You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/333838640

Evapotranspiration in green stormwater infrastructure systems

Article  in  Science of The Total Environment · June 2019


DOI: 10.1016/j.scitotenv.2019.06.256

CITATIONS READS

4 329

3 authors, including:

Ali Ebrahimian Robert Traver


Villanova University Villanova University
15 PUBLICATIONS   194 CITATIONS    102 PUBLICATIONS   1,449 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Effective Impervious area in urban watersheds View project

EPA STAR (Science To Achieve Results) Grant: Next Generation Volume Reduction Green Infrastructure Stormwater Control Measures in Support of Philadelphia’s Green
City Clean Waters Initiative View project

All content following this page was uploaded by Ali Ebrahimian on 08 July 2019.

The user has requested enhancement of the downloaded file.


Science of the Total Environment 688 (2019) 797–810

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Review

Evapotranspiration in green stormwater infrastructure systems


Ali Ebrahimian ⁎, Bridget Wadzuk, Robert Traver
Department of Civil and Environmental Engineering, Villanova University, 800 E. Lancaster Ave., Villanova, PA 19085, USA

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• ET is an important component of the hy-


drologic cycle in vegetated GSI.
• ET is dependent on both meteorological
factors and GSI properties.
• Dynamics of ET and infiltration should
be considered in vegetated GSI design.

a r t i c l e i n f o a b s t r a c t

Article history: Evapotranspiration (ET) is a viable runoff reduction mechanism and an important player in the hydrologic cycle
Received 9 April 2019 of vegetated green stormwater infrastructure (GSI). As a dynamic process, ET is dependent on both meteorolog-
Received in revised form 13 June 2019 ical factors (e.g., rainfall characteristics, relative humidity, and air temperature) and GSI properties (e.g., soil
Accepted 16 June 2019
media type). This paper investigates the role of ET in runoff volume reduction of green roofs and rain gardens
Available online 17 June 2019
through a comprehensive literature review. Evapotranspiration is mostly unaccounted in the design and
Editor: Ashantha Goonetilleke crediting of GSI systems because of the complex interaction of soil, plants, and climate that makes its quantifica-
tion difficult. To improve vegetated GSI design for runoff volume reduction, design methods should consider ET
Keywords: and infiltration processes concurrently. Two methods, complex and simple, are reviewed and discussed herein.
ET The simple method requires minimal input information compared to the more complex continuous simulation
Infiltration method; however continuous simulation yields volume reduction values more similar to field observations. It
Bioinfiltration is demonstrated that modifying the drainage structure and using fine-grained in-situ soils can potentially in-
Bioretention crease ET in vegetated GSI systems. None of the available ET predictive equations, mostly derived from agricul-
Rain garden
tural sciences, are found to precisely match observed GSI ET data. Until further research is conducted on GSI ET
Green roof
estimation methods, the 1985 Hargreaves method is recommended when performing continuous simulations.
Green infrastructure
The 1985 Hargreaves method is simple, requires limited input data that are readily available, and generates rea-
sonable results. Technical recommendations and directions for future research are provided.
© 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).

⁎ Corresponding author.
E-mail addresses: ali.ebrahimian@villanova.edu (A. Ebrahimian), bridget.wadzuk@villanova.edu (B. Wadzuk), robert.traver@villanova.edu (R. Traver).

https://doi.org/10.1016/j.scitotenv.2019.06.256
0048-9697/© 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
798 A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 798
2. ET significance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799
2.1. ET in urban hydrologic cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799
2.2. ET in GSI runoff volume control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799
2.2.1. Green roofs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799
2.2.2. Rain gardens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 800
3. ET determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 800
3.1. Potential ET versus actual ET. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 800
3.2. PET estimation methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 800
3.3. Comparison of PET and AET in GSI systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 802
3.3.1. Green roofs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 802
3.3.2. Rain gardens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 802
3.4. Stress point and available water for ET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 802
4. ET and GSI design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804
4.1. Static methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804
4.2. Dynamic methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805
4.2.1. Complex method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805
4.2.2. Simple method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805
4.3. Technical recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 806
4.3.1. Drainage condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 806
4.3.2. Native (in-situ) soils in the media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 806
5. Research needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 806
5.1. Continuous simulation of ET and exfiltration in GSI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
5.2. Simple tools to account for ET in GSI design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
5.3. ET predictive methods in GSI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
5.4. Stress point and PAW in GSI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
5.5. Long term monitoring of GSI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
5.6. Drainage conditions for improved ET and other benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
5.7. Soil media design for ET-infiltration balance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
6. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 808

1. Introduction incorporating ET into design when GSI is designed for other objectives,
particularly with finer temporal resolutions (e.g., flood control) or non-
Urban flooding and water pollution are common problems asso- surface systems (e.g., groundwater recharge, stormwater reuse). Focus-
ciated with stormwater management in urbanized areas (Shariat ing on rain gardens and green roofs, a comprehensive literature review
et al., 2019). Green stormwater infrastructure (GSI) mitigates urban- study was executed and shows the significance of ET in GSI system per-
ization impacts by disconnecting impervious areas from storm sewer formance. Further, different methods for estimating ET were investi-
systems and using the hydrologic cycle components to reduce runoff gated, and ideas for resolving the main research question were
volume and pollutant loads (Ebrahimian et al., 2016a, 2016b, 2018). developed. Technical recommendations regarding the design and
In cities, such as New York City, New York, Philadelphia, Pennsylva- crediting of rain gardens and green roofs, as well as directions for future
nia (PA), London, UK, and Paris, France, that widely utilize combined research, are also provided. It should be noted that the term “rain gar-
sewer systems to manage stormwater and sewage, urbanization im- den” is used in this manuscript to be inclusive of both bioinfiltration
pacts are intensified because even relatively small rainfall events can and bioretention systems.
cause combined sewer overflows (CSOs) (Montalto et al., 2007). As Peer-reviewed journal publications were the main references in
such, cities are using vegetated GSI systems (e.g., rain gardens and this article. However, other publications such as conference proceed-
green roofs) that reduce runoff entering the sewer system and, sub- ings and technical reports were reviewed and cited as appropriate.
sequently, CSOs. Evapotranspiration (ET) (i.e., evaporation from The references encompass a wide range of areas including hydrol-
plants and land surface plus transpiration from plants) is an impor- ogy, water resources science and engineering, environmental sci-
tant post-storm element of the hydrologic cycle and plays an impor- ence and engineering, irrigation and drainage engineering, soil
tant role in the performance of vegetated GSI systems (Berland et al., sciences, landscape planning, ecohydrology, and ecological engi-
2017; Muerdter et al., 2018; Feng, 2018). Eger et al. (2017) report the neering. While the article is within the context of urban GSI, refer-
average and median ET portion of water budgets as 61% and 64%, re- ences from agricultural research have been presented to support
spectively, in green roofs (N = 59) and 37% and 28%, respectively, in the statements and discussions where needed. Reviewing the role
bioretention cells (N = 10), where N is the number of green roofs of plant types on ET in vegetated urban GSI systems is beyond the
and bioretention cells. scope of this paper. Examples of such studies can be found in Wolf
When the goal of GSI design is to reduce the runoff volume and and Lundholm (2008), Lundholm et al. (2010), and Zhang et al.
knowing that ET is a substantial component of an annual water budget, (2018a) for green roofs, Hunt et al. (2015), and Payne et al. (2018)
ET needs to be considered in GSI design. Evapotranspiration contributes for rain gardens, and Tirpak et al. (2018) and Zhang and Chui
to the restoration of soil media pore storage in vegetated GSI through (2019) for other GSI systems. Additionally, this paper did not seek
moisture reduction after storm events. Hence, the main research ques- to evaluate and compare existing design guidelines or manuals ex-
tion becomes how to incorporate ET in the volume-reduction design plicitly, but they are mentioned as necessary to demonstrate where
and crediting of vegetated GSI systems. Care must be taken in ET is or is not considered in design.
A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810 799

2. ET significance regulations that require or include ET in the design and crediting nor
vegetation or soil media guidelines for maximizing the ET benefits.
2.1. ET in urban hydrologic cycle Hence, considering ET in the design and crediting of vegetated GSI sys-
tems deserves further investigation (Berland et al., 2017). In-situ ET
Simplified hydrologic models usually neglect ET in urban areas; measurement in vegetated GSI systems (e.g., using chamber systems-
however, ET has been found to be an important component of urban hy- Hamel et al., 2014) is difficult (Szota et al., 2018). Therefore, ET is usually
drologic cycle when trees and vegetated surfaces cover notable portions estimated using other methods such as 1) measurements of other water
of the urban area (Grimmond and Oke, 1991, 1999; Kuehler et al., 2017; budget components through lysimeter/column/box studies (see
Berland et al., 2017). For example, ET was found to constitute 38% of an- Tables 1 and 3), 2) potential ET (PET) estimation equations (see
nual and 81% of summer water balance losses in Vancouver, Canada Section 3), and 3) stomatal conductance measurements at the leaf-
(Grimmond and Oke, 1986). Scharenbroch et al. (2016) reported a 46 scale (e.g., Scharenbroch et al., 2016- beyond the scope of this paper).
to 72% share for transpiration in total water outputs of parking lot
bioswales with trees in Chicago, Illinois. While ET has been heavily stud- 2.2.1. Green roofs
ied in agriculture, it is largely unaccounted for in the design of urban GSI. Unlike other GSI systems that rely on infiltration as the primary
Several concepts and findings from agricultural literature can be modi- stormwater control mechanism, a green roof (with restricted
fied for and applied to the urban environments. underdrains) primarily retains water within its soil media and
evapotranspirates this water over time. Runoff volume reduction is
2.2. ET in GSI runoff volume control the most widely reported performance metric for green roofs. It is calcu-
lated as the percentage of total captured rainfall by the green roof dur-
The volume control in vegetated GSI is mostly attributed to infiltra- ing the study period and sometimes is called percent retention. Palla
tion in current practices, although ET can be considered a viable mech- et al. (2010) performed a review on green roof experimental studies
anism (Li et al., 2009; Denich and Bradford, 2010; Zaremba et al., 2016; in different cities in the US and Europe (Sweden, Belgium, and
Hess et al., 2017). After a rainfall event, there is a recovery period where Germany) and concluded that green roof volume retention may range
infiltration and ET decrease the soil moisture and restore the soil media from 40% to 80% of the total rainfall. Similarly, Carson et al. (2013)
pore volume available for the next rainfall event. The initially high infil- reviewed pilot scale green roof studies conducted on elevated test
tration rates after rainfall events reduce as the soil moisture increases boxes or similar modules with an area of 0.37 to 12 m2 that have re-
but ET continues throughout the no-rain period while adequate water sulted in 30% to 86% volume reduction. Overall, while there have been
is available for the ET process (Wadzuk et al., 2015). Maimone (2010) studies documenting ET in green roofs, there is wide variability and un-
investigated the effect of ET on urban tree pit overflows and identified certainty to the controlling mechanisms, so care should be taken in gen-
the need to account for ET if there are poor soil infiltration conditions eralizing the performance of any green roof to a proposed green roof
or lined systems. with different meteorological conditions or engineering properties
Accounting for ET as a volume control mechanism can provide a (Harper et al., 2015).
more process-based design for vegetated GSI systems (Berland et al., The variability of the reported runoff volume reductions in green
2017). While there are a few places around the world that consider ET roofs (Berndtsson, 2010) can be because of meteorological factors
in the design of vegetated GSI through continuous simulations, e.g., in (e.g., antecedent rainfall, antecedent dry time, rainfall depth and inten-
Australia (DPLG, 2009; Melbourne Water, 2018), New Zealand (Lewis sity, relative humidity, wind speed, solar radiation, and time of year)
et al., 2015), and Western US (WSDE, 2012), this is not the norm. In and green roof properties (e.g., media water retention capacity, vegeta-
the US, for example, Wadzuk (2013), LeBoon (2014), and Hess (2017) tion, slope, and age) (Fassman-Beck et al., 2013; Hakimdavar et al.,
looked at how different cities and states give volume reduction credit 2014; Harper et al., 2015; DiGiovanni-White et al., 2018). Hakimdavar
to rain gardens and green roofs. They found that there are often neither et al. (2014) identified rainfall depth and duration as the most

Table 1
Pilot scale studies that highlight the significance of ET in green roofs through measurements.

