Professional Documents
Culture Documents
Reuse of Urban Runoff in Australia: A Review of Recent Advances and Remaining Challenges
Reuse of Urban Runoff in Australia: A Review of Recent Advances and Remaining Challenges
Reuse of Urban Runoff in Australia: A Review of Recent Advances and Remaining Challenges
TECHNICAL REPORTS
Reuse of Urban Runoff in Australia: A Review of Recent Advances and Remaining Challenges
Tim D. Fletcher,* Ana Deletic, V. Grace Mitchell, and Belinda E. Hatt Monash University
S-116
sequence, demand for water in many urban areas is approach- channels with that of lining some or all drainage channels
ing, and in some cases exceeding, the available supply (Hatt et within the catchment. They concluded that 80 to 90% of
al., 2006a). For example, in 2006, six of the seven Australian the increase in peak flows is explained by the nature of the
state capital cities faced water restrictions, affecting an esti- drainage connection, rather than simply the proportion of
mated 70% of the Australian population of 21 million people the catchment that is impervious. More recently, Hatt et al.’s
(Australian Bureau of Statistics, 2006). Similar shortages face (2004b) investigation of water quality in relation to catch-
parts of Africa, China, Singapore, the USA, Israel, and Europe ment characteristics also showed that the nature of connection
(Niemczynowicz, 1996; Vlachos and Braga, 2001). of impervious areas to receiving waters was a strong indepen-
Although there are several potential alternative water sup- dent predictor of pollutant concentrations and loads.
plies that may help to address this shortage, such as recycling One aspect of the hydrologic changes brought about by urban-
of wastewater or desalination of sea water, the harvesting and ization that has been less studied is the change in evapotranspira-
treatment of urban runoff has the advantage of higher public tion. Generally, it is expected that the amount of evapotranspira-
acceptance than wastewater recycling (Brown and Davies, tion will be less after urbanization due to the loss of vegetation
2007; Coombes and Mitchell, 2006). Mitchell et al. (2003) (Oke, 1987). However, this effect may be altered by the impact of
showed that the annual storm water discharged (from separate garden watering and open-space irrigation (Grimmond and Oke,
storm water sewer systems) from three major Australian cit- 1991), particularly in areas with lower density and large gardens.
ies—Brisbane, Melbourne, and Sydney—was similar to their Although it has long been known that urbanization and the
total water demand (50% of which is for nonpotable uses). documented changes in hydrology and water quality lead to deg-
The other key driver for harvesting of storm water is an increas- radation of aquatic ecosystems (Aalderink et al., 1990; Arthing-
ing recognition of the impacts of unmitigated urban runoff. Per- ton et al., 1983; Pratt et al., 1981), the recent insights into the
haps more importantly, there has been substantial recent research importance of drainage connection has significantly informed
into the mechanisms that contribute to this degradation. Broadly recent ecological studies of the “urban stream syndrome” (Walsh
speaking, urbanization causes significant changes to a catchment’s et al., 2005b). For example, Taylor et al. (2004) showed that ben-
hydrology (e.g., Leopold, 1968; Rose and Peters, 2001) and water thic algal biomass was directly related to the degree of connec-
quality (Hatt et al., 2004b; Soranno et al., 1996). tion of impervious areas to receiving waters. Harbott and Grace
Changes to hydrology from urbanization are well docu- (2005) showed similar relationships for extracellular enzyme re-
mented and are related to two factors: (i) the proportion of the sponse, as did Walsh (2004) for macroinvertebrates. Walsh et al.
catchment, which is made up of impervious surfaces, and (ii) the (2005a, 2004) observed that the frequency of runoff into receiv-
hydraulic efficiency of flow paths from these impervious areas ing waters is a critical indicator of the likely degradation to receiv-
to receiving waters (Booth and Jackson, 1997; Leopold, 1968; ing waters. Consequently, they suggest that it is necessary to in-
Walsh et al., 2005b; Wong et al., 2000). For example, Leopold tercept and retain any rainfall up to the amount that would have
(1968) showed that urbanization leads to decreases in the lag caused surface runoff under predeveloped conditions. Ladson et
time between rainfall and runoff and increased peak flow, with al. (2007) further explore this, demonstrating that installation of
the greatest proportional increase occurring for the smaller, more rain water tanks on each property and the installation of storm
frequent events (because for larger events, pervious areas become water infiltration and biofiltration systems would be sufficient to
saturated and begin to produce runoff anyway). Leopold suggest- achieve the required predeveloped runoff frequency.
