Abstract
Little is known about how stormwater exfiltrates from green infrastructure and few efforts have been undertaken to address this question. This study used tensiometers to monitor water exfiltration from an aggregate-filled storage gallery installed under permeable pavement. An 80-space parking lot was built at Seitz elementary school in Fort Riley, KS under an agreement between EPA and US Army during the summer of 2015. A network of twelve tensiometers and twelve monitoring wells was installed under and south of the storage gallery. Tensiometers were installed a various depths and distances to monitor soil moisture tension. The installation was used to monitor subsurface water flow patterns from the storage gallery under the permeable pavement site. The results of the study showed that soil moisture tension is larger at the shallower depths, decreasing with depth from the ground surface. Larger soil tension was associated with increased distance from the permeable pavement storage gallery. The results showed exfiltration from both the sidewalls and the bottom of the gallery while the changes in soil tension were larger for the tensiometer monitoring exfiltration from the side walls. Both the accumulated water depth inside the storage gallery and groundwater level rise were positively correlated with total rainfall depth. The calculated vertical flow rate was larger than the horizontal flow. The soil water tension change associated with storage gallery exfiltration was measured in a radius of <5 m from the storage gallery. Long-term peak groundwater level rise should be considered for design of the storage gallery depth to maintain the effective exfiltration. Understanding the exfiltration pathways aids with the placement and design the storage gallery. Additional research is necessary to understand how specific local parameters and vadose zone characteristics would affect the long-term exfiltration process.
Keywords: Tensiometer, Exfiltration, Sidewalls, Permeable pavement, Stormwater control measure
Graphical Abstract
1. Introduction
Due to climate variability and increased use of impermeable surfaces in urban areas, the rate of flooding events has increased in recent years (Hamlet and Lettenmaier, 2007). Incorporating green infrastructure (GI) into built environments is one strategy to manage stormwater associated with urbanization. GI stormwater control measures (SCMs), including permeable pavement systems with infiltration, have been implemented in urban areas of the United States (US). By reducing runoff volume, these technologies can reduce the pressure on the urban drainage systems and the pollutant load to the receiving water bodies. However, the mechanisms of how stormwater infiltrates into the SCM and how water exfiltrates from the SCM are incompletely understood. Understanding these processes will help planners design and size the SCMs.
The available stormwater models conceptualize multiple approaches to describe the exfiltration mechanism. Several models, such as SLAMM (Source Loading and Management Model) and EPA SWMM (Stormwater Management Model), assume stormwater exfiltration from only the bottom of the SCM (Pitt and Voorhees, 1995; Rossman, 2015). Other models assume exfiltration from only the sidewalls (Staher and Urbonas, 1990; Bettess, 1996; Lee et al., 2015) and some incorporate exfiltration from both sidewalls and the bottom, such as Infoworks, HydroCAD, and the newer version of MUSIC (Model for Urban Stormwater Improvement Conceptualisation) (InfoWorks Software, 2002; HydroCAD Software, 2011; eWater, 2013). After water ponds inside a storage gallery, it is expected that water would exfiltrate from all wetted surfaces, including the sides and bottom (Lee et al., 2015).
Several studies have adopted Darcy’s law for calculating water exfiltration (Schlüter et al., 2007; Emerson et al., 2010; HydroCAD Software, 2011; Lee et al., 2015). Green-Ampt relation was used by Braga et al. (2007) who developed an exfiltration model of the sub surface storage unit from an infiltration BMP. Brown and Borst (2014) used water level data measured by pressure transducer and piezometers for calculating exfiltration rate from bottom of a permeable pavement infiltration trench. There is a lack of direct exfiltration study in the vadose zone.
Because the exfiltrating water from the SCM flows into unsaturated zones, tensiometers, which can be used to study the water and moisture patterns in the vadose zone, were selected for use in this study. Tensiometers measure soil particle moisture tension or soil water negative pressure (Kirkham, 2005). The soil water tension energy is the portion of the water potential that can be attributed to the attraction of the soil matrix for water (Kirkham, 2005). Tensiometers have been used to solve soil water movement problems, conduct groundwater recharge studies, find groundwater tables, determine irrigation scheduling, study nutrient and solution transport, investigate leakage from landfills, and establish stable soil slopes (Nielsen and Johnson, 1990; Kirkham, 2005; Todd and Mays, 2005; Das, 2008; Toll et al., 2011).
A tensiometer is comprised of a porous cup that allows water, but not air, to enter (Nielsen and Johnson, 1990). Water enters the evacuated ceramic cup of the tensiometer until there is a hydraulic equilibrium between the soil water and interior of the tensiometer. The equilibrium pressure, known as the matric pressure head, has been recorded using different devices, including manometers, burden gauges, and pressure transducers. When this technology was first developed, mercury was used in the manometer, but long response time and variability under warm and cold climate are disadvantages associated with mercury-based tensiometers (Kirkham, 2005). In recent generations of tensiometers, a pressure transducer has been used for pressure measurements, and they have been designed to work across a wider operational range with a quicker response time than the older generation (Sisson et al., 2002). Early tensiometers were able to record the soil moisture tension in the operational range of 0 to ±700 cm of water. Current tensiometers record a range of 0 to 867 cm of water (Sisson et al., 2002; Kirkham, 2005; UMS, 2009).