Study Location ET performance Remarks

Volume Rate
reduction (mm/day)
(%)

Bengtsson (2005) Malmö, Sweden Up to 64 July 2001–Dec. 2002


VanWoert et al. (2005) East Lansing, MI 49–83 Overall values in Aug. 2002-Oct. 2003.
49% for a standard commercial roof with gravel ballast and 83% for an extensive green roof
with vegetation
DeNardo et al. (2005) Centre County, PA 45 3 October and November 2002
Stovin (2010) Sheffield, UK 34 During spring 2006
Voyde et al. (2010a) Auckland, New 37–64 2.2 January to March 2007 and 2008
Zealand Minimum retention for unplanted trays
Stovin et al. (2012) Sheffield, UK 50 Almost-continuously over a 29-month period from 1/1/2007 to 5/31/2009
Ouldboukhitine et al. La Rochelle, France 2.5 Using an experimental setup during week 44 (November 1–7) of 2010
(2012)
Sherrard Jr and Jacobs Durham, NH 32 0.5–1.2 Between August to November 2009
(2012) 2.8 0.5 and 1.2 = ET rate for the coolest and warmest months, respectively
2.8 = Maximum daily ET
DiGiovanni et al. (2013) Bronx, NY 0.5–3.5 Between June 2009 to June 2010
Wadzuk et al. (2013) Villanova, PA 1 to 10 On average, 3, 3, and 4 mm/day in 2009, 2010, and 2011, respectively, between April and
November.
Zaremba et al. (2016) Villanova, PA 68–88 3.1 2009–2014 between April and November
Feng et al. (2018) Salt Lake City, UT 2.01, 2.52, Average values for non-vegetated, sedums, and grass covers during 2014
2.69
Loiola et al. (2018) Rio de Janeiro, 58 Average for modular-tray green roofs in January to March 2015
Brazil
800 A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810

influencing factors on overall green roof retention and peak runoff re- 2.5 mm/day on a typical green roof with a 1:1 loading ratio, ET accounts
duction performance when investigating two full-scale and one test- for 50% of the volume removal, while a typical rain garden with a load-
box green roof in New York City, although rainfall intensity and ante- ing ratio of 5:1 has 10% volume reduction by ET. Although a higher load-
cedent dry time had little influence. Conversely, in other studies, obser- ing ratio decreases the ET percentage by influent volume, it may provide
vations showed the performance of similar green roofs varies more available water for increased and prolonged ET (Wadzuk et al.,
substantially because retention ability is strongly affected by the local 2015). Other design factors such as the presence and characteristics of
climate (e.g., seasonal and short-term distribution of rainfall events liners, underdrains, and internal water storage (IWS) may alter the ob-
and net radiation dictating ET rates) and the antecedent dryness served ET percentage (Eger et al., 2017).
(Villarreal and Bengtsson, 2005; Stovin et al., 2012; Carson et al., Selbig and Balster (2010) measured micrometeorological factors in
2013; Krebs et al., 2016). It is evident that the stormwater retention two side-by-side rain gardens (four rain gardens in total) with the
ability is sensitive to the initial moisture condition of the green roof. same size and loading ratio but different vegetation and soil type in
Microclimates, along with specific system properties, may explain Madison, Wisconsin, to estimate reference ET for the FAO-56 Penman-
variability in green roof performance. DiGiovanni-White et al. (2018) Monteith (PM) equation (Allen et al., 1998). The ET ranged from 12%
found up to a 40% difference in estimated annual ET values (using the to 25% of the total influent volume during April–November 2005
ASCE standardized Penman-Monteith equation for short reference sur- (Selbig and Balster, 2010). In Table 3, the volume reductions for each
faces; ASCE, 2005) for six weather stations at four GSI sites (three green study are the percentage of total rainfall that is evapotranspirated. The
roofs and one rain garden) and two airports in New York City over a difference between volume reduction values in different studies is due
one-year period. While highlighting the importance of onsite microme- to the experiment configuration and drainage structure, soil and vegeta-
teorological data measurements in accurate ET estimation, DiGiovanni- tion types, study time (season), micrometeorological conditions in the
White et al. (2018) recommended a more comprehensive study that study location, and loading ratio.
also evaluates the effect of non-meteorological conditions, such as veg-
etation and soil media types, in GSI systems. Additionally, a better un- 3. ET determination
derstanding of the soil moisture content behavior during dry periods
due to ET should result in the development of more accurate approaches 3.1. Potential ET versus actual ET
for long-term simulations (Berretta et al., 2014).
Most published observations on the hydrologic performance of Evapotranspiration is a difficult phenomenon to quantify because of
green roofs are based on pilot scale studies. Greater areal plant coverage, the complex interaction between the ground surface, vegetation, and
higher transpiring plants, and warmer weather are among contributing atmosphere (Carbone et al., 2016). Actual ET (AET), or the quantity of
factors to increased ET rates in these studies (Carson et al., 2013). water that leaves a surface based on climatological demand and soil
Table 1 presents a review of pilot scale green roof studies (excluding water availability (i.e., the water-limited ET), depends on surface, sub-
modeling studies) that highlight the significance of ET. The ET rates surface, and meteorological conditions. Potential ET (PET) (i.e., non-
(Table 1) are in the range of 0.5 to 3.5 mm/day except for Wadzuk water-limited ET) is the maximum amount of water that would leave
et al. (2013) that reports a range of 1–10 mm/day. The high ET rates a surface from a healthy, well-watered crop growing uniformly in a
in the latter study were explained by extreme rainfall events in August large area based on available energy. The PET does not consider the dif-
and September 2011 and ponding at the undrained green roof followed ferences between plant species or environmental stresses (Wadzuk
by sunny days that provided high energy for the ET process. The average et al., 2013). The difference between PET and AET is primarily a function
daily ET in Wadzuk et al. (2013) was 3 to 4 mm in different years of of soil water availability (Marasco et al., 2015). The AET becomes less
study that agrees with other studies (Table 1). The reported volume re- than PET as soil moisture decreases from field capacity. Various soil
duction of reviewed studies (32–88%; Table 1), is within the range of moisture extraction functions based on a combination of actual soil
Carson et al. (2013) and Palla et al. (2010). moisture, field capacity, and wilting point can be used to relate the de-
The volume reduction in full-scale green roofs is generally lower gree of substrate saturation to AET (Zhao et al., 2013; Marasco et al.,
than pilot-scale ones. This might be explained by non-vegetated sec- 2015).
tions and irrigation requirements (Carson et al., 2013; Cipolla et al.,
2016). A review of the hydrologic performance of 44 full-scale green 3.2. PET estimation methods
roofs in the US, Canada, New Zealand, China, and Europe (Table 2)
yielded an overall retention between 11% to 77% with a median, aver- Several empirical and analytical methodologies are available to de-
age, and standard deviation of 57%, 50%, and 21%, respectively. These re- termine PET, including Hargreaves, Slatyer-McIlroy, Priestley-Taylor,
tention values are based on a median monitoring duration of one year of and Blaney-Criddle (empirical equations) and Penman and Penman-
green roofs with a median drainage area and substrate depth of 170 m2 Monteith (PM) (analytical equations). The PM equation (Monteith,
and 100 mm, respectively. 1965) supersedes the Penman equation (Penman, 1948). The Food
and Agriculture Organization (FAO) of the United Nations recommends
2.2.2. Rain gardens a simplified version of the ASCE-PM equation for a short crop reference
There are fewer published studies on ET measurement in rain gar- surface (similar to grass) with a vegetation height, surface resistance,
dens (Table 3- excludes modeling studies) compared to green roofs, and albedo of 0.12 m, 70 s/m, and 0.23, respectively, which is widely
which is likely due to difficulty measuring ET and the perception that known as FAO-56 PM equation (Allen et al., 1998). Jensen et al.
the ET in rain gardens is small (e.g., Brown and Hunt, 2011a; 2011b (1990) presented a PM-based ET equation through the American Soci-
found ET to be b5% of a bioretention water budget using the ety of Civil Engineers (ASCE) Manual 70 which is known as ASCE-PM
Thornthwaite (1948) method for PET and the Zhang et al. (2001) equa- equation. The ASCE recommends two versions of the ASCE-PM equation
tion for estimating actual ET from PET) (Szota et al., 2018). However, for short and tall crop reference surfaces which are known as ASCE stan-
there are a few studies that demonstrate that ET may be significant in dardized PM equations (ASCE, 2005). The reference vegetation height
a rain garden water budget (see Table 3; 19–84% volume reduction). and albedo parameters as well as the surface resistance for daily calcu-
Relative to green roofs, the hydraulic loading ratio (the ratio of directly lations of short reference surfaces are the same in both the ASCE stan-
connected impervious area to the rain garden area) for rain gardens dardized PM and FAO-56 PM methods. However, the ASCE
tends to be large, yielding more inflow and subsequently less ET standardized PM uses two different surface resistance values of 50 s/m
percentage-wise (Selbig and Balster, 2010; Brown et al., 2011). For ex- and 200 s/m for daytime and nighttime, respectively, in hourly ET calcu-
ample, using 5 dry days after a 25 mm rainfall and an ET rate of lations (Allen et al., 1998; Walter et al., 2000; ASCE, 2005; Irmak et al.,
A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810 801

Table 2
Hydrologic performance of full-scale green roofs.

Authors Publication type Study location Monitored area Substrate depth Length of monitoring Overall retention
(m2) (mm) (months) (%)

Hutchinson et al. (2003) Conference paper Portland, OR 240 100–125 15 69


Moran et al. (2005) Conference paper Goldsboro, NC 35 75 17 63
Raleigh, NC 65 100 2 55
Liu and Minor (2005) Conference paper Toronto, Canada 200 75 20 57
Toronto, Canada 200 100 20 57
Carter and Rasmussen (2006) Journal paper Athens, GA 21.3 76.2 13 77
Connelly et al. (2006) Technical Report Vancouver, Canada 33 75 11 29
Vancouver, Canada 33 150 11 26
TRCA (2006) Technical Report Toronto, Canada 240 140 15 65.3
Teemusk and Mander (2007) Journal paper Tartu, Estonia 120 100 1 19.6
Spolek (2008) Journal paper Portland, OR 290 100–150 30 12
Portland, OR 280 100–150 30 17
Portland, OR 500 150 33 25
Hathaway et al. (2008) Journal paper Goldsboro, NC 35 75 14 64
Kinston, NC 27 100 14 64
Kurtz (2008) Conference paper Portland, OR 246 125 73 56
Portland, OR 465 75 15 64
Berkompas et al. (2008) Conference paper Seattle, WA 743 150 10 30.5
Seattle, WA 1860 100–125 2 33
Seattle, WA 80 150 2 17.1
Bliss et al. (2009) Journal paper Pittsburg, PA 330 140 5 21.8
Berghage et al. (2009) Technical Report Chicago, IL 7000 76 23 74
Voyde et al. (2010a) Journal paper Auckland, NZ 41 50 12 66
Auckland, NZ 13 50 12 66
Auckland, NZ 46 70 12 66
Auckland, NZ 45 70 12 66
Auckland, NZ 12 70 12 66
Auckland, NZ 38 50 12 66
Gregoire and Clausen (2011) Journal paper Storrs, CT 307 102 2 51.4
Carpenter and Kaluvakolanu (2011) Journal paper Southfield, MI 325.2 101.6 6 68.25
Palla et al. (2011) Journal paper Genova, Italy 170 200 13 51.8
Genova, Italy 170 200 3 14.9
Fassman-Beck et al. (2013) Journal paper Auckland, NZ 171 100 8 72
Carson et al. (2013) Journal paper New York, NY 310 32 12 36
New York, NY 390 100–200 24 47
New York, NY 940 100 36 61
Speak et al. (2013) Journal paper Manchester, UK 408 170 11 65.7
Hakimdavar et al. (2014) Journal paper New York, NY 310 32 5 50.6
New York, NY 99 32 10 61.3
Yang et al. (2015) Journal paper Beijing, China 120 150 3 76.4
Versini et al. (2015) Journal paper Paris, France 35 30 14 17
Paris, France 35 150 14 11
Cipolla et al. (2016) Journal paper Bologna, Italy 58 100 12 51.9
Zaremba et al. (2016) Journal paper Villanova, PA 54 100 16 56–62

Table 3
Summary of studies that demonstrate the significance of ET in rain gardens through measurements.