ed that 20% imperviousness would result in a doubling of the It is also well established that the increased peak flows that
frequency of overbank flows, with 50% resulting in four times occur as a result of urbanization can lead to geomorphologic
the predeveloped frequency. Similar observations have been made changes, such as bed and bank erosion and channel enlarge-
by several studies (Codner et al., 1988; Driver and Troutman, ment (Booth, 1990; Booth and Jackson, 1997; Konrad et al.,
1989; Olivera and DeFee, 2007; Wong et al., 2000). Wong et al. 2005) and the destruction of benthic habitat due to sediment
(2000) showed that the 20% annual exceedence probability flow smothering (Lenat and Crawford, 1994; Olthof, unpublished
(i.e., the flow with an average recurrence interval of 5 yr) under data, 1994; Wang et al., 1997).
rural conditions would occur on average 2 to 3 times per year, The important implication of these recent research find-
with only 20% catchment imperviousness. ings is that retention and “loss” of storm water is necessary to
The proportion of a catchment that is impervious is not restore catchment hydrology toward predevelopment levels.
the only predictor of the degree of hydrologic disturbance Storm water reuse provides one potential tool to assist in this
that derives from urbanization. Rather, a long series of studies aim. By detaining and using urban runoff, there may be po-
have examined the extent to which hydraulic efficiency of the tential reductions in peak flow, annual runoff volume, and the
drainage network controls catchment response to urbaniza- frequency of runoff (Walsh et al., 2005a).
tion. Leopold (1968) showed that the post:predevelopment Despite this apparently strong rationale for the harvesting of
peak flow ratio depends on the proportion of impervious ar- urban runoff as a means for simultaneously tackling water supply
eas that are directly connected (via a pipe or constructed lined shortages and degradation of aquatic ecosystems, urban storm
channel, for example) to the receiving water. water remains a relatively neglected water resource (Thomas et al.,
Wong et al. (2000) continued this line of enquiry, com- 1997). One reason for this may be the relative lack of technology
paring the consequences of maintaining natural drainage and facilities already in place, relative to other components of
the urban water management system. For example, recycling of Pitt, 2001). Therefore, storm water treatment commonly uses
wastewater is facilitated by the fact that treatment infrastructure natural purification processes that also allow for temporary stor-
for wastewater treatment is well established, meaning that adding age and attenuation of flows, such as sedimentation in large open
infrastructure for additional treatment, along with storage and water bodies, filtration through soils (or any other filtration me-
distribution, is relatively straightforward (Thomas et al., 1997). dia), or biological uptake by plants and microorganisms. These
systems are usually energy efficient and highly diverse (Table 1).
Recent Advances and Trends These systems were originally known as Best Management Prac-
tices for storm water management (WEF and ASCE, 1998). They
Evolution of Technologies: From Storm water Treatment are now becoming known as Low Impact Development systems in
to Storm Water Harvesting the USA (Department of Environmental Resources, 1999), Water
Urban storm water harvesting systems are made up of four Sensitive Urban Design (WSUD) in Australia (Wong, 2006), and
main components: collection, treatment, storage, and distribu- Sustainable Urban Drainage Systems in the UK (CIRIA, 2000b).
tion, which may also play a role in flood protection (Fig. 1). Ur- There are numerous design guidelines for their construction, with
ban storm water is generated in large quantities within short peri- almost every developed country, state or, region issuing their own
ods, and the nature of this pollutant source is diffuse (Burton and manual (e.g., Auckland Regional Council, 2002; CIRIA, 2001;
Table 1. Storm water treatment measures and their pollutant removal effectiveness for pollution control and storm water harvesting applications
(Source: CIRIA, 2000a, 2000b; WEF and ASCE, 1998; Wong, 2006).