The revised form of Darcy’s law, which results from combining the original Darcy’s law with the equation of continuity in Richard’s equation, is the governing equation in solving water movement in unsaturated soil.
According to Bernoulli’s theorem, for any point in the soil, the total hydraulic head is comprised of the elevation head, pressure head, and velocity head (Kirkham, 2005; Das, 2008). The velocity head is very small in the soil media and can be neglected. The hydraulic head would then be limited to the elevation head and matric potential or pressure head. It is possible to determine the water movement between a pair of tensiometer. To estimate the exfiltrated water direction from the storage gallery or flow rate among multiple tensiometers, a network of tensiometers is necessary to record soil water moisture tension at multiple locations and elevation in an unsaturated zone. In this study, to monitor the soil water tensions in the vadose zone, a network of 12 tensiometers was installed near and under the pavement section storage gallery at three distances and depths. The designed system monitored the water exfiltration pathways from a subsurface storage gallery into the surrounding, unsaturated media. In addition, the measurements were used to study the effect of the depth and distance from the storage galley on soil moisture tension and to calculate the flow rate of exfiltrated water.
2. Methods
2.1. Study site
The study was conducted at the parking lot of the Seitz Elementary School (39°04′12.1″N latitude and 96°50′31.9″W longitude) on Fort Riley, Kansas (Fig. 1-A). This region of Kansas has a temperate continental climate characterized by hot summers, cold, dry winters, moderate winds, low humidity, and a pronounced peak in rainfall late in the spring and in the first half of summer. The average yearly rainfall and snowfall is 927 mm and 330 mm, respectively. Prevailing winds are from the south to southwest during most of the year. During February and March, the prevailing winds are from the north (US Army, 2014).
Fig. 1.
(A) Shows the location of the study area; (B) an 80-space parking-lot with PICP section.
US Environmental Protection Agency has been conducting green infrastructure implementation, research and outreach across the United States under various climate condition. This spans from the research conducted in the east coast of the United Stated to cover the high rainfall areas toward the west coast that contains areas with less rainfall. Fort Riley, Kansas that geographically is located in the middle of the United States is representing of an area with dry winter and wet late-spring through summers. The similar water exfiltration pattern from the storage gallery of the permeable pavement in similar climate and soil condition can be expected. Teaming with the US Army under the Net Zero program, Fort Riley provided a secure location for the research and the land area needed for the monitoring network.
An 80-space asphalt parking-lot with a permeable interlocking concrete pavement (PICP) section captures the parking lot runoff and allows it to drain into the underground layers (Fig. 1-B). The design of the storage gallery volume was based on capturing the runoff from the 5-year, 24-h rainfall event (107.8 mm). The contributing drainage area was 1833.7 m2, while the PICP surface area was 175.7 m2. The ratio of the contributing drainage area to PICP area was 10.4, which is about double the recommended ratio (Smith, 2006). Increasing the ratio was expected to increase the frequency of deep water accumulation in the storage gallery, thereby facilitating the research monitoring, albeit at the potential expense of more frequent surface maintenance.
During the monitoring period, the parking lot was mainly used for student pick up and drop off, with heavy traffic during school arrival and dismissal but relatively low use at other times. Because of the low infiltration rate of the surface soil at the study site, the permeable pavement section of the parking lot was designed to have a deep storage gallery. The storage galley is 3.2 m (width) × 54.9 m (length) × 3.3 m (depth). The three layers under the pavement section include a 5-cm thick bedding layer of AASHTO (American Association of State Highway and Transportation Officials) No. 8, a 15.2-cm thick choker layer of AASHTO No. 57, and the storage gallery (3.10 m of AASHTO No. 2) (Fig. 2-B). Fig. 2 also depicts the placement of monitoring wells and tensiometers and provides elevations relative to sea level. Before placing the compacted aggregate in the storage gallery, the walls and bottom were covered with permeable, non-woven, double-needle punched geotextile.
Fig. 2.
(A) Plan view and (B) cross-sectional view of the location of tensiometers (description in Table 1).
Infiltration tests during school construction showed exceedingly small infiltration rates (I < 0.15 cm/h) of the surface soils and it was confirmed by the geotechnical investigations that showed a surface clay layer (Kaw Valley Engineering Inc., 2011). Borings were drilled to 4.6-m (15 ft) below grade using a CME-55 drill rig, using 10 cm O.D. continuous flight augers. Sample was obtained utilizing a thin walled steel tube and a split-barrel sampler. All 21 borings were drilled to depth of 4.60 m below the existing ground surface. A fill layer generally <0.5-m was found across the site. The vadose zone texture was comprised of combination of silt, lean (low plasticity) clay, and sand through the full depth of the borings and this formation is the type that prone to frequent capillary rise (Kirkham, 2005; Das, 2008). The moisture contents ranging from 2.3% to 38.3% which the low moisture contents correspond to sandier soil. In-situ bulk densities ranged from 1.26 to 1.60 g/cm3.
A weather station installed on the southwest corner of the school roof monitored rainfall and other meteorological parameters. The data logger was programmed to record all readings at 1-min intervals during rain events and 10-min intervals otherwise. Rain events were defined as a period with at least 2.4 mm of rainfall measured by the on-site tipping bucket rain gauge and dry inter-event period of at least 6 h, as generally suggested by Driscoll et al. (1989). The time between the finish of the last rainfall and start of the new event considered as antecedent dry period (ADP) (Razzaghmanesh and Borst, 2018).
