The role of living roofs in holistic stormwater management systems - Living Roofs in Integrated Urban Water Systems - page 40

A computationally more intensive, but not necessarily more accurate, model
was calibrated and veriied for a full-scale extensive living roof in Portland, OR
(She and Pang 2010) and Auckland, New Zealand (She et al. 2010). Up to ten
parameters were required as input or determined as calibration parameters for
the Auckland site. Upper and lower bounds of moisture storage were assumed
similarly as above, and ET was assumed to decrease exponentially during dry
periods. A modiied form of Green-Ampt iniltration was used to predict the rate
of water movement through the media, advancing toward the drainage layer.
During a rain event, the conventional Green-Ampt model assumes that there is a
sharp horizontal boundary (a “wetting front”) between the wet growing media
initially near the top of the media proile, and drier media in the deeper proile.
Saturated conditions are assumed above the wetting front, while below it the
media is at a moisture content dependent on antecedent conditions (determined
by ET between events). Setting the upper bound of moisture storage at the ield
capacity modiies the Green-Ampt method in this model to allow for runoff to
occur from the system without it reaching saturation, albeit violating the meth-
od's theoretical underlying assumptions. As it rains, the wetting front advances
vertically downwards toward the drainage layer based on physical properties of
the media. Green-Ampt iniltration has also been embedded in the living roof
routine of the US EPA's Stormwater Management Model (SWMM) version 5.1,
which has the capability to automatically integrate living roofs into wider
watershed development simulations. Its current formulation has yet to be tested
in the academic literature. SWMM 5.1 is discussed in more detail in Section 3.5 .
In the ield, a living roof should only operate under unsaturated low condi-
tions because the growing media's coarse texture and large pore size prevents
physically achieving saturation while providing very high permeability compared
to rainfall intensity ( Section 4.1 ). Unsaturated low occurs at rates always less
than saturated low, but the actual rate depends on the moisture content, thus
adding another level of complexity to the problem. Unsaturated to saturated low
conditions in porous media can be described by Richard's Equation, which is com-
putationally demanding. The Richard's Equation is operationalized for event-
based simulation in one-, two- or three-dimensions in software packages such as
HYDRUS (Šim ů nek et al. 2012) and SWMS_2D (Šim ů nek et al. 1994). Hilten et al.
(2008) successfully calibrated HYDRUS 1-D to simulate runoff from a 37 m 2 living
roof in Georgia, United States. The authors noted limitations imposed by a lack
of data for storms greater than 50 mm and a poor ability to represent the
growing media composition. Palla et al. (2012) concluded HYDRUS 1-D was suc-
cessful in reproducing retention, peak low and overall hydrograph shape for
several events from a 350 m 2 , 200-mm deep living roof in Genoa, Italy. For the
same living roof, SWMS_2D was deemed suitable to reproduce the growing
media moisture content, and the runoff hydrograph shape, volume and timing
(Palla et al. 2011).
Implicit in the application of the Green-Ampt method or the Richard's
Equation is that water movement through living roof growing media is assumed