Study Location ET performance Remarks

Volume Rate
reduction (mm/day)
(%)

Culbertson and Topeka, KS 2.7–7.8 A column study from June 2003 to November 2003.
Hutchinson (2004) Minimum rate (2.7) in fall and maximum rate (7.8) in summer
Li et al. (2009) Louisburg, 19 Volume reduction due to ET in 2004 for one lined, underdrained cell with sandy loam soil
NC
Denich and Bradford Guelph, 4.2, 7.7 An average of 4.2 mm/day over August 2008 and 7.7 mm/day over a sunny, dry period of 5 days for an
(2010) Canada unconstrained draining rain garden lysimeter with sandy soil
Wadzuk et al. (2015) Villanova, 50 3.1 Data from April to November in 2010 and 2011 in two weighing lysimeters.
PA 78 6.1 Top: Freely draining sandy soil
Bottom: Freely draining sandy soil with IWS.
Nocco et al. (2016) Madison, 21–84 1.5–9 Data from 12 replicate, free-drainage bioretention cells with three vegetation types (prairie, shrub, and turfgrass)
WI and bare soil (control) subjected to three controlled stormwater applications in July, August, and September 2006
from a 6:1 loading-ratio drainage area. Minimum and maximum volume reduction values correspond to bare soil
in August and prairie in September, respectively.
Minimum and maximum rates correspond to prairie in July and bare soil in September, respectively
Hess et al. (2017) Villanova, 47 2.9 Data from July 2013 to July 2015 in three weighing lysimeters.
PA 43 2.7 Top: Sandy loam soil with unconstricted valve underdrain.
70 4.3 Middle: Sandy soil with unconstricted valve underdrain.
Bottom: Sandy soil with IWS.
802 A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810

2005; Suleiman and Hoogenboom, 2009). Lu et al. (2005) and best fit. Nocco et al. (2016) measured ET during three stormwater appli-
DiGiovanni et al. (2013) classify the PET models into temperature cations in July, August, and September 2006 to 12 replicate, free-
based (e.g., Thornthwaite, Hamon, and Hargreaves), radiation based draining bioretention lysimeters with different vegetation types (prai-
(e.g., Turc, Makkink, and Priestley-Taylor) and composite (combina- rie, shrub, turfgrass, and bare soil) in Madison, Wisconsin and compared
tion) models (e.g., Penman, and Penman-Monteith). Carbone et al. the AET results to Hamon, Priestley-Taylor, and FAO-56 PM models. All
(2016) also categorize the ET quantification methods based on the cli- of the PET models underestimated prairie AET values. The AET values for
mate input data: (1) methods that rely on temperature measurements other vegetation types were either overestimated or underestimated by
and require some estimation of solar energy (e.g., Hargreaves, 1975 the three PET models. Hess et al. (2017) studied the observed ET values
and Thornthwaite (Wilm et al., 1944)), and (2) combination methods during 2013–2015 from three weighing lysimeter rain gardens with dif-
based on the energy balance and mass transfer principles ferent drainage conditions at Villanova, PA and concluded that both the
(e.g., Penman and the FAO-56 PM (Allen et al., 1998)). Available agricul- ASCE standardized PM and 1985 Hargreaves equations can adequately
tural literature concludes that combination methods have variability in represent the free draining (unconstructed valve underdrain) lysime-
predicting AET but, in general, have better performance than other ters but underestimate the IWS (using an upturned elbow underdrain)
methods (Katul et al., 1992; ASCE, 2005). The downside of combination lysimeter. Hess et al. (2019) continued the study of Hess et al. (2017) for
methods is the extensive required input data that are not always justifi- the period of July 2013–July 2016 (both daily and storm based) by con-
able given the improvement in their ET results compared to those of the sidering modifications for water availability (soil moisture extraction
simpler methods (Katul et al., 1992; DiGiovanni et al., 2013). A review functions) and plant health (crop coefficients). The unmodified ASCE
and discussion about different PET methods can be found in Wadzuk standardized PM provided an adequate ET estimate for all lysimeters
et al. (2013), Zhao et al. (2013), and Feng (2018). on a storm basis but not for the IWS lysimeter on a daily basis. The un-
modified 1985 Hargreaves resulted in adequate ET estimates for all ly-
3.3. Comparison of PET and AET in GSI systems simeters in both daily and storm based cases. Considering soil
moisture extraction functions and crop coefficients resulted in increas-
3.3.1. Green roofs ing prediction efficiency of both equations for all lysimeters.
Comparing the measured ET results from different green roof studies In summary, the 1985 Hargreaves method (Hargreaves and Samani,
with available PET estimation methods has been the subject of several 1985; Hargreaves and Allen, 2003) generally overpredicts the measured
studies (Table 4). DiGiovanni et al. (2013) tested two variations of the ET but has a good accuracy for longer periods (N5 days). This method is
PM equation (ASCE standardized PM equation for short reference sur- simple and only uses temperature data and global position to estimate
faces (ASCE, 2005) and Meteorological Office Rainfall and Evaporation ET (Wadzuk et al., 2013; 2015; Hess et al., 2017; 2019). Hargreaves
Calculation System (MORECS) equation (DeGaetano et al., 1994)) models (Hargreaves, 1975; Hargreaves and Samani, 1982; Hargreaves
against ET measurements from a green roof lysimeter in the Bronx, and Samani, 1985) have also shown a good agreement with ET values
New York. Both equations overpredicted the measured ET over a one- from lysimeters and the FAO-56 PM model in various agricultural stud-
year period. However, during a 30-day subset of the one-year period ies, especially for non-arid climates and not very high ET rates (e.g., b
when water was not limited, the ASCE standardized PM equation per- 9 mm/day) (e.g., Jensen et al., 1990, 1997; López-Urrea et al., 2006;
formed well in predicting measured ET. Wadzuk et al. (2013) quantified Khoob, 2008; Tabari and Talaee, 2011; Bourletsikas et al., 2018). In
the ET component of a green roof water budget in Villanova, PA using arid and semiarid climates, Hargreaves can be calibrated to improve
weighing lysimeters for three climatologically different years its performance (Heydari and Heydari, 2014; Djaman et al., 2018). The
(2009–2011) and concluded that the ASCE standardized PM tends to a 1985 Hargreaves method has been demonstrated to produce compara-
slight underestimation, but occasionally overestimates measured ET ble ET results to the FAO-56 PM method for periods greater than five
during very dry periods. The Blaney-Criddle and ASCE standardized days in irrigated sites (Hargreaves and Allen, 2003) and has been rec-
PM gave similar estimations that were closest to the measured ET in ommended as an alternative for the FAO-56 PM method when sufficient
Wadzuk et al. (2013). In another study by Marasco et al. (2015), the or reliable input data for FAO-56 PM is not available (Allen et al., 1998).
Priestley-Taylor model was found to generate the closest results to ET
measurements from two green roof sites in New York City when assum- 3.4. Stress point and available water for ET
ing uniform moisture conditions. Berretta et al. (2014) compared the
observed data from green roofs in Sheffield, UK with simulated moisture Davis et al. (2012) define the bioretention abstraction volume (BAV)
content using a water balance model and two PET methods (Hargreaves as the volume of the water captured from the surface runoff that is not
and FAO-56 PM) combined with a soil moisture extraction function. returned to the surface waters and is either evapotranspirated to the at-
Since the two PET methods did not show any significant difference, mosphere or exfiltrated into surrounding soils. While Davis et al. (2012)
they suggested the 1985 Hargreaves method (Hargreaves and Samani, consider both the surface storage and soil media pore volume in the BAV
1985) due to requiring less input data. Based on the onsite data from definition, they only consider the difference between full saturation and
an experimental extensive green roof (irrigated and covered with field capacity as the maximum pore storage in the soil media. However,
short vegetation) in Rende, Italy, Carbone et al. (2016) demonstrated plant available water (PAW) (also called available water capacity or soil
that Hargreaves (1975) and Hargreaves and Samani (1982) models pro- moisture capacity), is determined as the difference between field capac-
duce results that are in good agreement with the FAO-56 PM model and ity and permanent wilting point for planning purposes (Fassman and
hence, can be used as simple practical PET methods when the required Simcock, 2012; Fassman-Beck et al., 2016). Soil water storage metrics
input data for combination models are not available. This conclusion and their corresponding tensions (suction heads) are summarized
agrees with that of Hilten (2005) where the author found no statistical from Fassman and Simcock (2012) as Fig. 1. As an example, the average
difference between the monthly ET estimates from FAO-56 PM and values of full saturation, field capacity and wilting point for three differ-
1985 Hargreaves methods in a green roof site in Athens, GA. ent texture classes (sand, sandy loam, and loamy sand) based on Saxton
and Rawls (2006) are presented in Table 5. The free water is the avail-
3.3.2. Rain gardens able moisture between full saturation and field capacity that is drained
There are limited studies comparing measured ET data from rain by gravity. The readily available water is the moisture between field ca-
gardens with PET estimation methods available in the literature pacity and stress point, which is a portion of PAW. The remaining part of
(Table 4). Wadzuk et al. (2015) measured ET from bioretention lysime- PAW is called stress water, which is the moisture between stress point
ters in Villanova, PA and found no equation to be an exact fit for the and permanent wilting point. Fassman and Simcock (2012) consider
measured data, although the 1985 Hargreaves model produced the the maximum potential available water for ET as readily available
Table 4
Comparison of the measured ET results from different green roof and rain garden studies with available PET estimation methods. ‘+’ denotes overprediction, ‘−’ underprediction, and ‘*’ closest to the observed ET values.