2.2. Tensiometers installation
Twelve UMS model TS1 (Decagon, Inc., Pullman, WA) were installed on the southern side of the storage galley. It was expected that the same pattern applies to the other (northern) side of the gallery that is under the parking lot. The frost depth in Kansas is around 1.1 m (USDOT, 2008). All tensiometers in this study were installed below the frost line.
TS1 are battery operated and self-filling, thus allowing unattended long-term monitoring, and compatible with the selected data logger (Campbell Scientific CR1000, Logan, UT). According to the manufacturer, TS1 combines the principles of a pressure transducer tensiometer to measure the soil water tension (UMS, 2009). The measurement range of TS1 is from 0 kPa to 85 kPa for soil water tension and from −100 kPa to 0 kPa for water pressure when submerged. The three main parts of a TS1 are a ceramic cup, a miniature pump, and a pressure transducer. To refill the cup, the pump creates a pressure lower than the water tension in the surrounding soil so that water flows from the surrounding soil into the ceramic cup. Excess water flows back to the surrounding soil through a release. As recommended by the manufacturer, the tensiometer was installed in boreholes at a 30° angle, allowing for the complete removal of air bubbles (UMS, 2009). Each borehole was advanced using 10.2-cm diameter continuous-flight augers by a CME-45C drill rig equipped for directional drilling (Fig. 3). After reaching the planned position, the augers were removed, and the cuttings from the borehole segregated, slurried, and used to backfill the borehole after inserting the tensiometer. The borings for all tensiometers were set perpendicular to the southern edge of the parking lot. The starting point was surveyed based on geometry using the drilling angle and the specified final depth assuming a perfectly straight boring pathway.
Fig. 3.
Drilling and installing tensiometers around storage gallery. Flags show the surveyed entry points.
Table 1 lists the names and locations of the installed tensiometers. The naming convention for the tensiometers was T(X, Y) format, where X is the distance (m) from the gallery centerline and Y is the depth below ground level (m). The naming convention for the four tensiometers below the center line of the gallery includes designation to indicate whether the tensiometer was installed under the east (E) or west (W) side of the storage gallery; all other tensiometers are along a line perpendicular to the midpoint of the storage gallery.
Table 1.
Tensiometer locations.
| Tensiometers name | Distance from the gallery centerline | Depth below ground level | Location with relative to the centerline of the pavement section |
|---|---|---|---|
| T(0,5)W | 0.0 | 5.0 | 13.70 m West |
| T(0,5)E | 0.0 | 5.0 | 13.70 m East |
| T(0,6.5)W | 0.0 | 6.5 | 13.70 m West |
| T(0,6.5)E | 0.0 | 6.5 | 13.70 m East |
| T(2,3) | 2.0 | 3.0 | Centerline South |
| T(2,4) | 2.0 | 4.0 | Centerline South |
| T(2,5) | 2.0 | 5.0 | Centerline South |
| T(4.5,3) | 4.5 | 3.0 | Centerline South |
| T(4.5,4) | 4.5 | 4.0 | Centerline South |
| T(4.5,5) | 4.5 | 5.0 | Centerline South |
| T(12.5,5) | 12.5 | 5.5 | Centerline South |
| T(12.5,6.5) | 12.5 | 6.5 | Centerline South |
To measure water ponding depth, a pair of shallow monitoring wells were installed in the storage gallery under the pavement section. These were installed at one- and two-thirds the length of the storage gallery. Monitoring wells were also drilled around the storage gallery. The monitoring wells were drilled into groundwater at set distances from the midpoint of the gallery. The locations of the four monitoring wells (designated MW1, MW2, MW3, and MW4) are shown in Fig. 2. Pressure transducers (CS451(Campbell Scientific CR1000, Logan, UT) hardwired to the data logger recorded the water level in monitoring wells to the data logger. The CS451 is a temperature-compensated, vented sensor that can measure water pressures of 0–50 kPa.
The maximum recorded soil moisture tension after the preceding rain event and before the current rainfall event was considered as the maximum soil moisture tension. The minimum soil moisture tension recorded after the current rain event and before the next rain event was defined as the minimum recorded tension. The tension change has calculated as the difference between the maximum and minimum soil moisture tension values (Supplementary Material 1).
2.3. Water flow investigation
At the macroscopic scale, flow through a porous medium is described by revised Darcy’s law for unsaturated media (Nielsen and Johnson, 1990; Das, 2008) and written as Eq. (1).
| (1) |
In this equation, Q is the flow rate (m3/s), K(Ɵ) is the unsaturated hydraulic conductivity (m/s), A is the area (m2), h is the pressure head or matric potential (m), and z is the elevation head and x is the distance between the two points of interest. ∂(z + h)/∂z is the hydraulic gradient (m/m).
