting practices for rain gardens only account for stormwater volume reduction by infiltration. Accrediting these systems for ET will help to reduce the size and therefore the cost of rain gardens, as well as to enable implementation where infiltration is not recommended.
At Villanova University, weighing lysimeters have been constructed to represent three different rain garden configurations in a side by side comparison of their ET and infiltration volume reduction potential. Two different soil media were implemented, as well as two different drainage regulation devices. The three lysimeters consist of native plants in a lined soil media basin with an underdrain that is attached to a drainage configuration. The weighing lysimeters are used to replicate and measure hydrologic processes in rain gardens with different designs (Figure 1).
A controlled valve outflow (CVO) allows for a temporary storage of water within a rain garden. This CVO mechanism is used for two differing media: typical engineered media recommended for an infiltrating system and native soil to Villanova, PA. A third lysimeter also contains engineered media with an Internal Water Storage (IWS) zone that acts as a permanent reservoir. ET from these systems are monitored and compared over time.
Flow into the system is measured by a rain gage, soil moisture is measured through the soil column, and flow out of the system via ET and infiltration is measured to provide a comprehensive view of how water is moving through each of the systems. Evapotranspiration loss is determined by a system mass balance via a tension load cell that measures the systems weight at 5 min intervals and precipitation and drainage measurements. Each lysimeter’s drainage regulation system is monitored for exfiltration with a custom-made flow rate device. To simulate the volume of rainfall a rain garden would obtain from the runoff of the surrounding catchment area, a distribution system that provides for a 5:1 and 10:1 impervious area: SCM area rainfall event was developed.
Lysimeters are devices that are built in order to assess the water balance budget of a vegetated or non-vegetated soil column. The use of lysimetry technology dates back to 1640 with the work done by Philippe de Hire for use in agriculture. This technology has many applications, including applying it to a rain garden design as a weighing lysimeter (Figure 1).
Weighing lysimeters work on the principle of mass balance, where the evapotranspiration (ET) is equal to the sum of precipitation, outflow and change in storage, as seen in the equation below:
Where ET is the actual evapotranspiration, P is the precipitation inflow, O is the percolate outflow and ∆S is the change in storage. At the Villanova University site, an S-Beam tension load cell measures the change in weight and is used as a proxy to the change in storage. Soil moisture monitoring will also give a clearer understanding of the change in storage component. The soil monitoring system includes soil moistures meters located at 10 cm, 35 cm, and 65 cm depths in each lysimeter. Precipitation is measured by an American Sigma rain gauge tipping bucket located approximately 5.5 m from the lysimeters. The outflow percolate is measured through a custom made percolate measurement collection system. With the inflow and outflow measured, and the change in weight of the lysimeter recorded, the change in storage can be determined and attributed as a loss to ET.
A weather station is located approximately 4.5 m away equipped with an anemometer for wind speed, a relative humidity and temperature sensor, and a solar radiation sensor (Figure 2).
These climatological parameters are collected in 5 minute intervals and are necessary in estimation equations of ET.
Previously, Villanova University has successfully quantified ET in rain gardens the use of two weighing lysimeters yielding 358 mm and 646 mm of water lost to ET during a period with a total precipitation of 1019 mm (Hickman 2011). One lysimeter was built to mimic the expected ET loss from a free flowing outflow with a similar soil mix to a bioinfiltration rain garden located at Villanova University (Figure 2). This soil mix was developed by mixing half of the “native” Glenelg Loam and half concrete sand. The second lysimeter consists of a lined sandy media mix with an IWS drainage layer of 36 cm (18 in) (Figure 1).
The IWS provides many water quality benefits, such as nitrogen removal, as well as prolonged water storage for plants to draw on during dryer periods and reduce stored stormwater volume via ET (Hunt 2006). Lined bioretention rain gardens are recommended when there are poorly infiltrating underlying soils, allowing for necessary volume reduction by mechanisms such as ET (Prince George’s County 2009). The results of J.M. Hickman’s work on the previously studied lysimeters indicate that both rain gardens were effective at producing considerable amounts of ET that can contribute to overall stormwater volume reduction. Many details about bioretention systems are still to be determined; these include when it is appropriate to use underdrains and depth of the IWS layer (Davis et al. 2009). Since bioretention technology increases the amount of ET activity, it is vital to continue to study this technology under different configurations. These lysimeters have been altered to compare differing outflow and media systems for bioretention design based on a better understanding of how ET is affected by water availability.
Design and Construction
The Villanova’s evapotranspiration (ET) study site consists of three identical lysimeter housings. Each lysimeter is housed inside of a concrete well 1.83 m deep and 1.22 mm square. There is a concrete ceiling with a 1.07 m diameter hole through which a 0.76 m diameter and 0.91 m deep galvanized steel weighing bucket hangs from an open 1.22 m cube galvanized steel structure. This ratio of diameter to depth of lysimeter meets the design recommendation that state that the lysimeter must be at least as deep as its diameter (Marek et al. 1998). The soil column fills approximately 0.76 m of the cylinder depth, leaving 0.15 m of free space at the surface to allow for ponded water within the lysimeters. The ponding depth is in compliance with the recommended range of 0.15 to 0.30 m (PA DEP 2006).