Temperature-based methods Radiation-based methods Combination methods Remarks

Study Location Blaney-Criddle Hamon Hargreaves Hargreaves 1985 Slatyer-McIlroy Priestley-Taylor Penman FAO-56 ASCE MORECS
(1975) and Samani Hargreaves PM Standardized
(1982) PM

Green roofs

A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810


Hilten Athens, N/A N/A N/A N/A * N/A N/A N/A * N/A N/A FAO-56 PM was used as the reference ET
(2005) GA
Wadzuk Villanova, −/+ N/A N/A N/A +1 – – N/A N/A ‐/+2 N/A 1
Better accuracy for longer periods (N5 days)
2
et al. PA * * Slightly underestimation, but occasionally
(2013) overestimation during very dry periods
DiGiovanni Bronx, NY N/A N/A N/A N/A N/A N/A N/A N/A N/A * + For non-water-limiting conditions
et al.
(2013)
Berretta Sheffield, N/A N/A N/A N/A * N/A N/A N/A * N/A N/A Hargreaves was suggested because of requiring less input
et al. UK data
(2014)
Marasco New York N/A N/A N/A N/A + N/A * + N/A + N/A Assuming uniform moisture conditions
et al. City, NY
(2015)
Carbone Rende, N/A N/A * * N/A N/A N/A N/A N/A * N/A ASCE standardized PM was used as the reference ET
et al. Italy
(2016)

Rain gardens
Wadzuk Villanova, -1 N/A N/A N/A + – – N/A N/A -1 N/A 1
Underestimation during summer and good performance
et al. PA *2 in low ET rates
2
(2015) Overestimation but better than other methods
Nocco et al. Madison, N/A −/+ N/A N/A N/A N/A −/+ N/A −/+ N/A N/A Variations based on vegetation type (prairie, shrub, and
(2016) WI turfgrass, and bare soil) and time of the year (July, August,
September) in 12 replicate, free-drainage bioretention
lysimeters
Hess et al. Villanova, N/A N/A N/A N/A *1 N/A N/A N/A N/A *1 N/A 1
For freely draining lysimeters
(2017) PA −2 −2 2
For IWS lysimeters
Hess et al. Villanova, N/A N/A N/A N/A *1 N/A N/A N/A N/A *2 N/A 1
Overestimation and underestimation in unconstructed
(2019) PA outflow and IWS lysimeters, respectively, but ultimately
adequate results
2
Underestimation in IWS lysimeters but ultimately
adequate results

803
804 A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810

Fig. 1. Soil water storage metrics and their corresponding tensions (not to scale).

water plus stress water (i.e., PAW). However, the actual available water dependent, and relies on the available soil water. Once soil water
for ET is greater than readily available water and less than PAW. reaches the permanent wilting point, no ET can happen (Allen et al.,
Modelers may make some simplifying assumptions to generate a 1998; Wadzuk et al., 2013), which must be considered in the design
parsimonious model. For example, the total water available for ET may time horizon.
be taken as PAW; however, Allen et al. (1998) discussed water stress
as a linear reduction in ET after the soil reaches a threshold between
field capacity and wilting point. Hence, ET decreases after this threshold 4.1. Static methods
and discontinues at the wilting point (Allen et al., 1998). Szota et al.
(2017) supported the presence of this stress threshold, based on a sim- Design and crediting of rain gardens are often based on static
ulated rainfall experiment over August to November 2011 on eighteen stormwater volume storages in the GSI system. For example, the design
green roof modules in Melbourne, Australia, where they found that guideline in Pennsylvania, which is similar to that in other states
laboratory-measured PAWs may overestimate ET and rainfall retention (e.g., Virginia (VADCR, 2011) and Maryland (MDESD, 2007)), only ac-
potential in green roofs. Hence, Szota et al. (2017) suggested those PAW counts for surface storage volume and soil media void space volume
values should be adjusted down to conform to real conditions. This find- during a rain event (PA DEP, 2006). This static approach considers the
ing agrees with Kasmin et al. (2010) where they state that some of the porosity values of soil media (sandy loam or loamy sand), sand, and
moisture between field capacity and permanent wilting point will not stone as 0.2, 0.3, and 0.4, respectively, and does not consider ET in vol-
be accessible by plant roots. A reduction in the ET rate after approxi- ume reduction calculations (PWD, 2015, 2018).
mately 50% of the PAW is lost has been reported by Rezaei et al. The difference between the average values of full saturation and
(2005), Berghage et al. (2009), and Voyde et al. (2010b). This threshold field capacity (which is assumed to be the available water for infiltration
might be corresponding to the stress point (associated moisture with or draining due to gravity) is calculated as 27% (i.e., 45.0%–17.9%) for
100 kPa tension), as Fassman and Simcock (2012) state that ET will be sandy loam soils using Saxton and Rawls (2006). The calculated value
reduced at tensions below the stress point because the remaining mois- (27%) is greater than the credited void space for rain garden soil
ture is held more tightly to the soil matrix in smaller pores. media (i.e., 20%). Also, according to Saxton and Rawls (2006), the differ-
ence between average values of field capacity and wilting point (which
4. ET and GSI design is assumed to be the maximum potential water for ET) is about 10%
(i.e., 17.9–8.1%) for the sandy loam soil. No portion of this potential stor-
Incorporation of ET in a GSI design depends on the design objective age is credited for ET in the static design guidelines. For loamy sand (an-
and the region. Evapotranspiration can be an important element in the other common planting soil texture), the difference between full
water budget of vegetated GSI systems and needs to be considered in saturation and field capacity is 34%, which is greater than the 20%
GSI design when the objective is annual runoff volume reduction credited void space, and a 6% difference between field capacity and
(e.g., for CSO control). Care must be taken when considering ET in de- wilting point that is not credited to ET. Similarly, the available water
sign that contradicts the design objective (e.g., groundwater recharge for infiltration and the maximum potential water for ET in sandy soils
and stormwater reuse), is not conservative (e.g., flood reduction), or are calculated as 37% and 5%, respectively, sum of which is greater
the design is in an area where ET loss is minimal (e.g., in a region with than the 30% credited void space. To assess the static design and
very frequent rainfall). Additionally, ET is both water- and energy- crediting approach, the actual available water for ET as void space was
considered as 50% of the PAW (i.e., ½ (field capacity – wilting point))
(Allen et al., 1998) and the total creditable void storage summed free
Table 5 water and 50% PAW using the values from Table 5. The creditable void
Soil water content values in percent (vol/vol) for different soil texture classes.
storage values were found as 39.1%, 36.8%, and 32.0% for sand, loamy
Metric Tension (kPa) Sand Loamy Sand Sandy Loam sand, and sandy loam, respectively, all of which are greater than the
Full saturation 0 46.3 45.7 45.0 aforementioned credited void storage. This indicates that the current
Field capacity 10–33 9.4 12.1 17.9 static approach is overly conservative. Increasing the credited void stor-
Stress point 100 age (e.g., for sandy loam from 20% to 30%) will yield more physically
Wilting point 1500 5.0 5.7 8.1 based design of rain gardens. This example assumes the actual available
Oven dry Very high 0.0 0.0 0.0
water for ET is 50% PAW. Calculating the portion of PAW that
A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810 805

corresponds to the measured ET values in vegetated GSI systems is rec- (5 mm/day) multiplied by a factor of 0.5 (equivalent to 2.5 mm/day,
ommended as future research. which is close to the average measured ET rates in the literature) over
a basin's drawdown time (time between rain events).
4.2. Dynamic methods To take advantage of accumulated ET through the GSI recovery pe-
riod, Hess (2017) looked at the balance between ET and infiltration in
Immediately after a storm event, the stormwater retained in the GSI three rain garden lysimeters with a media depth of 66 cm in Villanova,
soil (with a restricted or no underdrain) is lost primarily to infiltration PA: a sandy loam and a sandy media with an unconstricted outflow
and some ET. Evapotranspiration continues to reduce the soil water dur- (UO) valve and a sandy media with IWS (Hess et al., 2014, 2017).
ing the interevent dry time and restore the soil storage capacity for the Hess (2017) developed charts that estimate cumulative ET in the rain
next storm event. Although ET is generally greater after a rainfall event gardens during the dry period after large storm events (25 mm or
than during extended dry periods, it continues as long as water is avail- greater). This method could be applied in other regions and climate
able (Wadzuk et al., 2015; Nocco et al., 2016). This is a key point that has zones to develop charts that can be used as a simple tool to incorporate
been usually neglected in crediting and continuous modeling of GSI sys- dynamic losses from ET into static design. Fig. 2 demonstrates a concep-
tems (Wadzuk and Traver, 2012; Wadzuk et al., 2013; Nocco et al., tual example of these charts that allow the user to determine the cumu-
2016). To incorporate ET in the design and crediting of vegetated GSI lative amount of water loss due to ET from a rain garden up to eleven
systems, methods are needed that incorporate both ET and infiltration days after a storm event. For example, using Fig. 2 for the sandy loam
dynamically into the design process (Berland et al., 2017). Two UO, sand UO, and sand IWS systems would result in an approximate
methods, i.e., complex and simple methods, are reviewed. water loss of 25.2, 24.6, and 43.0 mm, respectively, due to ET over
6 days (the typical time between storm events in Philadelphia, PA-
4.2.1. Complex method Driscoll et al., 1989). The amount of ET from the simple tool may be con-
Evapotranspiration is a continuous process and does not easily fit sidered as an additional porosity to the current static porosity (e.g., 20%
into the existing static design of vegetated GSI systems. Evapotranspira- void space for engineered soil media- PWD, 2015, 2018) in storage vol-
tion rates vary by climate conditions over the year and considering this ume calculations. Considering this additional porosity for a bioretention
variation can result in a more realistic performance analysis that is not system with 66 cm media depth will yield a credited storage of 23.8%,
overly conservative (Zaremba et al., 2016; Eger et al., 2017). Available 23.7%, and 26.5% for sandy loam UO, sand UO, and sand IWS, respec-
literature demonstrates the dynamic role of ET in the hydrology of tively. Note that the maximum potential water for ET (i.e., the difference
green roofs and rain gardens and provide insights to incorporate the dy- between average values of field capacity and wilting point) for sandy
namics of ET in their design and crediting (Li et al., 2009; Denich and loam and sand was calculated based on Saxton and Rawls (2006) as
Bradford, 2010; Feller et al., 2010; Hickman et al., 2010, 2011; 10% and 5%, respectively. In this example, the actual values of ET
Schneider et al., 2011; Wadzuk and Traver, 2012; Wadzuk et al., 2013; (from Fig. 2) are less than the presented values of maximum potential
Hess et al., 2015a, 2015b; Zaremba et al., 2015a, 2015b; Zaremba et al., water for ET.
2016; Hess, 2017; Hess et al., 2017). Infiltration and ET are both dy- It should be noted that the chart in Fig. 2 was developed based on the
namic processes that should be considered simultaneously in the design specific conditions of the experimental setup (media depth, soil type,
of vegetated GSI systems (Wadzuk et al., 2013). drainage condition, and plant type), length of monitoring period or
The complex method that incorporates ET in GSI design is a contin- number of observed data corresponding to large events, and meteoro-
uous simulation approach that considers the dynamic aspects of infiltra- logical conditions of the study area (Villanova, PA), and is applicable
tion and ET by using long-term records of rainfall and climate data to systems with a similar climate and rain garden configurations to
(Traver and Ebrahimian, 2017). This approach is currently used in a Hess (2017). A similar approach can be adopted to develop simple
few places in Australia, New Zealand, and Western US by hydrologic methods (charts) for other climate zones and rain garden configuration
models such as MUSIC and SWMM (DPLG, 2009; Melbourne Water, and characteristics where annual volume reduction is the design objec-
2018; Lewis et al., 2015; WSDE, 2012). Moving towards continuous sim- tive. Another study, which is a bench scale study using non-weighing ly-
ulation is emphasized for areas that currently do not use this approach simeters, was performed by Hess et al. (2015b) and DelVecchio et al.
for GSI design. The complex method provides a robust design that cou- (2019) looking at two more media depths (46 and 20 cm) and four ad-
ples the soil matrix and hydrologic aspects through a continuous simu- ditional media types (loamy sand, loam, silt loam, clay loam) that pro-
lation of ET and exfiltration (i.e., infiltration from soil media to the vided additional information for expanding the simple method to
surrounding in-situ soil in infiltration-based rain gardens). Evapotrans-
piration is dependent on available soil water; hence, continuous simula-
tions with a water accounting system would provide results closest to
reality. Based on the available knowledge, the 1985 Hargreaves method
(Hargreaves and Samani, 1985) is recommended for continuous simula-
tions. Further research is needed to modify the ET predictive methods to
better represent the observed ET data in vegetated GSI systems
(Wadzuk et al., 2013, 2015; Eger et al., 2017).