3. Statistical analyses
All statistical analyses were computed using Statistica 9.1 with a significance level of α = 0.05 (Statsoft, 2009). Initially, the datasets were tested for normality using the Shapiro-Wilk (S-W) test. In cases when the data had a non-normal distribution, log-transformed datasets were then tested for normality. When the assumption of normality in the original and log-transformed datasets was not supported nonparametric methods were used. For analysis of the long-term tensiometer average recordings, because of the fairly large data sets, the deviation from normality is not critical (Lindeman, 1974; Statsoft, 2009). A two-way analysis of variance (ANOVA) was used to study of effects of depth and distance on mean tension recording of the tensiometers. For correlation studies, a nonparametric test (Spearman rank order) was carried out for the tensiometers’ minimum, maximum, and tension changes recordings against rainfall depth and depth of water in the gallery. The nonparametric Kruskal-Wallis ANOVA test was employed for the statistical study of tension changes among tensiometers. Multiple comparisons p-values (2-tailed) for all analysis were used to test for difference between the variables. To test the effects of depth and distance on tensiometer recording changes, a one-way ANOVA on original or log-transferred data was used with Tukey HSD post hoc multiple comparisons.
The distance-weighted least squares method (DWLS) was used for displaying the three-dimensional counter lines of soil moisture tension changes. DWLS fits a curve to the data by using the following procedure. A polynomial (second-order) regression is calculated for each value on the X variable scale to determine the corresponding Y value such that the influence of the individual data points on the regression decreases with their distance from the particular X value.
A distributed lag analysis (DLA) is a model for time series data in which a regression equation is used to predict current values of a dependent variable based on both the current values of an explanatory variable and the lagged (past period) values of this explanatory variable. A Distributed Lag Analysis was used to study the lags between hourly data that extracted from a 10-min recording of rainfall and groundwater level in the monitoring wells.
4. Results and discussion
4.1. Tensiometers recording comparisons
Therefore, this analysis assumed that tensiometers did not respond to direct rainfall and they only responded to water exfiltrating from the storage gallery.
The long-term 10-min average of tensiometers readings was used to determine the pattern and variation of the soil moisture tensions in the vadose zone around the permeable pavement storage gallery. This information provided a detailed understanding of the vadose zone drainage behavior. The same dataset was used to study the small and large soil moisture tensions zones and the effects of distance and depths from the storage gallery on recorded soil moisture tensions.
The results are shown in Fig. 4. A two-way ANOVA test was used to investigate the effects of depth and distance from the storage gallery. The two-way ANOVA results showed significant main effects of depth (F(2,458,177) = 1,790,000, p < 0.001) and horizontal distance (F(3,458,177) = 936,000, p < 0.001) and the interaction of depth × horizontal distance (F(2,458,177) = 81,000, p < 0.001) on recorded mean 10-min soil moisture tensions. These results show that, although both the depth and horizontal distance are statistically significant on recorded soil moisture tensions, these two factors also interact. This means that for the locations deeper and farther from the storage gallery, smaller moisture tensions were recorded, and this supports the two-dimensional (horizontal and vertical) exfiltration of water from the storage gallery. The results, surprisingly, showed T(4.5,3) (Mean = 198.60 hPa) was the driest point because of small soil water tension changes and T(2,5) (Mean = 28.25 hPa) was the wettest point because of large soil moisture changes during this study. T(4.5,3), at the same depth as T(2,3), but farther from the gallery, had smaller mean tension. These results and range of recorded tensions are shown in Fig. 4. For the groups (1 to 4 in Fig. 4) of tensiometers installed at the same distances but at different depths, larger soil moisture tensions were associated with the shallow group and decreased from the shallowest to the deepest. The arrows on the figure show that the soil moisture tension is decreasing with an increase with depth.
Fig. 4.
Mean 10-min soil moisture tension of all tensiometers.
These results indicate that the tensiometers located near to the existing ground surface never get highly wet, supporting the assumptions that the tensiometers did not respond to direct rainfall and they only responded to water exfiltrating from the storage gallery.
4.2. The relationship between groundwater fluctuation and tensiometers responses
Tensiometer T(2,5) recorded negative soil moisture tension from mid-December 2015 through early March 2016 (74 days) and from late April through late May 2016 (36 days). Comparing the elevation of T(2,5) with groundwater level in nearby wells (MW3 and MW4) shows that the tensiometer was not submerged, but capillary rise caused the negative readings (deepest tensiometers elevations: 321.64 m ASL (above sea level)).
The tensiometers 6.5 m below the ground surface (T(12.5,6.5), T(0,6.5)W and T(0,6.5)E) recorded negative soil moisture tension for a week from late July through early August 2016. This was because the groundwater was at the highest level in July 2016 (Groundwater level > 321.64 m ASL) and reached these tensiometers (Fig. 5) and, therefore, these tensiometers were submerged.
Fig. 5.
Groundwater levels during the study.
Infiltration is one component of the water cycle. Due to the addition of GI technologies, changes in urban land use, and less demand for groundwater as industries moved to the outskirt of the cities; the groundwater recharge pattern has changed as reported in the recent studies (Lerner, 1990; Pitt et al., 1999; Wolf et al., 2006). There are several demonstrations of local urban groundwater mounding. While not a case in this study, these results suggest that the potential long-term effects of implementing green infrastructure on groundwater level rise in the urban areas must be considered. For example, a prediction of the long-term maximum groundwater level in urban areas is a consideration in the storage gallery design depth. The groundwater levels in MW3 and MW4 showed the measurable (1 cm) groundwater rise after rainfall of at least 2.5 to 4.6 mm. This can be because recharge water from the storage gallery flows away from the storage gallery into the nearby unsaturated zones. Fig. 5 shows the flow direction was usually from the MW3 (6.6 m from the gallery) to MW4 (18.5 m from the gallery).
