Planning considerations - Living Roofs in Integrated Urban Water Systems - page 77

with depths ranging ~25-100 cm, Alfredo et al. (2010) determined a CN range
of 92-95.
The more recent advice in Hawkins et al. (2009) provides an alternative
method for determining the CN . Fassman-Beck et al. (in preparation) compiled
data from 22 sources in the literature and some previously unpublished data to
calculate CN s according to Hawkins et al. (2009). Data originated from extensive
living roofs mostly from the United States including New York, Illinois, Pennsylva-
nia, North Carolina, Georgia, Michigan and Oregon, along with data from Auck-
land (New Zealand), Toronto (Canada), Shefield (UK) and Genoa (Italy). In many
cases, multiple extensive living roofs were monitored in each city or state. All sites
were ield studies subject to natural rainfall, while the living roof footprints
ranged from test plot or garden shed-scale experiments (1-4 m 2 ) to full-scale
roofs (approximately 40-7,000 m 2 ). Monitored roofs ranged from “lat” to 10
percent pitch, with one site at 25 percent. One pre- fabricated modular tray
system was included. Full site descriptions and many of the data sets are found in
Berghage et al. (2010), Carpenter and Isenberg (2012), Carpenter and Kaluva-
kolanu (2011), Carter and Rasmussen (2006), Fassman-Beck et al. (2013), Hatha-
way et al. (2008), Hoffman et al. (2010), Hutchinson et al. (2003), Kurtz et al.
(2010), Palla et al. (2011) and Stovin et al. (2012). Researchers contributed addi-
tional or previously unpublished data for CN and runoff coeficient analysis for
sites in Portland, Oregon (Kurtz, personal communication 2013) and Villanova,
Pennsylvania (Wadzuk, personal communication 2014).
Calculated CN s were organized according to Köppen Geiger climate zone. 8
Signiicant caution is advised in applying the resultant CN s, for several reasons.
Table 3.3 summarizes the average result per climate zone, but site-to-site results
varied, as evidenced by the standard deviations. This may relect differences in
living roof coniguration and/or the signiicant differences in climate that may be
observed within individual climate zones. For example, Stovin et al. (2013) used a
validated hydrological model to demonstrate that retention performance for the
same extensive roof coniguration varied widely between four UK (Cfb) locations.
Averaged CN presented in Table 3.3 are somewhat surprisingly high, given the
exceptional retention performance evidence in the monitoring literature. Availa-
ble living roof data is relatively limited; only a few (at most) living roofs represent
each climate zone (with the exception of Cfb). Most importantly, many individual
sites provide relatively small data sets dominated by small events. The CN meth-
odology itself has been observed to be less accurate for storms that generate less
than 12.5 mm of rainfall (uSDA 1986). Rainfall- runoff data from relatively larger
events should be the focus of CN determination. Hawkins et al. (1985) concluded
that CN s should be determined from storm events with rainfall ( P ) that is at least
0.46 times greater than storage ( S ) in Equations 3.1 and 3.2. In fact, Hawkins et
al. (2009) directly observe the potential for elevated CN s derived from small rain-
falls. As was described in Section 2.1 , the majority of storms produce relatively
small amounts of rain across a range of climates, thus biasing the living roof data
sets toward small storm performance. The data limitation is further exacerbated