In the present study, the three lysimeters are built to mimic three different rain garden configurations (Figure 1).
The first, labeled 1, is a “native” soil with a controlled valve outflow (CVO) system. The second, labeled 2, is an “engineered” soil with the same CVO. The third, labeled 3, is an “engineered” soil with an internal water storage (IWS) zone (Figure 2).
These three lysimeters designs were constructed to compare how different outflow and media systems affect water availability and ET. A controlled valve outflow (CVO) temporarily stores water within the rain garden’s soil to slow percolation rates and increase the opportunity for ET. The controlled outflow was implemented in two lysimeters with different media: typical engineered media and native soil to Villanova’s campus. A third lysimeter contains engineered media with an internal water storage (IWS) zone, a permanent reservoir created by an upturned drainage pipe.
Lysimeter 1 is comprised of a “native” soil media that had been taken from Ithan Ave construction on Villanova’s campus. The outflow system is comprised of a 5.08 cm gate valve that connects to a 10.16 cm perforated plastic pipe, representing the underdrain. The underdrain system rests on a 1.14 mm (45 mil) EPDM liner atop 10.16 cm of 2 cm coarse gravel, as seen in Figure 3. The liner is in compliance with the recommended thickness of 0.76 mm (30 mil) or greater for bioretention technology (MD DES 2007). Wrapped around the 10.16 cm slotted PVC pipe is more of coarse gravel to prevent cogging. A 3.81 cm cleanout pipe attached to the perforated underdrain pipe in order to release accumulated sediment trapped in the underdrain should it occur (Figure 3).
The cube hanger arrangement that holds the lysimeter is optimized to improve the amount of solar radiation and rainfall reaching the buckets to better simulating the actual conditions without sacrificing structural support. Four chains attach the bucket to the upper corners of the cube and converge at a Sentran S-beam tension load cell (Figure 3).
The handle is extended to the top of the metal bucket. The controlled valve allows for user regulation of percolate flow out of the system. During dryer months the valve may be completely closed to act as a temporary internal water storage zone for plants. However, for the present study the valve will remain at half full flow capacity. A hose attachment is placed on the outflow side of the gate valve to funnel into the percolate collection system (Figure 4).
Lysimeter 2 has essentially the same set up as Lysimeter 1; however the media is comprised of a more typical rain garden soil mix. The “engineered” media placed in Lysimeter 2 is designed to match existing media in Lysimeter 3. Lysimeter 3 was previously built in 2010. Modifications were introduced to assimilate this lysimeter with the newly implemented ones. For this design, instead of a valve, a 5.08 cm upturned elbow is connected to the underdrain device (Figure 4). The upturned elbow creates approximately 35.56 cm layer of saturated soil to house denitrifying bacteria, and act as a reservoir. The top of the upturned elbow is attached to a hose. This hose is used to funnel the percolate into the outflow system bucket, similar to that of lysimeter 1 and 2. Lysimeter 1 is comprised of a loamy sand with a controlled valve outflow (CVO). Lysimeter 2 is the same CVO device with a sand media. The CVOs remain at half open through this study. Lysimeter 3 is comprised of a sand media with an IWS. Lysimeter 3 was previously studied and established 3 years prior to lysimeter 1 and 2.
Below each lysimeter there is an outflow measurement bucket to capture the percolate from the rain garden microcosm. This device consists of a constant diameter bucket, with the following attachments: a ToughSonic Senix ultrasonic measurement device, the percolate hose, and a solenoid outflow valve (Figure 5).
In the field, the outflow hose is placed opposite to the distance measure sensor to avoid detecting high ripple effects from the hose (Figure 5). The ultrasonic outflow measurement device monitors water height level within the bucket. The distance sensor is optimized to operate within the depth of the bucket and internally compensated for temperature. This particular sensor functions on the speed of light, which varies in relation to temperature. The water level reading is converted to the percolation rate of the soil column through a relationship between the diameter of the bucket and the diameter of the soil column. The ultrasonic setup allows for recording large to very small flows that is necessary to capture the peak of the storm simulations to the dripping of percolate hours after the rain event.
The plant selection was based on the plants that were chosen for the Villanova University bioinfiltration rain garden as well as recommendations from the horticulturalist from Villanova’s Facilities and Maintenance Department. Switch grass (Anicum Virgatum), Seaside Goldenrod (Solidago Sempervirens), and Black Chokeberry (Photinia Melanocarpa) were planted in the summer of 2013 with a layer of topsoil for nutrients (Figure 6).
These plants are indigenous to the New Jersey coast and are resistant to saline environments. Salt resistivity is necessary for rock salt accumulation in rain gardens from de-icing of pavements during the winter months.