If the gallery infiltration water is flowing from the gallery to MW3 and then to MW4, the increase in measured groundwater elevation at the wells should show a lag. DLA was used to test the presumptive relationship. The hourly rainfall and groundwater time series (extracted from 10-min recordings) of MW3 and MW4 were investigated using DLA. The results showed that the groundwater depth in both MW3 and MW4 significantly lagged the time of the rainfall by 2 and 3 h respectively, as reported in Table 2. The results showed that the groundwater in MW4 lagged the groundwater recording of the MW3 by 30 min. The recorded lag times of groundwater level data in both wells supported this lag time between wells.
Table 2.
Reports distributed lag analysis results between rainfall depth and groundwater level rise in monitoring wells.
| Independent variable | Dependent variable | Lag | Standard error | p-Value |
|---|---|---|---|---|
| Groundwater level rise in monitoring well 3 (MW3) | Rainfall depth (hourly) | 0 | 3.392 | 0.442 |
| 1 | 9.632 | 0.305 | ||
| 2 | 9.491 | 0.032 | ||
| Groundwater level rise in monitoring well 4 (MW4) | Rainfall depth (hourly) | 0 | 4.427 | 0.547 |
| 1 | 9.865 | 0.404 | ||
| 2 | 9.902 | 0.583 | ||
| 3 | 9.941 | 0.047 | ||
| Groundwater level rise in monitoring well 4 (MW4) | Groundwater level rise in monitoring well 3 (MW3) | 0 | 0.157 | 1.638 |
| 1 | 0.114 | 0.045 | ||
Bold values indicates statistically significance at p <0.05.
4.3. Effect of rainfall characteristics on tensiometer tension recording
Minimum (Tmin), maximum (Tmax), and the difference between the maximum and minimum soil moisture tensions (ΔT) from November 18, 2015 to August 7, 2016 were used for Spearman rank correlation analysis against rain event characteristics and the maximum depths of water in the storage gallery. The period was selected as the events before the surface of the PICP section clogged (Razzaghmanesh and Borst, 2018). During this period, there were twenty storm events (Supplementary Material 2). The correlations were examined using the Spearman rank order correlation because of non-normal distribution of the data. The tested parameters include total rainfall depth (as a surrogate for runoff volume), maximum accumulated water depths in the gallery (providing the required hydraulic head for exfiltration), and antecedent dry weather period (ADP, to investigate soil moisture tensions variations or vadose zone drainage ability) recognizing that the gallery water depth and total rainfall may be covariant.
Four of the correlated tensiometers were located below the pavement section storage gallery and three on the south side, both below and above the elevation of the bottom of the storage gallery. The correlation results (significant correlation highlighted in Table 3) with rainfall depth and maximum water depth in the storage gallery suggest captured water exfiltrates through both the storage gallery bottom and the sidewalls (i.e., with horizontal and vertical flow components). The correlation results suggest if exfiltrated water pattern comprised of two flow components, during exfiltration, the horizontal flow is the primary driver on the sidewall of the storage gallery and the vertical flow is more dominant at the bottom.
Table 3.
Correlation between recorded soil moisture tension and water depths.
| Tensiometers | Correlation with rainfall depth | Correlation with maximum water depth in the storage gallery | ||||
|---|---|---|---|---|---|---|
| Tmax | Tmin | ΔTa | Tmax | Tmin | ΔT | |
| T(0,5)W | −0.12 | −0.60 | 0.76b | 0.05 | −0.46 | 0.92 |
| T(0,6.5)W | −0.15 | −0.48 | 0.64 | 0.09 | −0.20 | 0.74 |
| T(0,5)E | −0.02 | −0.55 | 0.79 | 0.19 | −0.24 | 0.69 |
| T(0,6.5)E | −0.03 | −0.31 | 0.69 | 0.05 | −0.17 | 0.80 |
| T(2,3) | 0.04 | −0.25 | 0.70 | 0.20 | −0.12 | 0.77 |
| T(2,4) | −0.10 | −0.58 | 0.68 | 0.08 | −0.29 | 0.64 |
| T(2,5) | 0.17 | −0.21 | 0.65 | 0.29 | −0.12 | 0.57 |
| T(4.5,3) | 0.04 | −0.01 | 0.03 | 0.37 | −0.27 | −0.13 |
| T(4.5,4) | 0.20 | −0.20 | 0.10 | 0.40 | −0.40 | −0.20 |
| T(4.5,5) | −0.10 | −0.10 | 0.87 | 0.00 | −0.00 | 0.97 |
| T(12.5,5) | 0.34 | −0.39 | 0.20 | 0.47 | −0.46 | −0.17 |
| T(12.5,6.5) | 0.02 | −0.04 | 0.31 | 0.2 | −0.18 | 0.03 |
Bold values indicates statistically significance at p <0.05.
Difference in tension calculated from maximum tension minus minimum tension.
Highlighted correlations are significant at p < 0.050.
It was expected that the minimum tension would decrease with increasing ADP as the soil drained, a period particularly associated with soil gravity drainage. However, ADP was not correlated with three types of soil moisture tensions.
The other results, given in Table 3, show that most of the significant correlations are related to tensiometers located directly under or near the gallery. Maximum soil water tension neither correlated with rainfall depth nor depth of water inside the storage gallery. Minimum soil water tension of just three of the tensiometers beneath the gallery (T(0,5)W, T(0,6.5)W and T(0,5)E) correlated with rainfall depth. It is surprising that the deeper tensiometer on the east side of the gallery did not show a significant correlation with either the rainfall depth or the gallery water level, and this is mainly because of the different infiltration rates at the bottom of the gallery in east and west sides.