Effectiveness of treatment measure
Storm water Pollution control Storm water harvesting
treatment measure Pollutant removal mechanisms Sediment Nutrients Heavy metals Sediment Nutrients Heavy metals Pathogens
Swales and filter strips sedimentation 99 9 99 9 9 9 ?
coarse filtration
infiltration
Ponds and wetlands sedimentation 999 99 999 99 9 99 99
filtration
biological processes
infiltration
evapotranspiration
Biofiltration systems sedimentation 999 99 999 999 9 999 ?
filtration
biological processes
infiltration
evapotranspiration
Infiltration systems sedimentation 999 99 999 99 9 99 99
filtration
biological processes
infiltration
evapotranspiration
for irrigation of nearby trees, and therefore the system did not acceptable. Among the studies of biofiltration system perfor-
have storage. A small system built in the UK is used to collect, mance, there is a lack of studies into the removal efficiency for
treat, and store storm water runoff from a car park (375 m2) pathogens. Given the public health implications of high levels
that is then used for flushing toilets in a youth hostel (Pratt, of bacteria, viruses, or protozoans in harvested storm water,
1999). The interesting feature here is that the sub-base of the this is an important gap that needs to be addressed.
pavement was used to treat and store the harvested water. There are several storm water harvesting systems that use
It has been concluded that porous pavements can provide the biofiltration as their treatment mechanism currently being
basis for further developments in treatment and storage of storm trialed in Australia, including one at Monash University in
water for reuse. Technologies for achieving an effective imple- Melbourne, which has been extensively monitored. This sys-
mentation of porous pavements for storm water reuse are still tem captures runoff from a carpark; the runoff passes through
to be refined, but they seem to have great potential, particularly pretreatment settling tanks and then into a fully lined biofiltra-
given their ability to provide treatment and storage within the tion system (Fig. 2). The treated water is stored in an adjacent
existing impervious surface, thus overcoming space constraints. pond before being used to irrigate a nearby sports oval. Three
Biofiltration Technologies different filter media types were trialed in separate filter strips:
sandy loam; 80% sandy loam, 10% vermiculite, 10% perlite;
Biofiltration is an emerging technology that has been used as and 80% sandy loam, 10% composted pine bark, and 10%
an effective treatment of storm water, primarily for protection hardwood mulch. The entire biofilter is densely vegetated with
of aquatic ecosystems, and more recently for nonpotable use. native species that are tolerant of variable wetting and drying.
For example, reductions in heavy metal and total suspended The hydraulic and treatment performance of the biofilter
solids concentrations through biofiltration media of more than was monitored via a fully automated monitoring system. Lev-
90% are commonly observed (Davis et al., 2003; Hatt et al., eled V-notch weirs and ultrasonic depth sensors were used to
2008). However, nutrient removal tends to be far more vari- continuously (1-min time step) measure flow rates, whereas au-
able, depending on the filter media (Hatt et al., 2006b, 2007; tosamplers collected flow-weighted water quality samples at the
Henderson et al., 2007), the presence of vegetation (Henderson inlet and three outlet points (i.e., from each filter strip; Fig. 2).
et al., 2007), plant type (Read et al., 2007), and the presence of On average, 30 to 35% of inflow volume was lost via evapo-
an anaerobic zone to promote coupled nitrification–denitrifi- transpiration. Although this is beneficial in terms of pollution con-
cation within the filter medium (Hsieh and Davis, 2005; Kim trol (reducing loads and ensuring that contaminants in the filter
et al., 2003). Most studies show good removal of total Kjeldahl are held there for longer periods, thus allowing treatment processes
nitrogen and NH3 that is often offset by leaching of NO3 to occur), this loss of water is an important consideration when
(Davis et al., 2001, 2003; Henderson et al., 2007; Hatt et al., designing a storm water harvesting system. In terms of pollutant
2008). Phosphorus removal is typically high (e.g., Davis et al., removal, removal of sediment and heavy metals was consistently
2006) and may increase with depth (Hatt et al., 2008). high, outflow concentrations of nutrients were variable, and all
Results from laboratory studies are not always repeated three biofilter strips tended to leach P and N (Table 2). In addition
by monitoring of field systems, particularly due to variations to the regular water quality monitoring, two spiking tests were per-
in soil properties (Hunt et al., 2006) and climate conditions. formed to evaluate the pathogen removal efficiency of the biofilter.