Soil Moisture Monitoring
Soil moisture monitoring was implemented throughout the profile of the three weighing lysimeters as well as in the adjacent in situ soil. The in situ profile is defined as the dense native soil that lies next to the lysimeter wells under a grass cover. Steven’s Hydraprobe II soil moisture sensors are placed at 10, 35, and 65 cm depth, with a duplicate sensor at 35 m depth. The placement of these sensors was selected to be at the top, middle and bottom to get a full profile of the water path. The 10 cm placement is conjectured to be in the dense root zone, the 35 cm is approximately placed at the interface of the top of the water reservoir in lysimeter 3 and the 65 cm was the deepest that the moisture meters could feasibly be placed without conflicting with the drainage pipe. A duplicate sensor was placed at the 35 cm mark for quality assurance purposes. In recent years, soil moisture monitoring technology has significantly improved, but still only provides an estimate of soil moisture. Therefore, soil specific custom calibration were developed to obtain more meaningful results.
Soil moisture meters were calibrated based on soil specific properties. Duplicate meters were placed at 35 cm depth and recorded soil moisture at 5 minute intervals; these duplicates were used to verify the output data. The mean moisture contents with the maximum standard deviation are 0.324±0.026, 0.247±0.031, and 0.210±0.030 vol/vol for lysimeters 1, 2, and 3, respectively, which corresponds to 2.6%, 3.1%, and 3.0% variability in the readings of raw soil moisture data. The manufactured specifications state the accuracy of the soil moisture meters is ±0.01 vol/vol for most soils and ±0.03 vol/vol maximum for fine textured soil (Stevens 2007). The maximum deviation experimentally determined on the raw soil moisture meter readings conforms quite well to the out of factory variability. This variability could also be attributed to differential settlement between the meters after installation.
The plant selection was based on the plants that were chosen for the Villanova University bioinfiltration rain garden as well as recommendations from the horticulturalist from Villanova’s Facilities and Maintenance Department. Switch grass, Seaside Goldenrod, and Black Chokeberry were planted in the summer of 2013 with a layer of topsoil for nutrients. These plants are indigenous to the New Jersey coast and are resistant to saline environments. Salt resistivity is necessary for rock salt accumulation in rain gardens from de-icing of pavements during the winter months.
Storm Simulation System
Rain gardens typically collect runoff from surrounding impervious area. The PA BMP Manual (2006) indicates suitable loading ratios for rain gardens range from 5:1. A loading ratio is defined as impervious runoff area to SCM area. Since there only exists a 1:1 loading on each lysimeter, a water distribution system was developed to mimic the runoff quantity from surrounding impervious surfaces. The distribution system was built and calibrated to handle a range of storm volumes and intensities. Storm events of 19, 38, and 76 mm (0.75, 1.5, and 3 in) over a 24 hour period were chosen, as they mimic design criteria for PA (PA DEP 2006). The 19 and 38 mm (0.75 and 1.5 in) storms relate to typical design critera for SCMs and the 76 mm (3 in) storm relates to the 2 year storm for PA, which ranges from 61 to 84 mm (2.4 to 3.3 in) across the state (PA DEP 2006). The smallest storm intensity, that of a 19 mm (0.75 in) storm at an 5:1 ratio yields a loading rate of 30 ml/min (0.008 gpm). The largest storm intensity is 120 ml/min (0.032 gpm) for the 76 mm (3 in) storm at a 5:1 loading ratio. The distribution system was design and calibrated to produce flow rates between 30 and 241 ml/min in case higher loading ratios (10:1) wanted to be explored.
The location of the site adjacent to the Constructed Stormwater Wetland (CSW) provides for a supply of water for the distribution system. A 12.7 l/min (3.35 gpm) delivery pump is utilzed to overcome the height differentinal between the lysimters and the base of the sediment forebay of the CSW. The 1.91 cm (0.75 in) suction tube is elevated about 45 cm from the base covered with a mesh filter to avoid suspeneded soilds. Another easily accessable 150 micron (No. 100) mesh filter is in place before the delivery pump to ensure longveity of the pump as recommended by the manufacturer (Aquatec 2009). This set up provides the 380 liter (100 gal) storage tank with a reservoir of water (Figure 1).
The storage tank provides water to a distribution head regulated by a peristaltic pump. The peristaltic pumps are able to produce variable flows of 3.8 ml/min to 380 ml/min (0.001 to 0.1 gpm), which incorporates the goal simulation range of 30 to 241 ml/min (0.008 to 0.064 gpm). The distribution head is made from 9.5 mm (0.37 in) flexible plastic tubing formed into a circle and preforated with approximatly twenty 1 mm (0.04 in) holes. The tubing is supported by 12.7 mm (0.5 in) PVC pipe (Figure 2).
Each of the three rain garden lysimeters has their own distribution system. Each distribution head and pump were calibrated together to account for any minor losses specific to the system. For very low flows, a slight incline to the distribution head could produce inaccurate flow amounts, so it is important that the distribution heads remain level over the duration of the test. To properly mimic weather conditions, as they affect ET, storm simulations will be performed during actual rainfall events. The inflow of the natural event will be added to the total volume of the storm simulation via the rain gage tipping bucket.