The change in tension showed a significant correlation with rainfall depth and depth of water in the gallery for all of the tensiometers (7 in total) located below and within 2 m of the storage gallery. The significant correlation between T(4.5, 5) tension difference (ΔT) and maximum water depth in the storage gallery (Table 3) shows that the flow movement pattern was not horizontal after the T(2,3), T(2,4) and T(2,5) locations because the correlation was not significant with T(4.5,3) and T(4.5,4). Beyond 4.5 m south of the gallery, the vertical component is the primary flow components in the area far from the storage gallery. In the area around the storage gallery, there is a hydraulic gradient from the bottom and side of the storage gallery into the surrounding media. Similar to the seepage pattern under the hydraulic structures for solving Laplace’s flow equation, some assumptions were considered, including that the shape of the flow path should be a parabola (Das, 2008). Hence, it is possible to say that, in this study, exfiltrated water followed a downward parabola shape below and around the storage gallery. This parabola elongates under the bottom of the storage gallery when the ponding depth is small inside the gallery, and, because of deeper rainfall, the elongation goes to the sidewalls when there is a larger water ponding inside the storage gallery.
4.4. Tensiometer responses to storm events
The changes in soil moisture tension were compared for all the tensiometers during 20 rain events. The tensiometers located below the storage gallery and those located near the south side showed larger changes than those further away (4.5 m from the gallery and beyond).
Tensiometer T(2,3), located 2 m from the south edge of the gallery, showed the largest change (Fig. 6). This result supports the hypothesis that water exfiltrates through both the sidewalls and the bottom of the storage gallery as the tensiometer located on the south side and those located under the gallery show some degree of soil moisture tension changes or wetting. This result proves that, if the bottom of the storage gallery becomes clogged, the water will exfiltrate from the sidewalls and it will not eliminate the long-term storage gallery exfiltration.
Fig. 6.
(A) Changes in tensiometer for the 20 rain events. (B) Contour plot of tension changes against distance and depth.
The nonparametric Kruskal-Wallis ANOVA test was employed for statistical study among tensiometers tension changes. A statistical significant difference was detected among tension differences (H (11, N = 194) = 65.35763, p < 0.01). The tensiometers nearer the sidewall and under the gallery had larger tension changes than those located farther from the gallery.
To study the water movement directions, a distance weighted least squares method (Helsel and Hirsch, 2002) was used to visualize the tension difference zones. A contour area plot of the tension changes against distance and depth of the tensiometers are shown in Fig. 6B.
Three zones were recognized from the contours plot of soil moisture tension differences. First, the area on the sidewall (south) and below the storage gallery. The gradient in tensiometer in Fig. 6A and B shows the water moving from the storage gallery into this area, and maximum differences in soil moisture changes were recorded in this zone. The second area goes from 5 m to 12 m from the storage gallery, which it is not under the influence of the storage gallery. The negative tension changes happened in this zone which indicates the tensiometers were submerged in this zone. For the third zone, beyond 12 m, small changes were recorded, and it appears that water is flowing from the south.
4.5. Effects of depth and distance on soil moisture tension changes
In this study, tensiometers were installed at selected depths and distances from the permeable pavement storage gallery to determine the effects of depth and distance on tensiometer soil moisture tension changes. A series of tensiometers that was located at a same depth and another series that was installed at the same distances from the storage gallery were selected for this section. For depth investigations, two groups comprised of those tensiometers located at 2 m and 4.5 m south of the gallery and in three depths of 3 m, 4 m and 5 m below the ground surface were selected. The results are shown in Fig. 7.
Fig. 7.
Differences in soil moisture tensions differences (A) various depths distance 2 m, (B) various depths in distance of 4.50 m, (C) various distances and same depths of 6.50 m
A one-way ANOVA test was employed on log-transformed tension changes. No significant differences were detected in tension changes with depth (F(2, 77) = 3.0350, p > 0.054) among the three sensors installed 2 m south of the infiltration gallery. There were statistically significant differences among the tension changes located at 4.5 m (F(2, 30) = 3.5607, p = 0.04). The soil moisture tension changes decreasing with increased depth and distance from the storage gallery. The shallower tensiometers showed larger tensions changes than the deeper tensiometers. This reduction in tension changes with increasing depth and distance from the storage gallery suggests that the parabolic flow path did not extend far from the gallery.
The tensiometers located 5 m below the surface at four distances [0 (the midpoint of the gallery as the datum), 2 m, 4.50 m, and 12.50 m south of the gallery] were used for the depth study. The results (Fig. 7C) show that the maximum tension change occurs to the tensiometer located at 2 m from the gallery, followed by the tensiometers located below the gallery and all other tensiometers located south of the gallery. A one-way ANOVA test was employed on tension changes and in this condition, there were statistically significant differences among the tensiometer data (F(3, 89) = 11.267, p < 0.01). The tensiometers located at the same depth but various distances from the gallery showed that there was a significant difference among tension changes of those tensiometers. Larger soil tension change was associated with the tensiometers farther from the permeable pavement storage gallery.