4.2.2. Simple method


Traditional methods for the design and crediting of GSI systems are
based on design storm events analyzed over a narrow time window.
Also, ET is not accounted for because of perceived unimportance. How-
ever, identifying how ET contributes to the volume reduction between
storm events can improve the event-based designs and is recognized
as a research need (Berland et al., 2017). For example, Minnesota
stormwater manual (MPCA, 2017) uses an event-based method to cal-
culate the volume credit due to ET by considering the ET volume as
the minimum of the measured ET and the amount of water stored be- Fig. 2. Conceptual example of cumulative ET (as water depth) versus days after large rain
tween field capacity and wilting point. MPCA (2017) considers the mea- events (25 mm or greater) for sandy loam and sandy soils (based on data from Hess,
sured ET as the average daily pan evaporation rate of 0.2 in./day 2017).
806 A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810

different soil media and depths. Typical (e.g., average) length of the dry of increased groundwater drainage from rain gardens due to shallow
periods between storm events should be determined by statistical anal- groundwater tables (Zhang et al., 2018b). In a study by Li et al. (2009),
ysis of long-term rainfall records in the study area. Such a study is per- a bioretention cell with an IWS zone produced outflow during 37% of
formed by Driscoll et al. (1989) for 160 locations throughout the US. observed events, while a neighboring identical cell with no IWS zone
Two years of monitoring data, including ten storm events equal to or generated outflow for 65% of events. The exfiltration rate is reported
N25 mm, were used for developing the charts in Fig. 2. Long-term mon- to be dependent on the surrounding in-situ soil hydraulic conductivity.
itoring is needed to increase the number of large events in the observed While highly conductive surrounding soils can greatly increase the
dataset. If large storm events (flood volume reduction) is of concern, a exfiltration rate, in sites with low infiltration capacity in-situ soils, the
study of back-to-back storm events needs to be performed to evaluate water is more restricted within the bioretention media and fills the
the appropriateness of accounting for ET in the design. Such a study void space (Davis et al., 2012). However, Winston et al. (2016) showed
has been done by Wadzuk et al. (2017) that shows a very low likelihood that the inclusion of IWS in three rain gardens constructed in low per-
for back-to-back storm events that impact GSI recovery performance meability soils (clay, hydrologic soil group D) in Ohio resulted in a sub-
between events in the study area for Fig. 2. stantial reduction in runoff volume. Wadzuk et al. (2015) quantified ET
in two weighing lysimeters in Villanova, PA mimicking two different
4.3. Technical recommendations drainage conditions in bioretention systems: freely draining (low stor-
age capacity) and lined system with IWS (high storage capacity). Al-
Technical recommendations presented and discussed herein are though the IWS system had a sandier soil with a lower water storage
based on the available literature and encompass the effects of two im- capacity, it showed more ET in comparison to the freely draining system
portant items, drainage conditions and soil media, on the amount of (78% of the direct rain falling on the system versus 50%). Also, the mea-
ET in green roofs and rain gardens. sured daily average ET during the study period (April–November in
2010–2011) in the IWS system was found to be double that of the freely
4.3.1. Drainage condition draining system (i.e., 6.1 mm/day versus 3.1 mm/day). Hess et al.
Zaremba et al. (2016) compared an undrained green roof (i.e., no (2017) monitored three rain garden lysimeters with two different
drainage system) to a drained green roof under the same microclimate drainage structures: underdrain with unconstructed valve (freely
and rainfall conditions in Villanova, PA. While the drained green roof draining) and IWS (with upturned elbow underdrain) in Villanova,
evapotranspirated 62% and 56% of the total rainfall in 2013 and 2014 PA. The average ET rate and volume reduction due to ET during the en-
(April to November), respectively, with an average ET rate of tire monitoring period (July 2013 to July 2015) for the IWS system with
2.1 mm/day, the undrained system had greater ET performance with sandy soil media (4.3 mm/day and 70%, respectively) were greater than
an annual ET range of 68% to 88% and an average ET rate of those of the free draining systems with sandy loam (2.9 mm/day and
3.1 mm/day during 2009–2014 (April to November). The different 47%) and sandy media (2.7 mm/d and 43%). These results are attributed
drainage configurations determined the available water for ET. The un- to the prolonged water storage in an IWS system enabling more ET than
drained green roof had more stored water and resulted in more ET. a freely draining system. Therefore, one recommendation for rain gar-
Availability of sufficient dry time between storm events in the study den design is modifying the outlet structure (e.g., underdrains equipped
area allowed ET to restore the storage capacity of the green roof with an end cap) or including an IWS to encourage extended water stor-
media and, hence, system performance was not compromised in the un- age and increase the role of ET where appropriate.
drained green roof. The other advantage of the undrained system in
Zaremba et al. (2016) was the increased storage capacity, which re- 4.3.2. Native (in-situ) soils in the media
sulted in less outflow. The undrained system could become fully satu- Native (in-situ) soils often have a higher percentage fines and
rated and only had measurable outflow from rainfall events N35 mm. greater water-holding capacity than the typically sandier engineered
The drained green roof could only reach field capacity and generated soils designed for expedited infiltration. One recommendation is to
outflow for rainfall events N13.5 mm, similar to studies reviewed in use native (in-situ) soils for the rain gardens' media design, where
Fassman-Beck et al. (2016). One recommendation for green roof design they are finer than engineered soils, have sufficient infiltration rates,
is to consider the need for an underdrain. Given adequate structural ca- and are not contaminated (Davis et al., 2009; Lee et al., 2016). Utilizing
pacity, omitting underdrains in extensive green roofs or adding a valve fine-grained native (in-situ) soil encourages more ET (Hess et al., 2014)
to regulate outflow based on the weather conditions can be investigated and will result in improved pollutant removal in some cases (Hsieh and
to enhance ET. Alternatively, the roof drain could be slightly raised to Davis, 2005; Sickles et al., 2007), as well as cost savings and environ-
encourage more storage within the drainage layer, or other mechanisms mental protection due to reducing the need for soil extraction and
could be implemented to reduce conveyance to the roof drain (Zaremba transportation to the site (Flynn and Traver, 2013). However, care
et al., 2016). This approach may require more maintenance needs and must be taken when using in-situ soils for the rain garden media as
considerations for freezing conditions but would substantially increase they may include large amounts of dead dirt or ash (i.e., with few living
annual volume reduction. Moreover, it could minimize the watering organisms). The rain garden location is important in the soil mix design.
needs except for areas with long dry weather periods. Higher fines content will result in more water retention in the media
Similar to green roofs, design modifications, such as excluding the and may increase the risk of flooding, which is a major concern in
underdrain (bioinfiltration) or including an IWS, could be made to some highly-populated urban areas. It is suggested that the in-situ soil
rain gardens to encourage controlled (delayed) release of stormwater is evaluated first to see if it is a living soil (i.e., with adequate amount
to the drainage system and increase ET from the system (Brown and of living organisms, organic matter, minerals, and water). Then, the
Hunt, 2011a; Davis et al., 2012; Winston et al., 2016). An IWS can create soil texture should be evaluated using particle size distribution of the
a semi-permanent saturated zone at the bottom of the rain garden by soil. Compaction and plasticity of the soil should also be evaluated as
adding an upturned elbow underdrain pipe or elevated outlet. Capping they are limiting factors for infiltration rates (Pitt et al., 2008;
the underdrain pipe end and making an orifice by drilling the cap when Carpenter and Hallam, 2010; Olson et al., 2013).
needed is another way of providing IWS (PWD, 2018). The IWS provides
increased hydrologic performance via less outflow, prolonged water 5. Research needs
storage for plants to draw on during dryer periods, increased runoff vol-
ume reduction via ET, as well as water quality benefits (Davis et al., With the purpose of incorporating ET in the design and crediting of
2009; Brown and Hunt, 2011a; Wadzuk et al., 2015; Hess et al., 2017). rain gardens and green roofs, future research directions are identified
Excluding or elevating the underdrain may also address the problem as following:
A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810 807

5.1. Continuous simulation of ET and exfiltration in GSI 5.7. Soil media design for ET-infiltration balance

Continuous simulation models often neglect exfiltration during Further research is needed to determine the best soil mix design
storm events and make empirical assumptions about how the soil mois- looking at the balance between ET and infiltration in different areas as
ture declines and the infiltration capacity recovers over time between well as the balance between runoff reduction and pollutant removal
events with little connection to the physics of ET (Traver and (Davis et al., 2009; Liu et al., 2014). A comprehensive cost-benefit
Ebrahimian, 2017). Because of the importance of rainfall characteristics study and risk analysis is suggested to investigate the benefits, limita-
in runoff retention, rainfall distribution needs to be considered in eval- tions, issues and concerns regarding using in-situ (native) soils as the
uating GSI performance. The effects of climate change on rainfall distri- soil media in GSI systems.
bution is another issue that makes the problem more complex. Further
research is needed for the continuous simulation of ET and infiltration 6. Conclusions
processes and exploring the balance between them in a GSI system.
A literature review was performed on the role of ET in green roofs
5.2. Simple tools to account for ET in GSI design and rain gardens. The results indicate that:
– Evapotranspiration is a viable runoff reduction mechanism in the
Simple tools are needed to explore the interrelation between ET, hydrologic cycle of vegetated green stormwater infrastructure (GSI).
exfiltration, and pore storage in GSI systems and to incorporate this re- – The annual volume retention in green roofs (mostly due to ET) can
lationship in the design and crediting of those systems. Pilot scale stud- range from 11% to 77% of the total rainfall volume (with a median
ies such as Denich and Bradford (2010), Wadzuk et al. (2015), Hess et al. of 57%) in different studies depending on meteorological conditions
(2017), and DelVecchio et al. (2019) need to be expanded to consider and green roof properties (Table 2).
different soil and plant types as well as media depths in different cli- – The annual volume retention due to ET in rain gardens depends on
mates to cover the whole range of rain garden and green roof systems the GSI configuration, design, and hydraulic loading ratio in each
in practice. rain garden, but is still considerable in the overall water balance
(e.g., 19% in Li et al., 2009 that is for an actual rain garden not a lysim-
5.3. ET predictive methods in GSI eter or column study where ET portions in the annual water balance
tend to be higher).
Currently, the most common predictive models for ET do not prop- – The ET rates in green roofs and rain gardens in different climates
erly fit the measured ET rates in vegetated GSI. More research is needed and experimental setups are found to vary between 1 and
to find out modification factors such as soil moisture extraction func- 10 mm/day; however, the average ET rate can be assumed about
tions, which are useful in determining a proper predictive method. 2–3 mm/day in climates similar to southeastern Pennsylvania
(i.e., between Cfa and Dfa climate zones of the Koppen-Geiger
5.4. Stress point and PAW in GSI classification system).
– Modifying the drainage structure and using fine-grained in-situ
Further research is needed to calculate the portion of PAW that cor- soils can potentially improve ET in vegetated GSI systems.
responds to the measured ET values in vegetated GSI systems. By using
Evapotranspiration is a dynamic process that is highly dependent
laboratory-derived soil moisture characteristics curve for the GSI soil
on available water in the soil media; hence, ET and infiltration should
media, it can be examined to see how close the measured ET with a cer-
be considered simultaneously in the design of vegetated GSI systems.
tain plant type is to the readily available water (field capacity minus
Continuous modeling of ET and infiltration, which couples the soil
stress point) and whether the measured ET is close to 50% PAW in dif-
matrix and hydrologic aspects using rainfall and climate data,
ferent soil textures. This will provide a reasonable estimate of the por-
would provide the most appropriate design. None of the available
tion of PAW that corresponds to actual ET and can be added to the
methods (equations) for estimating ET are an exact fit for observed
credited pore space, when designing and crediting GSI systems are
ET data. Based on the available knowledge, the 1985 Hargreaves
based on a static surface storage plus void space volume in the soil
method (Hargreaves and Samani, 1985), due to its simplicity, rea-
media.
sonable results without modifications, and ease of access to the re-
quired input data, is recommended when performing continuous
5.5. Long term monitoring of GSI simulations.
Two methods, simple and complex, for incorporating ET in the de-
The varying performance of GSI systems over the design life due to sign and crediting of vegetated GSI systems were reviewed. The simple
factors such as clogging and compaction indicates a need for long method (empirical, porosity-based approach based on certain soil,
term monitoring of those systems. More on-site data collection is neces- plant, and climate conditions that can be reproduced for other condi-
sary to fully evaluate the dynamic performance of the ET equations over tions) requires little input information but gives conservative volume
different seasons and in different locations. Also, more observed data reductions compared to the complex method (continuous simulation)
through long term monitoring programs from different green roof and that requires more input information and provides more realistic vol-
rain garden systems is needed to verify the continuous simulations. ume reductions.
Long term measurements (e.g., 5 years or more- Eger et al., 2017) are
recommended to investigate the discussed issues and ideas in this Acknowledgements
paper. Further, long term studies will help define a GSI' design life.
The authors would like to acknowledge Philadelphia Water De-
5.6. Drainage conditions for improved ET and other benefits partment, Philadelphia, Pennsylvania, USA Pennsylvania Depart-
ment of Environmental Protection, USA Villanova Center for
Further investigations are needed to improve the performance of the Resilient Water Systems, Villanova, Pennsylvania, USA and partners
green roofs and rain gardens by allowing more ET through the modifica- of Villanova Urban Stormwater Partnership, Villanova, Pennsylvania,
tion of drainage systems. Using automated controls on a GSI drainage USA for funding this study. The opinions expressed in this paper are
structure could provide another mechanism to manipulate the stored those of the authors and do not necessarily represent the funding or-
water in the soil and make it available for ET. ganizations' views.
808 A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810