For storm water harvesting systems, this variability may be The pathogen surrogates—baker’s yeast (Saccharomyces cerevisiae),
more problematic than it is for the protection of aquatic eco- E. coli, and the coliphage MS2—were used in these tests to rep-
systems, where some periods of poor water quality may be resent the behavior of the protozoan pathogen Cryptosporidium,
bution of harvested rain water was around 56% for the occupied the construction of alternative water supply infrastructure (such
lot monitored. Coombes et al. (1999) monitored a system where as a new dam). Tools such as Triple Bottom Line Assessment
rain water was collected in a centralized tank, whereas storm water (Taylor and Wong, 2002) may help to overcome the challenge
(from roads, pathways, carparks, etc.) was conveyed to a detention of accounting for all costs and benefits.
basin (but not used). Based on the use of rainwater alone, savings
in potable water demand of around 60 to 65% were obtained. Environmental Benefits
Hatt et al. (2006a) reviewed the performance data of 77 Two categories of environmental benefit are commonly
storm water harvesting systems. Of these, useable data on water demonstrated from storm water harvesting systems: (i) at-
collected and used were available for seven sites (Table 4). At tenuation of postdevelopment hydrology and (ii) reductions
these sites, the percent collected ranged from 20 to 100% of the in pollutant concentrations and/or loads. Most of the studies
mean annual runoff, and the reductions in potable water demand of these impacts rely on modeling, with little or no monitor-
ranged from 17 to 65%. Generally, the sites that collected a low ing data available (Hatt et al., 2006a).
proportion of mean annual runoff were those that were retrofit- The harvesting of storm water will affect downstream water
ted to an existing storm water treatment facility rather than pur- quality. It will reduce pollutant loads to receiving waterways as a
pose built for the harvesting of storm water. Both figures depend function of the water taken away (depending on the use and dis-
on the size of the storage relative to the mean annual runoff vol- posal of that harvested water). For example, if urban storm water
ume, which in turn depends on site constraints. has a typical concentration of around 2.5 mg L−1 total N (Duncan,
Hatt et al.’s review showed that the capital cost of storm 2006; Taylor et al., 2005; Brombach et al., 2005), each mL of wa-
water harvesting systems was related not only to catchment ter harvested will reduce the N load to the receiving water by 2.5
size; factors such as treatment method, land characteristics kg. Hatt et al. (2006a) found significant reductions in N loads to
(slope, presence of rock, etc.), and storage type also affected receiving water and quantified the equivalent cost of this reduction
the cost. The value of potable water savings (Table 4) varied using well established costs for constructing storm water treatment
greatly, from 0.1 to 30.8% of the system’s capital cost. Based wetlands. In the best case, this resulted in an equivalent reduction
on these figures, the payback period (based simplistically on in required storm water treatment of A$37,000 per hectare.
the capital cost alone) ranged from 3 to 950 yr. The influence of storm water harvesting on pollutant concen-
Hatt et al. (2006a) note that there are few data available on trations is more complex. If only roof water, with its lower concen-
the costs and benefits of such systems. Of the systems reviewed trations of particulate pollutants (Duncan, 2006), were harvested,
by Hatt et al. (2004a; 2006a), few gave indications of the unit the concentration at the receiving water may increase. This is un-
cost of water produced. Each of the three that did (Homebush likely to be the case when harvesting of overall storm water runoff.
Bay and Parafield in Australia and Santa Monica in the USA) However, no known monitoring data exist on the impact of storm
showed different outcomes. For example, the Homebush Bay water harvesting on downstream pollutant concentrations.
system cost twice as much as the retail cost of normal reticulat- Storm water harvesting systems can have a significant benefit
ed town water (but was sold at 15% less than the mains water in terms of reducing flow peaks downstream. For example, even
price to encourage uptake). The Santa Monica system provided a distributed network of rainwater tanks specially designed to
a near-neutral outcome, whereas the Parafield system produced provide permanent freeboard by having a “trickle-outlet” at a
water at 30% of the cost of the alternative (pumping from a predetermined point below the top of the tank has been shown to
nearby river). This diversity shows that local context is critical significantly reduce peak flows (Hardy et al., 2004). More detailed
in determining the financial outcomes. modeling showed reductions of around 40 to 50% in the 3-mo
Several authors have asserted that analysis of financial bene- average recurrence interval peak flow, dropping to around 5 to
fits purely in terms of potable water savings provides an incom- 10% for the 100-yr average recurrence interval event. The influ-
plete picture because it excludes the externalities (Coombes et ence of storm water harvesting on flow behavior was also examined
al., 2002), such as reductions in the size of downstream storm by Fletcher et al. (2007). They used the concept of “environmen-
water conveyance and treatment infrastructure and avoiding tal flows,” which is more typically used in the rural context, to