4.6. Water flow variations around storage gallery
The unsaturated form of Darcy’s law was used to calculate the flow rate between selected tensiometers to the flow after each event as reported in Table 4. Following the recommendations by Wösten and Van Genuchten (1988) for unsaturated hydraulic conductivity, a unit cross section was considered to solve the equation. Hydraulic heads were calculated from the elevation and measured soil moisture tensions. The elevation of deepest tensiometer (321.64 m ASL) was the datum used to solve the flow equation. The results show exfiltration from both the sidewalls and the bottom of the gallery. Higher flow rates were calculated from the tensiometers located near the sidewalls than under the bottom of the storage gallery. The vertical water flux is 4–12-times greater than the horizontal fluxes between 5 and 6.5 m below the existing ground level.
Table 4.
Estimated maximum flow (m/s/per square meter) between selected tensiometers.
| Selected tensiometers | Location | Mean | SDa |
|---|---|---|---|
| T(0,5)W to T(0,5)E | Below storage gallery | −0.0003 | 0.00032 |
| T(0,5)W to T(0,6.5)W | Below storage gallery | 0.0195 | 0.01221 |
| T(0,6.5)W to T(0,6.5)E | Below storage gallery | 0.00015 | 0.000167 |
| T(2,3) to T(2,4) | Sidewall | 0.0309 | 0.010593 |
| T(2,4) to T(2,5) | Sidewall | 0.03230 | 0.00906 |
SD = standard deviation.
5. Conclusions
In this study, tensiometers and ground monitoring wells were used to measure subsurface water flow patterns from the storage gallery under a permeable pavement site. Based on this study, the following conclusion can be made. The exfiltration potentially followed a downward parabolic shape, and this shape depended on the depth of accumulated water in the gallery. The sidewalls are capable of effective exfiltration and, if the bottom of the storage gallery clogged, the sidewalls will be an alternative flow path. The sidewalls had an exfiltration potential similar to the bottom of the storage gallery. Accumulated water inside the galley and total rainfall depth were positively correlated with the soil tension changes. There was a positive correlation between groundwater rise and total rainfall depth. Tensiometers beyond 4.5 m far from the storage gallery have shown little change in tension after the rain events, indicating that the effective recharge radius of the storage gallery was <4.5 m, this is important in the highly urbanized urban environment before the construction of a pavement section with storage gallery for infrastructure offsets. The distance between the bottom of the storage gallery and the predicted long-term peak groundwater level rise should be considered for successful exfiltration and effective groundwater recharge. These results, like most SCM studies, are site specific. However, these results show the pattern to expect at similar sites using designs. Further, the results support the narrow-deep shape of the storage gallery to allow for side wall exfiltration. The results also provide guidance for placement of instruments at future monitoring sites. This study was conducted in a secure, controlled environment with enough land space. The installation was done without any interaction with traffic and no vandalism happened during the monitoring period. The same methodology and instruments can be used in monitoring green infrastructure in dense urban environment. It is recommended that the tensiometers be installed during the construction of the GI and the street sidewalks can be used for drilling boreholes.
Supplementary Material
Supplementary material 2. Rainfall event characteristics during this study.
Supplementary material 3. Tensiometrs Quality Control
Supplementary material 1. Maximum, minimum and soil moisture tension difference definition.
Highlights.
Tensiometer approach used for monitoring performance of urban green infrastructure.
Exfiltration was evident from both the sidewalls and the bottom of the storage gallery.
Exfiltration potentially followed a downward parabolic shape.
Accumulated water inside the galley and total rainfall depth were positively correlated with the soil tension changes.
Tensiometers beyond 4.5 m far from the storage gallery have shown little change in tension after the rain events.
Acknowledgments
This article was supported in part by an appointment to the ORISE Research Participant Program supported by an interagency agreement between EPA and DOE and might not necessarily reflect the views of the EPA. The authors would like thank Dr. Douglas Beak and Mr. Steven Acree of the US EPA’s ORD for their advice. The writers also would also like to thank the following parties for their assistance with this project: the US Army, School District, EPA Region 7, and Corps of Engineers.
Disclaimer
The US EPA, through its Office of Research and Development, funded and managed, or partially funded and collaborated in, the research described in this paper. It has been subjected to the Agency’s administrative review and has been approved for external publication. Any opinions expressed in this paper are those of the writers and do not necessarily reflect the views of the Agency; therefore, no official endorsement should be inferred. Any mention of trade names or commercial products does not constitute endorsement or recommendation for use.
Footnotes
The following are the supplementary data related to this article.