References Driscoll, E.D., Palhegyi, G.E., Strecker, E.W., Shelley, P.E., 1989. Analysis of Storm Event
Characteristics for Selected Rainfall Gages throughout the United States.
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration-guidelines for Woodward-Clyde Consultants, Oakland, CA. Prepared for the. US Environmental Pro-
computing crop water requirements. FAO Irrigation and Drainage Paper No. 56. tection Agency Available at. https://nepis.epa.gov.
FAO, Rome, Italy. Ebrahimian, A., Wilson, B.N., Gulliver, J.S., 2016a. Improved methods to estimate the effec-
ASCE, 2005. The ASCE Standardized Reference Evapotranspiration Equation, Task Com- tive impervious area in urban catchments using rainfall-runoff data. J. Hydrol. 536,
mittee on Standardization of Reference Evapotranspiration of the Environmental 109–118.
and Water Resources Institute. ASCE, Reston, VA. Ebrahimian, A., Gulliver, J.S., Wilson, B.N., 2016b. Effective impervious area for runoff in
Bengtsson, L., 2005. Peak flows from thin sedum-moss roof. Nord. Hydrol. 36, 269–280. urban watersheds. Hydrol. Process. 30 (20), 3717–3729.
Berghage, R., Beattie, D., Jarrett, A., Thuring, C., Razaei, F., O'Connor, T., 2009. Green Roofs Ebrahimian, A., Gulliver, J.S., Wilson, B.N., 2018. Estimating effective impervious area in
for Stormwater Runoff Control. Rep. No. EPA/600/R-609/026. Environmental Protec- urban watersheds using land cover, soil character and asymptotic curve number.
tion Agency, Cincinnati. Hydrol. Sci. J. 63 (4), 513–526.
Berkompas, B., Marx, K.W., Wachter, H.M., Beyerlein, D., Spencer, B., 2008. A Study of Eger, C.G., Chandler, D.G., Driscoll, C.T., 2017. Hydrologic processes that govern
Green Roof Hydrologic Performance in the Cascadia Region. 2008 Low Impact Devel- stormwater infrastructure behaviour. Hydrol. Process. 31 (25), 4492–4506.
opment Conference. ASCE, pp. 1–10. Fassman, E., Simcock, R., 2012. Moisture measurements as performance criteria for exten-
Berland, A., Shiflett, S.A., Shuster, W.D., Garmestani, A.S., Goddard, H.C., Herrmann, D.L., sive living roof substrates. J. Environ. Eng. 138 (8), 841–851.
Hopton, M.E., 2017. The role of trees in urban stormwater management. Landsc. Fassman-Beck, E., Voyde, E., Simcock, R., Hong, Y.S., 2013. 4 living roofs in 3 locations:
Urban Plan. 162, 167–177. does configuration affect runoff mitigation? J. Hydrol. 490, 11–20.
Berndtsson, J.C., 2010. Green roof performance towards management of runoff water Fassman-Beck, E., Hunt, W., Berghage, R., Carpenter, D., Kurtz, T., Stovin, V., Wadzuk, B.,
quantity and quality: a review. Ecol. Eng. 36 (4), 351–360. 2016. Curve number and runoff coefficients for extensive living roofs. J. Hydrol.
Berretta, C., Poë, S., Stovin, V., 2014. Moisture content behaviour in extensive green roofs Eng. 21 (3), 04015073.
during dry periods: the influence of vegetation and substrate characteristics. Feller, M., Traver, R., Wadzuk, B., 2010. Estimation of green roof evapotranspiration: ex-
J. Hydrol. 516 (4), 37–49. perimental results. Proc. of 2010 International Low Impact Development Conference:
Bliss, D.J., Neufeld, R.D., Ries, R.J., 2009. Storm water runoff mitigation using a green roof. Redefining Water in the City, San Francisco, CA, pp. 11–14.
Environ. Eng. Sci. 26 (2), 407–418. Feng, Y., 2018. Evapotranspiration from green infrastructure: benefit, measurement, and
Bourletsikas, A., Argyrokastritis, I., Proutsos, N., 2018. Comparative evaluation of 24 refer- simulation. In: Bucur, Daniel (Ed.), Advanced Evapotranspiration Methods and Appli-
ence evapotranspiration equations applied on an evergreen-broadleaved forest. cations. IntechOpen https://doi.org/10.5772/intechopen.80910.
Hydrol. Res. 49 (4), 1028–1041. Feng, Y., Burian, S.J., Pardyjak, E.R., 2018. Observation and estimation of evapotranspira-
Brown, R.A., Hunt, W.F., 2011a. Underdrain configuration to enhance bioretention tion from an irrigated green roof in a rain-scarce environment. Water 10 (3), 262.
exfiltration to reduce pollutant loads. J. Environ. Eng. 137 (11), 1082–1091. Flynn, K., Traver, R., 2013. Green infrastructure life cycle assessment: a bio-infiltration
Brown, R.A., Hunt, W.F., 2011b. Impacts of media depth on effluent water quality and hydro- case study. J. Ecol. Eng. 55, 9–22.
logic performance of undersized bioretention cells. J. Irrig. Drain. Eng. 137 (3), 132–143. Gregoire, B.G., Clausen, J.C., 2011. Effect of a modular extensive green roof on stormwater
Brown, R.A., Hunt, W.F., Skaggs, W.R., 2011. Long-term Modeling of Bioretention Hydrol- runoff and water quality. Ecol. Eng. 37 (6), 963–969.
ogy with DRAINMOD. Water Resources Research Institute of the University of North Grimmond, C.S.B., Oke, T.R., 1986. Urban water balance: 2. Results from a suburb of Van-
Carolina, Raleigh, NC Report No. 415. https://repository.lib.ncsu.edu/handle/1840.4/ couver, British Columbia. Water Resour. Res. 22 (10), 1404–1412.
8155. Grimmond, C.S.B., Oke, T.R., 1991. An evapotranspiration-interception model for urban
Carbone, M., Principato, F., Garofalo, G., Piro, P., 2016. Comparison of evapotranspiration areas. Water Resour. Res. 27 (7), 1739–1755.
computation by FAO-56 and Hargreaves methods. J. Irrig. Drain. Eng. 142 (8), Grimmond, C.S.B., Oke, T.R., 1999. Evapotranspiration rates in urban areas. Impacts of
06016007. urban growth on surface water and groundwater quality, proceedings of lUGG 99
Carpenter, D.D., Hallam, L., 2010. Influence of planting soil mix characteristics on symposium. IAHS Publication No 259, 235–244.
bioretention cell design and performance. J. Hydrol. Eng. 15 (6), 404–416. Hakimdavar, R., Culligan, P.J., Finazzi, M., Barontini, S., Ranzi, R., 2014. Scale dynamics of
Carpenter, D.D., Kaluvakolanu, P., 2011. Effect of roof surface type on storm-water runoff extensive green roofs: quantifying the effect of drainage area and rainfall characteris-
from full-scale roofs in a temperate climate. J. Irrig. Drain. Eng. 137 (3), 161–169. tics on observed and modeled green roof hydrologic performance. Ecol. Eng. 73,
Carson, T.B., Marasco, D.E., Culligan, P.J., McGillis, W.R., 2013. Hydrological performance of 494–508.
extensive green roofs in New York City: observations and multi-year modeling of Hamel, P., Mchugh, I., Coutts, A., Daly, E., Beringer, J., Fletcher, T.D., 2014. Automated
three full-scale systems. Environ. Res. Lett. 8 (2), 024036. chamber system to measure field evapotranspiration rates. J. Hydrol. Eng. 20 (2),
Carter, T.L., Rasmussen, T.C., 2006. Hydrologic behavior of vegetated roofs. J. Am. Water 04014037.
Resour. Assoc. 42 (5), 1261–1274. Hargreaves, G.H., 1975. Moisture availability and crop production. Trans. ASAE 18 (5),
Cipolla, S.S., Maglionico, M., Stojkov, I., 2016. A long-term hydrological modelling of an ex- 980–0984.
tensive green roof by means of SWMM. Ecol. Eng. 95, 876–887. Hargreaves, G.H., Allen, R.G., 2003. History and evaluation of Hargreaves evapotranspira-
Connelly, M., Liu, K., & Schaub, J. (2006). BCIT Green Roof Research Program, Phase 1 Sum- tion equation. J. Irrig. Drain. Eng. 129 (1), 53–63.
mary of Data Analysis: Observation Period – Jan. 1, 2005 to Dec. 31, 2005. Canada Hargreaves, G.H., Samani, Z.A., 1982. Estimating potential evapotranspiration. J. Irrig.
Mortgage and Housing Corporation. Ottawa, Ontario, Canada. http://publications.gc. Drain. Div. 108 (3), 225–230.
ca/site/eng/389231/publication.html. Hargreaves, G.H., Samani, Z.A., 1985. Reference crop evapotranspiration from tempera-
Culbertson, T.L., Hutchinson, S.L., 2004. Assessing bioretention cell function in a Midwest ture. Appl. Eng. Agric. 1 (2), 96–99.
continental climate. SCGR Dynamic Partnership for an Environmentally Safe Healthy Harper, G.E., Limmer, M.A., Showalter, W.E., Burken, J.G., 2015. Nine-month evaluation of
World, ASAE Annual Int. Meeting, Ottawa, pp. 7814–7852. runoff quality and quantity from an experiential green roof in Missouri, USA. Ecol.
Davis, A.P., Hunt, W.F., Traver, R.G., Clar, M.E., 2009. Bioretention technology: an overview Eng. 78, 127–133.
of current practice and future needs. J. Environ. Eng. 135 (3), 109–117. Hathaway, A.M., Hunt, W.F., Jennings, G.D., 2008. A field study of green roof hydrologic
Davis, A.P., Traver, R.G., Hunt, W.F., Lee, R., Brown, R.A., Olszewski, J.M., 2012. Hydrologic and water quality performance. Trans. ASABE 51 (1), 37–44.
performance of bioretention storm-water control measures. J. Hydrol. Eng. 17 (5), Hess, A., 2017. Rain Garden Evapotranspiration Accounting. Ph.D. Dissertation. Villanova
604–614. University, Villanova, Pennsylvania.
DeGaetano, A.T., Eggleston, K.L., Knapp, W.W., 1994. Daily Evapotranspiration and Soil Hess, A., Wadzuk, B.M., Traver, R.G., 2014. Construction and soil monitoring plan for an
Moisture Estimates for the Northeast United States. Northeast Regional Climate Cen- ET-infiltration rain garden comparison site. World Environmental and Water Re-
ter Research Series Publication Number RR941. Cornell University, Ithaca, NY. sources Congress 2014: Water without Borders. World Environmental and Water Re-
DelVecchio, T., Welker, A., Wadzuk, B., 2019. Balance of infiltration and evapotranspira- sources Congress. Proceedings vol. 2014, pp. 105–114.
tion for different soil types: volume removal. J. Sustain. Water Built Environ. (in Hess, A., Wadzuk, B., Welker, A., 2015a. Evapotranspiration and infiltration in rain garden
press). systems. In World Environmental and Water Resources Congress: Floods, Droughts,
DeNardo, J.C., Jarrett, A.R., Manbeck, H.B., Beattie, D.J., Berghage, R.D., 2005. Stormwater and Ecosystems 261–270.
mitigation and surface temperature reduction by green roofs. Trans. ASAE 48 (4), Hess, A., DelVecchio, T., Welker, A., Wadzuk, B.M., 2015b. An experimental program to
1491–1496. evaluate soil mixes for rain gardens. World Environmental and Water Resources Con-
Denich, C., Bradford, A., 2010. Estimation of evapotranspiration from bioretention areas gress: Floods, Droughts, and Ecosystems, pp. 221–230.
using weighing lysimeters. J. Hydrol. Eng. 15 (6), 522–530. Hess, A., Wadzuk, B., Welker, A., 2017. Evapotranspiration in rain gardens using weighing
Department of Planning and Local Government (DPLG), 2009. Water sensitive urban de- lysimeters. J. Irrig. Drain. Eng. 143 (6), 04017004.
sign technical manual for the greater Adelaide region. Government of South Australia. Hess, A., Wadzuk, B., Welker, A., 2019. Predictive evapotranspiration equations in rain
Adelaide, Australia https://www.sa.gov.au/topics/planning-and-property/land-and- gardens. J. Irrig. Drain. Eng. 145 (7), 04019010.
property-development/planning-professionals/water-sensitive-urban-design. Heydari, M.M., Heydari, M., 2014. Calibration of Hargreaves–Samani equation for estimat-
DiGiovanni, K., Montalto, F., Gaffin, S., Rosenzweig, C., 2013. Applicability of classical pre- ing reference evapotranspiration in semiarid and arid regions. Arch. Agron. Soil Sci.
dictive equations for the estimation of evapotranspiration from urban green spaces: 60 (5), 695–713.
green roof results. J. Hydrol. Eng. 18 (1), 99–107. Hickman, J., Scheider, D., Wadzuk, B., Traver, R., 2010. Determination of evapotranspira-
DiGiovanni-White, K., Montalto, F., Gaffin, S., 2018. A comparative analysis of micromete- tion in SCM's using a weighing Lysimeter: an experimental approach. In World Envi-
orological determinants of evapotranspiration rates within a heterogeneous urban ronmental and Water Resources Congress: Challenges of Change 3990–3999.
environment. J. Hydrol. 562, 223–243. Hickman, J., Wadzuk, B., Traver, R., 2011. Evaluating the role of evapotranspiration in the
Djaman, K., O'Neill, M., Diop, L., Bodian, A., Allen, S., Koudahe, K., Lombard, K., 2018. Eval- hydrology of a bioinfiltration basin using a weighing lysimeter. Proceedings, World
uation of the Penman-Monteith and other 34 reference evapotranspiration equations Environmental and Water Resources Congress: Bearing Knowledge for Sustainability.
under limited data in a semiarid dry climate. Theor. Appl. Climatol. 1–15. ASCE/EWRI, Reston, VA.
A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810 809