References
- Bettess R, 1996. Infiltration Drainage: Manual of Good Practice. Construction Industry Research and Information Association (CIRIA), London, UK: C. I. R. a. I. A. C. Rep. R156. [Google Scholar]
- Braga A, Horst M, Traver RG, 2007. Temperature effects on the infiltration rate through an infiltration basin BMP. J. Irrig. Drain. Eng 133 (6), 593–601. [Google Scholar]
- Brown RA, Borst M, 2014. Evaluation of surface infiltration testing procedures in permeable pavement systems. J. Environ. Eng 140 (3), 04014001. [Google Scholar]
- Das BM, 2008. Advanced Soil Mechanics. Taylor & Francis, 270 Madison Ave, New York, NY: 10016, USA. [Google Scholar]
- Driscoll ED, Palhegyi GE, Strecker EW, Shelly P.e., 1989. Analysis of Storm Event Characteristics for Selected Rainfall Gages Throughout the United States. Environmental Protection Agency, Washington DC. [Google Scholar]
- Emerson CH, Wadzuk BM, Traver RG, 2010. Hydraulic evolution and total suspended solids capture of an infiltration trench. Hydrol. Process. Int. J 24 (8), 1008–1014. [Google Scholar]
- eWater, 2013. Model for Urban Stormwater Improvement Conceptualisation (MUSIC). Melbourne, Australia. [Google Scholar]
- Hamlet AF, Lettenmaier DP, 2007. Effects of 20th century warming and climate variability on flood risk in the Western U.S. Water Resour. Res 43 (6) (n/a-n/a). [Google Scholar]
- Helsel DR, Hirsch RM, 2002. Statistical methods in water resources Techniques of Water-Resources Investigations of the United States Geological Survey, Book 4, Hydrologic Analysis and Interpretation. U.S. Department of the Interior. [Google Scholar]
- HydroCAD Software, 2011. HydroCAD Stormwater Modeling System Version 10 Owner’s Manual. HydroCAD Software Solutions LLC. [Google Scholar]
- InfoWorks Software, 2002. Modelling Sustainable Urban Drainage Structures. [Google Scholar]
- KAW VALLEY Engineering Inc., 2011. Geotechnical Engineering Report Forsyth Elementary School Rifle Range Road and Potomal Drive Fort Riley, Kansas (Lenexa, Kansas). [Google Scholar]
- Kirkham MB, 2005. Principles of Soil and Plant Water Relations. Elsevier Academic Press, 200 Wheeler Road, 6th Floor, Burlington, MA: 01803, USA. [Google Scholar]
- Lee JG, Borst M, Brown RA, Rossman L, Simon MA, 2015. Modeling the hydrologic processes of a permeable pavement system. J. Hydrol. Eng 20 (5). [Google Scholar]
- Lerner DN, 1990. Groundwater recharge in urban areas. Atmos. Environ. B Urban Atmos 24 (1), 29–33. [Google Scholar]
- Lindeman HR, 1974. Analysis of Variance in Complex Experimental Designs. W.H. Freeman & Co., San Francisco. [Google Scholar]
- Nielsen D, Johnson A, 1990. Ground Water and Vadose Zone Monitoring. ASTM International, West Conshohocken, PA. [Google Scholar]
- Pitt R, Voorhees J, 1995. Source Loading and Management Model (SLAMM). [Google Scholar]
- Pitt R, Clark S, Field R, 1999. Groundwater contamination potential from stormwater infiltration practices. Urban Water 1 (3), 217–236. [DOI] [PubMed] [Google Scholar]
- Razzaghmanesh M, Borst M, 2018. Investigation clogging dynamic of permeable pavement systems using embedded sensors. J. Hydrol 557, 887–896. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rossman LA, 2015. Storm Water Management Model User’s Manual Version 5.1. EPA/ 600/R14/413b. United States Environmental Protection Agency, Cincinnati, OH. [Google Scholar]
- Schlüter W, Jefferies C, Zhang X, 2007. Modelling of flow through gravel-filled trenches. Urban Water J. 4 (4), 241–251. [Google Scholar]
- Sisson JB, Gee GW, Hubbell JM, Bratton WL, Ritter JC, Ward AL, Caldwell TG, 2002. Advances in tensionmetery for long-term monitoring of soil water pressures. Vadose Zone 1, 310–315. [Google Scholar]
- Smith DR, 2006. Permeable Interlocking Concrete Pavements, Selection, Design, Construction and Maintenance. 62. I. C. P. Institute, Washington, D.C. [Google Scholar]
- Staher P, Urbonas B, 1990. Stormwater Detention for Drainage, Water Quality, and CSO Management. Prentice Hall, Englewood Cliffs, NJ. [Google Scholar]
- Statsoft, 2009. STATISTICA, General Conventions and Statistic I. V. 9.1. Tulsa, OK. [Google Scholar]
- Todd DK, Mays LW, 2005. Groundwater Hydrology. John Wiley & Sons, Inc., Hoboken, NJ. [Google Scholar]
- Toll DG, Lourenço SDN, Mendes J, Gallipoli D, Evans FD, Augarde CE, Cui YJ, Tang AM, Rojas JC, Pagano L, et al. , 2011. Soil suction monitoring for landslides and slopes. Q. J. Eng. Geol. Hydrogeol 44 (1), 23–33. [Google Scholar]
- UMS, 2009. TS1 Tensiometer Manual. UMS GmbH, München, Germany. US Army, 2014 Programmatic Environmental Assessment Military Engineering Fort Riley, Kansas. [Google Scholar]
- USDOT, 2008. Long Term Pavement Performance Computed Parameter: Frost Penetration. P. N. FHWA-HRT-08–057; McLean, VA. [Google Scholar]
- Wolf L, Klinger J, Held I, Hötzl H, 2006. Integrating groundwater into urban water management. Water Sci. Technol 54 (6), 395–403. [DOI] [PubMed] [Google Scholar]
- Wösten J, Van Genuchten MT, 1988. Using texture and other soil properties to predict the unsaturated soil hydraulic functions. Soil Sci. Soc. Am. J 52 (6), 1762–1770. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary material 2. Rainfall event characteristics during this study.
Supplementary material 3. Tensiometrs Quality Control
Supplementary material 1. Maximum, minimum and soil moisture tension difference definition.