Hilten, R.N., 2005. An analysis of the energetics and stormwater mediation potential of Nocco, M.A., Rouse, S.E., Balster, N.J., 2016. Vegetation type alters water and nitrogen bud-
greenroofs. M.S. thesis. University of Georgia, Athens, Georgia. gets in a controlled, replicated experiment on residential-sized rain gardens planted
Hsieh, C.H., Davis, A.P., 2005. Evaluation and optimization of bioretention media for treat- with prairie, shrub, and turfgrass. Urban Ecosyst. 19 (4), 1665–1691.
ment of urban storm water runoff. J. Environ. Eng. 131 (11), 1521–1531. Olson, N.C., Gulliver, J.S., Nieber, J.L., Kayhanian, M., 2013. Remediation to improve infiltra-
Hunt, W.F., Lord, B., Loh, B., Sia, A., 2015. Plant Selection for Bioretention Systems and tion into compact soils. J. Environ. Manag. 117, 85–95.
Stormwater Treatment Practices. Springer Briefs in Water Science and Technology Ouldboukhitine, S.E., Belarbi, R., Djedjig, R., 2012. Characterization of green roof compo-
https://www.oapen.org/download?type=document&docid=1001994. nents: measurements of thermal and hydrological properties. Build. Environ. 56,
Hutchinson, D., Abrams, P., Retzlaff, R., & Liptan, T. (2003). Stormwater Monitoring Two 78–85.
Ecoroofs in Portland, Oregon, USA. Greening Rooftops for Sustainable Communities, Palla, A., Gnecco, I., Lanza, L.G., 2010. Hydrologic restoration in the urban environment
Chicago, IL. 1–18 www. portlandoregon.gov/bes/article/63098. using green roofs. Water 2 (2), 140–154.
Irmak, S., Howell, T.A., Allen, R.G., Payero, J.O., Martin, D.L., 2005. Standardized ASCE Palla, A., Sansalone, J.J., Gnecco, I., Lanza, L.G., 2011. Storm water infiltration in a moni-
Penman-Monteith: impact of sum-of-hourly vs. 24-hour timestep computations at tored green roof for hydrologic restoration. Water Sci. Technol. 64 (3), 766–773.
reference weather station sites. Trans. ASAE 48 (3), 1063–1077. Payne, E.G., Pham, T., Deletic, A., Hatt, B.E., Cook, P.L., Fletcher, T.D., 2018. Which species?
Jensen, M.E., Burman, R.D., Allen, R.G., 1990. Evapotranspiration and irrigation water re- A decision-support tool to guide plant selection in stormwater biofilters. Adv. Water
quirements. ASCE Manuals and Reports on Engineering Practice, No 70, 360. Resour. 113, 86–99.
Jensen, D.T., Hargreaves, G.H., Temesgen, B., Allen, R.G., 1997. Computation of ETo under Penman, H.L., 1948. Natural evaporation from open water, bare soil and grass. Proc. R. Soc.
nonideal conditions. J. Irrig. Drain. Eng. 123 (5), 394–400. Lond. A 193 (1032), 120–145.
Kasmin, H., Stovin, V.R., Hathway, E.A., 2010. Towards a generic rainfall-runoff model for Pennsylvania Department of Environmental Protection (PA DEP). (2006). Stormwater
green roofs. Water Sci. Technol. 62 (4), 898–905. Best Management Practices Manual, Chapter 6: Structural BMPs.
Katul, G.G., Cuenca, R.H., Grebet, P., Wright, J.L., Pruitt, W.O., 1992. Analysis of evaporative Pitt, R., Chen, S.E., Clark, S.E., Swenson, J., Ong, C.K., 2008. Compaction's impacts on urban
flux data for various climates. J. Irrig. Drain. Eng. 118 (4), 601–618. storm-water infiltration. J. Irrig. Drain. Eng. 134 (5), 652–658.
Khoob, A.R., 2008. Comparative study of Hargreaves's and artificial neural network's PWD, 2015. Stormwater Regulation Updates, July 2015, Philadelphia Water Department.
methodologies in estimating reference evapotranspiration in a semiarid environ- PA, Philadelphia http://www.phillywatersheds.org/StormwaterRegulations.
ment. Irrig. Sci. 26 (3), 253–259. PWD, 2018. Stormwater Management Guidance Manual Version 3.1. Philadelphia Water
Krebs, G., Kuoppamäki, K., Kokkonen, T., Koivusalo, H., 2016. Simulation of green roof test Department, Philadelphia, PA https://www.pwdplanreview.org/manual-info/guid-
bed runoff. Hydrol. Process. 30 (2), 250–262. ance-manual.
Kuehler, E., Hathaway, J., Tirpak, A., 2017. Quantifying the benefits of urban forest systems Rezaei, F., Jarrett, A.R., Berghage, R.D., Beattie, D.J., 2005. Evapotranspiration rates from ex-
as a component of the green infrastructure stormwater treatment network. tensive green roof plant species. 2005 ASAE Annual Meeting. American Society of Ag-
Ecohydrology 10 (3), e1813. ricultural and Biological Engineers, p. 1.
Kurtz, T., 2008. Flow monitoring of three ecoroofs in Portland, Oregon. 2008 Low Impact Saxton, K.E., Rawls, W.J., 2006. Soil water characteristic estimates by texture and organic
Development Conference. ASCE, pp. 1–10. matter for hydrologic solutions. Soil Sci. Soc. Am. J. 70 (5), 1569–1578.
LeBoon, Megan, 2014. Evapotranspiration Research Update, Memorandum, File No. Scharenbroch, B.C., Morgenroth, J., Maule, B., 2016. Tree species suitability to bioswales
781450002. AMEC, Philadelphia Water Department. and impact on the urban water budget. J. Environ. Qual. 45 (1), 199–206.
Lee, R.S., Traver, R.G., Welker, A.L., 2016. Evaluation of soil class proxies for hydrologic per- Schneider, D., Wadzuk, B.M., Traver, R.G., 2011. Using a weighing lysimeter to determine a
formance of in situ bioinfiltration systems. J. Sustain. Water Built Environ. 2 (4), crop coefficient for a green roof to predict evapotranspiration with the FAO standard-
04016003. ized Penman-Monteith equation. World Environmental and Water Resources Con-
Lewis, M., James, J., Shaver, E., Blackbourn, S., Leahy, A., Seyb, R., Simcock, R., Wihongi, P., gress, pp. 3629–3638.
Sides, E., Coste, C., 2015. Water sensitive design for stormwater. Auckland Council Selbig, W.R., Balster, N., 2010. Evaluation of turf-grass and prairie-vegetated rain gardens
Guideline Document GD2015/004 http://www.aucklanddesignmanual.co.nz/regula- in a clay and sand soil, Madison, Wisconsin, water years 2004–08. U.S. Geological Sur-
tions/technical-guidance/wsd. vey Scientific Investigations Report 2010–5077 , p. 72. https://pubs.usgs.gov/sir/
Li, H., Sharkey, L.J., Hunt, W.F., Davis, A.P., 2009. Mitigation of impervious surface hydrol- 2010/5077/.
ogy using bioretention in North Carolina and Maryland. J. Hydrol. Eng. 14 (4), Shariat, R., Roozbahani, A., Ebrahimian, A., 2019. Risk analysis of urban stormwater infra-
407–415. structure systems using fuzzy spatial multi-criteria decision making. Sci. Total Envi-
Liu, K., Minor, J., 2005. Performance Evaluation of an Extensive Green Roof. Greening ron. 647, 1468–1477.
Rooftops for Sustainable Communities, Washington DC, pp. 1–11. http://seedengr. Sherrard Jr., J.A., Jacobs, J.M., 2012. Vegetated roof water-balance model: experimental
com/documents/Performance%20evaluation%20of%20an%20extensive%20green% and model results. J. Hydrol. Eng. 17 (8), 858–868.
20roof.pdf. Sickles, L., Parker, N., Wu, J.S., Hilger, H., 2007. Evaluation of regionally appropriate and
Liu, J., Sample, D., Bell, C., Guan, Y., 2014. Review and research needs of bioretention used cost effective bioretention media mixes. World Environmental and Water Resources
for the treatment of urban stormwater. Water 6 (4), 1069–1099. Congress, ASCE 1–10.
Loiola, C., Mary, W., & da Silva, L. P. (2018). Hydrological performance of modular-tray Speak, A.F., Rothwell, J.J., Lindley, S.J., Smith, C.L., 2013. Rainwater runoff retention on an
green roof systems for increasing the resilience of mega-cities to climate change. J. aged intensive green roof. Sci. Total Environ. 461, 28–38.
Hydrol.. in press. https://doi.org/ https://doi.org/10.1016/j.jhydrol.2018.01.004. Spolek, G., 2008. Performance monitoring of three ecoroofs in Portland, Oregon. Urban
López-Urrea, R., de Santa Olalla, F.M., Fabeiro, C., Moratalla, A., 2006. Testing evapotrans- Ecosyst. 11 (4), 349–359.
piration equations using lysimeter observations in a semiarid climate. Agric. Water Stovin, V., 2010. The potential of green roofs to manage urban stormwater. Water Envi-
Manag. 85 (1–2), 15–26. ron. J. 24, 192–199.
Lu, J., Sun, G., McNulty, S.G., Amatya, D.M., 2005. A comparison of six potential evapo- Stovin, V., Vesuviano, G., Kasmin, H., 2012. The hydrological performance of a green roof
transpiration methods for regional use in the southeastern United States. J. Am. test bed under UK climatic conditions. J. Hydrol. 414, 148–161.
Water Resour. Assoc. 41 (3), 621–633. Suleiman, A.A., Hoogenboom, G., 2009. A comparison of ASCE and FAO-56 reference
Lundholm, J., MacIvor, J.S., MacDougall, Z., Ranalli, M., 2010. Plant species and functional evapotranspiration for a 15-min time step in humid climate conditions. J. Hydrol.
group combinations affect green roof ecosystem functions. PLoS One 5 (3), e9677. 375 (3–4), 326–333.
Maimone, Mark, 2010. Evaporation and Transpiration Impacts on Green Stormwater In- Szota, C., Fletcher, T.D., Desbois, C., Rayner, J.P., Williams, N.S., Farrell, C., 2017. Laboratory
frastructure, Memorandum. Philadelphia Water Department. tests of substrate physical properties may not represent the retention capacity of
Marasco, D.E., Culligan, P.J., McGillis, W.R., 2015. Evaluation of common evapotranspira- green roof substrates in situ. Water 9 (12), 920.
tion models based on measurements from two extensive green roofs in new York Szota, C., McCarthy, M.J., Sanders, G.J., Farrell, C., Fletcher, T.D., Arndt, S.K., Livesley, S.J.,
City. Ecol. Eng. 84, 451–462. 2018. Tree water-use strategies to improve stormwater retention performance of
MDESD (2007). Bioretention manual. Maryland Environmental Services Division, Depart- biofiltration systems. Water Res. 144, 285–295.
ment of Environmental Resources, Prince George's County, Maryland. Tabari, H., Talaee, P.H., 2011. Local calibration of the Hargreaves and Priestley-Taylor
Melbourne Water, 2018. MUSIC Guidelines; Input Parameters and Modelling Approaches equations for estimating reference evapotranspiration in arid and cold climates of
for MUSIC Users in Melbourne Water's Service Area. State Government of Victoria, Iran based on the Penman-Monteith model. J. Hydrol. Eng. 16 (10), 837–845.
Melbourne, Australia https://www.melbournewater.com.au/planning-and-building/ Teemusk, A., Mander, Ü., 2007. Rainwater runoff quantity and quality performance from a
stormwater-management. greenroof: the effects of short-term events. Ecol. Eng. 30 (3), 271–277.
Montalto, F., Behr, C., Alfredo, K., Wolf, M., Arye, M., Walsh, M., 2007. Rapid assessment of Thornthwaite, C.W., 1948. An approach toward a rational classification of climate. Geogr.
the cost-effectiveness of low impact development for CSO control. Landsc. Urban Rev. 38 (1), 55–94.
Plan. 82 (3), 117–131. Tirpak, R.A., Hathaway, J.M., Franklin, J.A., 2018. Evaluating the influence of design strate-
Monteith, J.L., 1965. Evaporation and environment. In: Fogg, G.E. (Ed.), Symposium of the gies and meteorological factors on tree transpiration in bioretention suspended pave-
Society for Experimental Biology, the State and Movement of Water in Living Organ- ment practices. Ecohydrology 11 (8), e2037.
isms. vol. 19. Academic Press, Inc., NY, pp. 205–234. Traver, R.G., Ebrahimian, A., 2017. Dynamic design of green stormwater infrastructure.
Moran, A.C., Hunt, W.F., Smith, J.T., 2005. Green roof hydrologic and water quality perfor- Front. Environ. Sci. Eng. 11 (4), 15.
mance from two field sites in North Carolina. Watershed Management Conference. TRCA, 2006. Evaluation of an Extensive Greenroof: York University, Toronto, Ontario. To-
ASCE, pp. 1–12. ronto and Region Conservation Authority, Sustainable Technologies Evaluation Pro-
MPCA, 2017. Minnesota Stormwater Manual, Minnesota Pollution Control Agency, St. gram. https://sustainabletechnologies.ca/app/uploads/2013/03/GR_york_fullreport.
Paul, MN http://stormwater.pca.state.mn.us/index.php/Main_Page. pdf.
Muerdter, C.P., Wong, C.K., LeFevre, G.H., 2018. Emerging investigator series: the role of VADCR, 2011. Virginia DCR Stormwater Design Specification No. 9: Bioretention, Version
vegetation in bioretention for stormwater treatment in the built environment: pol- 1.9. Virginia Department of Conservation and Recreation, Richmond, VA.
lutant removal, hydrologic function, and ancillary benefits. Environ. Sci.: Water Res. VanWoert, N.D., Rowe, D.B., Andresen, J.A., Rugh, C.L., Fernandez, R.T., Xiao, L., 2005. Green
Technol. 4 (5), 592–612. roof stormwater retention. J. Environ. Qual. 34 (3), 1036–1044.
810 A. Ebrahimian et al. / Science of the Total Environment 688 (2019) 797–810

Versini, P.A., Ramier, D., Berthier, E., De Gouvello, B., 2015. Assessment of the hydrological Winston, R.J., Dorsey, J.D., Hunt, W.F., 2016. Quantifying volume reduction and peak flow
impacts of green roof: from building scale to basin scale. J. Hydrol. 524, 562–575. mitigation for three bioretention cells in clay soils in northeast Ohio. Sci. Total Envi-
Villarreal, E.L., Bengtsson, L., 2005. Response of a Sedum green-roof to individual rain ron. 553, 83–95.
events. Ecol. Eng. 25 (1), 1–7. Wolf, D., Lundholm, J.T., 2008. Water uptake in green roof microcosms: effects of plant
Voyde, E., Fassman, E., Simcock, R., 2010a. Hydrology of an extensive living roof under species and water availability. Ecol. Eng. 33 (2), 179–186.
sub-tropical climate conditions in Auckland, New Zealand. J. Hydrol. 394 (3), Yang, W.Y., Li, D., Sun, T., Ni, G.H., 2015. Saturation-excess and infiltration-excess runoff
384–395. on green roofs. Ecol. Eng. 74, 327–336.
Voyde, E., Fassman, E., Simcock, R., Wells, J., 2010b. Quantifying evapotranspiration rates Zaremba, G., Molina, S., Wadzuk, B.M., Traver, R.G., 2015a. Optimizing green roof Design
for New Zealand green roofs. J. Hydrol. Eng. 15 (6), 395–403. for Evapotranspiration. In International Low Impact Development Conference: It
Wadzuk, B., 2013. How best to take credit for ET in green roofs-a roundtable discussion. Works in All Climates and Soils, ASCE 186–195.
2013 VUSP-PA Stormwater Symposium, Villanova, PA. Zaremba, G., Wadzuk, B.M., Traver, R.G., 2015b. Application of the Hargreaves equation
Wadzuk, B.M., Traver, R.G., 2012. Soil moisture balance in bioretention systems. Proc. for green roof evapotranspiration. In World Environmental and Water Resources
Water Environ. Fed. 2012 (5), 231–240. Congres: Floods, Droughts, and Ecosystems, ASCE 231–240.
Wadzuk, B., Schneider, D., Feller, M., Traver, R., 2013. Evapotranspiration from a green- Zaremba, G., Traver, R., Wadzuk, B., 2016. Impact of drainage on green roof evapotranspi-
roof storm-water control measure. J. Irrig. Drain. Eng. 139 (12), 995–1003. ration. J. Irrig. Drain. Eng. 142 (7), 04016022.
Wadzuk, B., Hickman Jr., J., Traver, R., 2015. Understanding the role of evapotranspiration Zhang, K., Chui, T.F.M., 2019. Linking hydrological and bioecological benefits of green in-
in bioretention: Mesocosm study. J. Sustain. Water Built Environ. 1 (2), 04014002. frastructures across spatial scales–a literature review. Sci. Total Environ. 646,
Wadzuk, B.M., Lewellyn, C., Lee, R., Traver, R.G., 2017. Green infrastructure recovery: anal- 1219–1231.
ysis of the influence of back-to-back rainfall events. Journal of sustainable water in Zhang, L., Dawes, W.R., Walker, G.R., 2001. Response of mean annual evapotranspiration
the built. Environment 3 (1), 04017001. to vegetation changes at catchment scale. Water Resour. Res. 37 (3), 701–708.
Walter, I.A., Allen, R.G., Elliott, R., Jensen, M.E., Itenfisu, D., Mecham, B., Howell, T.A., Zhang, Z., Szota, C., Fletcher, T.D., Williams, N.S., Werdin, J., Farrell, C., 2018a. Influence of
Snyder, R., Brown, P., Echings, S., Spofford, T., Hattendorf, M., Cuenca, R.H., Wright, plant composition and water use strategies on green roof stormwater retention. Sci.
J.L., Martin, D., 2000. ASCE's standardized reference evapotranspiration equation. Wa- Total Environ. 625, 775–781.
tershed Management and Operations Management 2000, pp. 1–11. Zhang, K., Chui, T.F.M., Yang, Y., 2018b. Simulating the hydrological performance of low
Washington State Department of Ecology (WSDE), 2012. Stormwater Management Man- impact development in shallow groundwater via a modified SWMM. J. Hydrol. 566,
ual for Western Washington. Department of Ecology, State of Washington, Lacey, 313–331.
Washington Publication Number 12-10-030. http://www.ecy.wa.gov/programs/wq/ Zhao, L., Xia, J., Xu, C.Y., Wang, Z., Sobkowiak, L., Long, C., 2013. Evapotranspiration estima-
stormwater/manual.html. tion methods in hydrological models. J. Geogr. Sci. 23 (2), 359–369.
Wilm, H.G., Thornthwaite, C.W., Colman, E.A., Cummings, N.W., Croft, A.R., Gisborne, H.T.,
Harding, S.T., Hendrickson, A.H., Hoover, M.D., Houk, I.E., Kittredge, J., 1944. Report of
the committee on transpiration and evaporation, 1943-44. Eos, Transactions
American Geophysical Union 25 (5), 683–693.

View publication stats

You might also like